<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html lang="en"> <head> <meta http-equiv="content-type" content="text/html; charset=ISO-8859-1"> <meta name="language" content="en"> <meta name="date" content="2013-10-11T14:24:03"> <meta name="generator" content="deplate.rb 0.8.5"> <title>User’s Guide for VirtualGL 2.3.3</title> <link rel="start" href="index.html" title="Frontpage"> <link rel="chapter" href="index.html#hd001" title="1 Legal Information"> <link rel="chapter" href="index.html#hd002" title="2 Conventions Used in This Document"> <link rel="chapter" href="index.html#hd003" title="3 Overview"> <link rel="chapter" href="index.html#hd004" title="4 System Requirements"> <link rel="chapter" href="index.html#hd005" title="5 Obtaining and Installing VirtualGL"> <link rel="chapter" href="index.html#hd006" title="6 Configuring a Linux or Unix Machine as a VirtualGL Server"> <link rel="chapter" href="index.html#hd007" title="7 Configuring a Windows Machine as a VGL Transport Client"> <link rel="chapter" href="index.html#hd008" title="8 Using VirtualGL with the VGL Transport"> <link rel="chapter" href="index.html#hd009" title="9 Using VirtualGL with X Proxies Such as VNC"> <link rel="chapter" href="index.html#hd0010" title="10 Support for the X Video Extension"> <link rel="chapter" href="index.html#hd0011" title="11 Transport Plugins"> <link rel="chapter" href="index.html#hd0012" title="12 Using VirtualGL with setuid/setgid Executables"> <link rel="chapter" href="index.html#hd0013" title="13 Using VirtualGL with Chromium"> <link rel="chapter" href="index.html#hd0014" title="14 Using VirtualGL with VirtualBox"> <link rel="chapter" href="index.html#hd0015" title="15 Using VirtualGL with VMWare Workstation"> <link rel="chapter" href="index.html#hd0016" title="16 Other Application Recipes"> <link rel="chapter" href="index.html#hd0017" title="17 Advanced OpenGL Features"> <link rel="chapter" href="index.html#hd0018" title="18 Performance Measurement"> <link rel="chapter" href="index.html#hd0019" title="19 The VirtualGL Configuration Dialog"> <link rel="chapter" href="index.html#hd0020" title="20 Advanced Configuration"> <link rel="stylesheet" type="text/css" href="virtualgl.css" title="virtualgl"> </head> <body > <a name="#pagetop"></a> <div class="title"> <p class="title">User’s Guide for VirtualGL 2.3.3</p> </div> <a name="file000"></a> <p><em>Intended audience:</em> System Administrators, Graphics Programmers, Researchers, and others with knowledge of Linux or Unix operating systems, OpenGL and GLX, and X windows.</p> <div id="Table_of_Contents"> <div id="Table_of_ContentsBlock" class="toc"> <h1 class="toc">Table of Contents</h1> <ul class="toc"> <li class="Itemize-1 toc"> <a href="#hd001" class="toc">1 Legal Information</a> </li> <li class="Itemize-1 toc"> <a href="#hd002" class="toc">2 Conventions Used in This Document</a> </li> <li class="Itemize-1 toc"> <a href="#hd003" class="toc">3 Overview</a> </li> <li class="Itemize-1 toc"> <a href="#hd004" class="toc">4 System Requirements</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd004001" class="toc">4.1 Linux/x86 and Other x86 Un*x Operating Systems</a> </li> <li class="Itemize-3 toc"> <a href="#hd004002" class="toc">4.2 Mac/x86</a> </li> <li class="Itemize-3 toc"> <a href="#hd004003" class="toc">4.3 Windows</a> </li> <li class="Itemize-3 toc"> <a href="#hd004004" class="toc">4.4 Additional Requirements for Stereographic Rendering</a> </li> <li class="Itemize-3 toc"> <a href="#hd004005" class="toc">4.5 Additional Requirements for Transparent Overlays</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd005" class="toc">5 Obtaining and Installing VirtualGL</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd005001" class="toc">5.1 Installing VirtualGL on Linux</a> </li> <li class="Itemize-3 toc"> <a href="#hd005002" class="toc">5.2 Installing the VirtualGL Client on OS X</a> </li> <li class="Itemize-3 toc"> <a href="#hd005003" class="toc">5.3 Installing the VirtualGL Client on Windows (Exceed)</a> </li> <li class="Itemize-3 toc"> <a href="#hd005004" class="toc">5.4 Installing the VirtualGL Client on Windows (Cygwin/X)</a> </li> <li class="Itemize-3 toc"> <a href="#hd005005" class="toc">5.5 Installing VirtualGL from Source</a> </li> <li class="Itemize-3 toc"> <a href="#hd005006" class="toc">5.6 Uninstalling VirtualGL</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd006" class="toc">6 Configuring a Linux or Unix Machine as a VirtualGL Server</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd006001" class="toc">6.1 Granting Access to the 3D X Server</a> </li> <li class="Itemize-3 toc"> <a href="#hd006002" class="toc">6.2 Using VirtualGL with Multiple GPUs</a> </li> <li class="Itemize-3 toc"> <a href="#hd006003" class="toc">6.3 SSH Server Configuration</a> </li> <li class="Itemize-3 toc"> <a href="#hd006004" class="toc">6.4 Un-Configuring the Server</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd007" class="toc">7 Configuring a Windows Machine as a VGL Transport Client</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd007001" class="toc">7.1 Configuring and Optimizing Exceed</a> </li> <li class="Itemize-3 toc"> <a href="#hd007002" class="toc">7.2 Optimizing Cygwin/X</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd008" class="toc">8 Using VirtualGL with the VGL Transport</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd008001" class="toc">8.1 VGL Transport with X11 Forwarding</a> </li> <li class="Itemize-3 toc"> <a href="#hd008002" class="toc">8.2 VGL Transport with a Direct X11 Connection</a> </li> <li class="Itemize-3 toc"> <a href="#hd008003" class="toc">8.3 VGL Transport with X11 Forwarding and SSH Tunneling</a> </li> <li class="Itemize-3 toc"> <a href="#hd008004" class="toc">8.4 VGL Transport over Gigabit Networks</a> </li> <li class="Itemize-3 toc"> <a href="#hd008005" class="toc">8.5 VGL Transport with XDMCP</a> </li> <li class="Itemize-3 toc"> <a href="#hd008006" class="toc">8.6 The VirtualGL Client Application: Nuts and Bolts</a> <ul class="toc"> <li class="Itemize-5 toc"> <a href="#hd008006001" class="toc">8.6.1 The VirtualGL Client and Firewalls</a> </li> </ul> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd009" class="toc">9 Using VirtualGL with X Proxies Such as VNC</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd009001" class="toc">9.1 Using VirtualGL with an X Proxy on the Same Server</a> </li> <li class="Itemize-3 toc"> <a href="#hd009002" class="toc">9.2 Using VirtualGL with an X Proxy on a Different Machine</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd0010" class="toc">10 Support for the X Video Extension</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd0010001" class="toc">10.1 The VGL Transport with YUV Encoding</a> </li> <li class="Itemize-3 toc"> <a href="#hd0010002" class="toc">10.2 The XV Transport</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd0011" class="toc">11 Transport Plugins</a> </li> <li class="Itemize-1 toc"> <a href="#hd0012" class="toc">12 Using VirtualGL with setuid/setgid Executables</a> </li> <li class="Itemize-1 toc"> <a href="#hd0013" class="toc">13 Using VirtualGL with Chromium</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd0013001" class="toc">13.1 Configuration 1: Sort-First Rendering (Image-Space Decomposition)</a> </li> <li class="Itemize-3 toc"> <a href="#hd0013002" class="toc">13.2 Configuration 2: Sort-First Rendering (Image-Space Decomposition) with Readback</a> </li> <li class="Itemize-3 toc"> <a href="#hd0013003" class="toc">13.3 Configuration 3: Sort-Last Rendering (Object-Space Decomposition)</a> </li> <li class="Itemize-3 toc"> <a href="#hd0013004" class="toc">13.4 A Note About Performance</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd0014" class="toc">14 Using VirtualGL with VirtualBox</a> </li> <li class="Itemize-1 toc"> <a href="#hd0015" class="toc">15 Using VirtualGL with VMWare Workstation</a> </li> <li class="Itemize-1 toc"> <a href="#hd0016" class="toc">16 Other Application Recipes</a> </li> <li class="Itemize-1 toc"> <a href="#hd0017" class="toc">17 Advanced OpenGL Features</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd0017001" class="toc">17.1 Stereographic Rendering</a> <ul class="toc"> <li class="Itemize-5 toc"> <a href="#hd0017001001" class="toc">17.1.1 Quad-Buffered Stereo</a> </li> <li class="Itemize-5 toc"> <a href="#hd0017001002" class="toc">17.1.2 Anaglyphic Stereo</a> </li> <li class="Itemize-5 toc"> <a href="#hd0017001003" class="toc">17.1.3 Passive Stereo</a> </li> <li class="Itemize-5 toc"> <a href="#hd0017001004" class="toc">17.1.4 Selecting a Stereo Mode</a> </li> </ul> </li> <li class="Itemize-3 toc"> <a href="#hd0017002" class="toc">17.2 Transparent Overlays</a> </li> <li class="Itemize-3 toc"> <a href="#hd0017003" class="toc">17.3 Color Index (PseudoColor) Rendering</a> </li> <li class="Itemize-3 toc"> <a href="#hd0017004" class="toc">17.4 Troubleshooting</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd0018" class="toc">18 Performance Measurement</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd0018001" class="toc">18.1 VirtualGL’s Built-In Profiling System</a> </li> <li class="Itemize-3 toc"> <a href="#hd0018002" class="toc">18.2 Frame Spoiling</a> </li> <li class="Itemize-3 toc"> <a href="#hd0018003" class="toc">18.3 VirtualGL Diagnostic Tools</a> </li> </ul> </li> <li class="Itemize-1 toc"> <a href="#hd0019" class="toc">19 The VirtualGL Configuration Dialog</a> </li> <li class="Itemize-1 toc"> <a href="#hd0020" class="toc">20 Advanced Configuration</a> <ul class="toc"> <li class="Itemize-3 toc"> <a href="#hd0020001" class="toc">20.1 Server Settings</a> </li> <li class="Itemize-3 toc"> <a href="#hd0020002" class="toc">20.2 Client Settings</a> </li> </ul> </li> </ul> </div></div> <p><br /></p> <hr class="break" /> <h1 id="hd001"><a name="file001"></a>1 Legal Information</h1> <p><img src="somerights20.png" alt="somerights20" class="inline" id="imgid_0" name="imgid_0"/></p> <p>This document and all associated illustrations are licensed under the <span class="remote"><a href="http://creativecommons.org/licenses/by/2.5/" class="remote">Creative Commons Attribution 2.5 License</a></span><a name="idx001"></a>. Any works that contain material derived from this document must cite The VirtualGL Project as the source of the material and list the current URL for the VirtualGL web site.</p> <p>The official VirtualGL binaries contain libjpeg-turbo, which is based in part on the work of the Independent JPEG Group.</p> <p>The VirtualGL server components include software developed by the <span class="remote"><a href="http://www.fltk.org/" class="remote">FLTK Project</a></span><a name="idx002"></a> and distributed under the terms of the <a href="LICENSE-FLTK.txt">FLTK License</a><a name="idx003"></a>.</p> <p>The package for the VirtualGL Client for Exceed includes <span class="remote"><a href="http://www.chiark.greenend.org.uk/~sgtatham/putty/" class="remote">PuTTY</a></span><a name="idx004"></a>, which is released under <a href="LICENSE-PuTTY.txt">this license</a><a name="idx005"></a>.</p> <p>VirtualGL includes portions of <span class="remote"><a href="http://www.x.org" class="remote">X.org</a></span><a name="idx006"></a>, which is released under <a href="LICENSE-xauth.txt">this license</a><a name="idx007"></a>.</p> <p>VirtualGL is licensed under the <a href="LICENSE.txt">wxWindows Library License, v3.1</a><a name="idx008"></a>, a derivative of the <a href="LGPL.txt">GNU Lesser General Public License (LGPL), v2.1</a><a name="idx009"></a>.</p> <p><br /></p> <hr class="break" /> <h1 id="hd002"><a name="file002"></a>2 Conventions Used in This Document</h1> <p>This document assumes that VirtualGL will be installed in the default directory (<code>/opt/VirtualGL</code> on Un*x and Mac systems and <code>c:\Program Files\VirtualGL-</code><em><code>{version}</code></em><code>-</code><em><code>{build}</code></em> on Windows systems, where <em><code>{version}</code></em> and <em><code>{build}</code></em> are the version and build number of VirtualGL, respectively.) If your installation of VirtualGL resides in a different directory, then adjust the instructions accordingly.</p> <p><br /></p> <hr class="break" /> <h1 id="hd003"><a name="file003"></a>3 Overview</h1> <p><a name="Overview"></a></p> <p>VirtualGL is an open source toolkit that gives any Unix or Linux remote display software the ability to run OpenGL applications with full 3D hardware acceleration. Some remote display solutions cannot be used with OpenGL applications at all. Others force OpenGL applications to use a slow, software-only renderer, to the detriment of performance as well as compatibility. The traditional method of displaying OpenGL applications to a remote X server (indirect rendering) supports 3D hardware acceleration, but this approach causes all of the OpenGL commands and 3D data to be sent over the network to be rendered on the client machine. This is not a tenable proposition unless the data is relatively small and static, unless the network is very fast, and unless the OpenGL application is specifically tuned for a remote X-Windows environment.</p> <p>With VirtualGL, the OpenGL commands and 3D data are instead redirected to a 3D graphics accelerator (AKA “graphics processing unit” or “GPU”) in the application server, and only the rendered 3D images are sent to the client machine. VirtualGL thus “virtualizes” 3D graphics hardware, allowing it to be co-located in the “cold room” with compute and storage resources. VirtualGL also allows GPUs to be shared among multiple users, and it provides “workstation-like” levels of performance on even the most modest of networks. This makes it possible for large, noisy, hot 3D workstations to be replaced with laptops or even thinner clients. More importantly, however, VirtualGL eliminates the workstation and the network as barriers to data size. Users can now visualize huge amounts of data in real time without needing to copy any of the data over the network or sit in front of the machine that is rendering the data.</p> <p>Normally, a Unix OpenGL application would send all of its drawing commands and data, both 2D and 3D, to an X-Windows server, which may be located across the network from the application server. VirtualGL, however, employs a technique called “split rendering” to force the 3D commands and data from the application to go to a GPU in the application server. VGL accomplishes this by pre-loading a dynamic shared object (DSO) into the OpenGL application at run time. This DSO intercepts a handful of GLX, OpenGL, and X11 commands necessary to perform split rendering. When the application attempts to use an X window for OpenGL rendering, VirtualGL intercepts the request, creates a corresponding 3D pixel buffer (“Pbuffer”) in video memory on the application server, and uses the Pbuffer for OpenGL rendering instead. When the application swaps the OpenGL drawing buffers or flushes the OpenGL command buffer to indicate that it has finished rendering a frame, VirtualGL reads back the pixels from the Pbuffer and sends them to the client.</p> <p>The beauty of this approach is its non-intrusiveness. VirtualGL monitors a few X11 commands and events to determine when windows have been resized, etc., but it does not interfere in any way with the delivery of 2D X11 commands to the X server. For the most part, VGL does not interfere with the delivery of OpenGL commands to the GPU, either (there are some exceptions, such as the implementation of color index rendering.) VGL merely forces the OpenGL commands to be delivered to a GPU that is attached to a different X server (the “3D X server”) than the X server to which the 2D drawing commands are delivered (the “2D X server.”) Once the OpenGL rendering has been redirected to a Pbuffer, everything (including esoteric OpenGL extensions, fragment/vertex programs, etc.) should “just work.” If an application runs locally on a 3D server/workstation, then that same application should run remotely from that same machine using VirtualGL.</p> <p>VirtualGL has two built-in “image transports” that can be used to send rendered 3D images to the client machine:</p> <p><a name="VGL_Transport"></a></p> <dl class="Description"> <dt class="Description-1 Description">1. VGL Transport</dt> <dd class="Description-1 Description"> The VGL Transport is most often used whenever the 2D X server (the X server used to draw the application’s GUI and transmit keyboard and mouse events back to the application server) is located across the network from the application server, for instance if the 2D X server is running on the user’s desktop machine. VirtualGL uses its own protocol on a dedicated TCP socket to send the rendered 3D images to the client machine, and the VirtualGL Client application decodes the images and composites them into the appropriate X window. The VGL Transport can either deliver uncompressed images (RGB-encoded), or it can compress images in real time using a high-speed JPEG codec. It also supports the delivery of stereo image pairs, which can be reconstructed into a stereo image by the VirtualGL Client. </dd> </dl> <a name="fig003001"></a> <div class="figure"> <p class="caption">Figure 3.1: The VGL Transport with a Remote 2D X Server</p> <img src="vgltransport.png" alt="vgltransport" class="figure" id="imgid_9" name="imgid_9"/> </div> <p><a name="X11_Transport"></a></p> <dl class="Description"> <dt class="Description-1 Description">2. X11 Transport</dt> <dd class="Description-1 Description"> The X11 Transport simply draws the rendered 3D images into the appropriate X window using <code>XPutImage()</code> and similar X-Windows commands. This is most useful in conjunction with an “X proxy”, which can be one of any number of Unix remote display applications, such as VNC. These X proxies are essentially “virtual” X servers. They appear to the application as a normal X server, but they perform X11 rendering to a virtual framebuffer in main memory rather than to a real framebuffer (video memory.) This allows the X proxy to send only images to the client machine rather than fine-grained X-Windows rendering commands. When using the X11 Transport, VirtualGL does not perform any image compression or encoding itself. It instead relies upon an X proxy to encode and deliver the images to the client(s). Since the use of an X proxy eliminates the need to send X-Windows commands over the network, this is the best means of using VirtualGL over high-latency or low-bandwidth networks. </dd> </dl> <a name="fig003002"></a> <div class="figure"> <p class="caption">Figure 3.2: The X11 Transport with an X Proxy</p> <img src="x11transport.png" alt="x11transport" class="figure" id="imgid_10" name="imgid_10"/> </div> <p>VirtualGL also provides an API that can be used to develop custom image transport plugins.</p> <p><br /></p> <hr class="break" /> <h1 id="hd004"><a name="file004"></a>4 System Requirements</h1> <h2 id="hd004001">4.1 Linux/x86 and Other x86 Un*x Operating Systems</h2> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard"></th> <th class="head standard">Server (x86)</th> <th class="head standard">Server (x86-64)</th> <th class="head standard">Client</th> </tr> </thead> <tr class="standard"> <td class="high standard">Recommended CPU</td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> For optimal performance, the CPU should support SSE2 extensions. </li> <li class="Itemize-0"> Dual processors or dual cores recommended </li></ul></td> <td class="standard">Dual processors or dual cores recommended</td> <td class="standard">For optimal performance, the CPU should support SSE2 extensions.</td> </tr> <tr class="standard"> <td class="high standard">Graphics</td> <td class="standard" colspan="2">AMD or nVidia GPU <ul class="Itemize"><li class="Itemize-0"> For optimal performance, particularly with multiple simultaneous users, a professional-grade GPU such as the AMD FirePro or nVidia Quadro is recommended. </li> <li class="Itemize-0"> Install the AMD or nVidia proprietary drivers. Open source drivers for these GPUs do not generally provide full 3D acceleration, and some of those drivers do not provide Pbuffer support. </li></ul></td> <td class="standard">Any graphics adapter with decent 2D performance <ul class="Itemize"><li class="Itemize-0"> If using a 3D graphics adapter, install the vendor drivers for that 3D graphics adapter. </li></ul></td> </tr> <tr class="standard"> <td class="high standard">Other Software</td> <td class="standard" colspan="3">X server configured to export True Color (24-bit or 32-bit) visuals</td> </tr> </table> </div> <h2 id="hd004002">4.2 Mac/x86</h2> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard"></th> <th class="head standard">Client</th> </tr> </thead> <tr class="standard"> <td class="high standard">Recommended CPU</td> <td class="standard">Any Intel-based Mac</td> </tr> <tr class="standard"> <td class="high standard">O/S</td> <td class="standard">OS X 10.4 (“Tiger”) or later</td> </tr> <tr class="standard"> <td class="high standard">Other Software</td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> <em>VGL Transport Only</em>: Mac X11 application (in the “Optional Installs” package on the OS X 10.7 and earlier install discs) or <span class="remote"><a href="http://xquartz.macosforge.org" class="remote">XQuartz</a></span><a name="idx0010"></a> </li></ul></td> </tr> </table> </div> <h2 id="hd004003">4.3 Windows</h2> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard"></th> <th class="head standard">Client</th> </tr> </thead> <tr class="standard"> <td class="high standard">Recommended CPU</td> <td class="standard">For optimal performance, the CPU should support SSE2 extensions.</td> </tr> <tr class="standard"> <td class="high standard">Graphics</td> <td class="standard">Any graphics adapter with decent 2D performance</td> </tr> <tr class="standard"> <td class="high standard">O/S</td> <td class="standard">Windows 2000 or later</td> </tr> <tr class="standard"> <td class="high standard">Other Software</td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> <em>VGL Transport Only</em>: <span class="remote"><a href="http://x.cygwin.com" class="remote">Cygwin/X</a></span><a name="idx0011"></a> or <span class="remote"><a href="http://connectivity.opentext.com" class="remote">OpenText</a></span><a name="idx0012"></a> Exceed v8 or newer </li> <li class="Itemize-0"> Client display must have a 24-bit or 32-bit color depth (True Color.) </li></ul></td> </tr> </table> </div> <h2 id="hd004004">4.4 Additional Requirements for Stereographic Rendering</h2> <p><a name="Stereo_Requirements"></a></p> <div class="important"><p class="important"> The client requirements do not apply to anaglyphic stereo. See Chapter <a href="#Advanced_OpenGL" class="ref">17</a> for more details. </p></div> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard"></th> <th class="head standard">Server</th> <th class="head standard">Client</th> </tr> </thead> <tr class="standard"> <td class="high standard">Linux/Unix</td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> AMD or nVidia GPU that supports stereo (examples: AMD FirePro W Series, nVidia Quadro) </li> <li class="Itemize-0"> The 3D X server should be configured to export stereo visuals. </li></ul></td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> GPU that supports stereo </li> <li class="Itemize-0"> The 2D X server should be configured to export stereo visuals. </li></ul></td> </tr> <tr class="standard"> <td class="high standard">Mac/x86</td> <td class="standard">N/A</td> <td class="standard">GPU that supports stereo (example: nVidia Quadro)</td> </tr> <tr class="standard"> <td class="high standard">Windows</td> <td class="standard">N/A</td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> GPU that supports stereo. Stereo support should be enabled in the 3D driver configuration. </li> <li class="Itemize-0"> 2D X server that supports and is configured to export stereo visuals. (As of this writing, <span class="remote"><a href="http://connectivity.opentext.com" class="remote">OpenText</a></span><a name="idx0013"></a> Exceed 3D v8 or newer is the only known Windows X server solution that supports stereo drawing with sufficient performance to be used with VirtualGL.) </li></ul></td> </tr> </table> </div> <h2 id="hd004005">4.5 Additional Requirements for Transparent Overlays</h2> <p><a name="Overlay_Requirements"></a></p> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard"></th> <th class="head standard">Client</th> </tr> </thead> <tr class="standard"> <td class="high standard">Linux/Unix</td> <td class="standard" rowspan="2">GPU that supports transparent overlays. The 2D X server should be configured to export overlay visuals.</td> </tr> <tr class="standard"> <td class="high standard">Mac/x86</td> </tr> <tr class="standard"> <td class="high standard">Windows</td> <td class="standard"><ul class="Itemize"><li class="Itemize-0"> GPU that supports transparent overlays. Overlay support should be enabled in the 3D driver configuration. </li> <li class="Itemize-0"> 2D X server that supports and is configured to export transparent overlay visuals. (As of this writing, <span class="remote"><a href="http://connectivity.opentext.com" class="remote">OpenText</a></span><a name="idx0014"></a> Exceed 3D v8 or newer is the only known Windows X server solution that supports transparent overlays.) </li></ul></td> </tr> </table> </div> <p><br /></p> <hr class="break" /> <h1 id="hd005"><a name="file005"></a>5 Obtaining and Installing VirtualGL</h1> <div class="important"><p class="important"> VirtualGL must be installed on any machine that will act as a VirtualGL server or as a client for the <a href="#VGL_Transport">VGL Transport</a><a name="idx0015"></a>. It is not necessary to install VirtualGL on the client machine if using VNC or another type of X proxy. </p></div> <h2 id="hd005001">5.1 Installing VirtualGL on Linux</h2> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered" value="1"> Download the appropriate VirtualGL binary package for your system from the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl/files/VirtualGL/" class="remote">Files area</a></span><a name="idx0016"></a> of the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl" class="remote">VirtualGL SourceForge project page</a></span><a name="idx0017"></a>. Packages are provided for Linux distributions in the <span class="remote"><a href="http://www.redhat.com/" class="remote">Red Hat</a></span><a name="idx0018"></a>, <span class="remote"><a href="http://www.novell.com/linux/" class="remote">SuSE</a></span><a name="idx0019"></a>, or <span class="remote"><a href="http://www.debian.org/" class="remote">Debian</a></span><a name="idx0020"></a> families that contain GLIBC 2.3.2 or later (including <span class="remote"><a href="http://fedoraproject.org/" class="remote">Fedora</a></span><a name="idx0021"></a>, <span class="remote"><a href="http://www.centos.org/" class="remote">CentOS</a></span><a name="idx0022"></a> 3 or later, <span class="remote"><a href="http://www.oracle.com/us/technologies/linux/" class="remote">Oracle Linux</a></span><a name="idx0023"></a>, and <span class="remote"><a href="http://www.ubuntu.com/" class="remote">Ubuntu Linux</a></span><a name="idx0024"></a> v6.0 or later.) <br /> <div class="important"><p class="important"> If you want to run both 32-bit and 64-bit OpenGL applications using VirtualGL on 64-bit Linux systems, then you will need to install both the i386 and x86_64 VirtualGL RPMs or both the “virtualgl” and “virtualgl32” amd64 DEBs. (The virtualgl32 DEB is a supplementary package that contains only the 32-bit server components.) </p></div> </li> <li class="Ordered-1 Ordered" value="2"> Log in as root, cd to the directory where you downloaded the binary package, and issue the following commands: <dl class="Description"> <dt class="Description-3 Description">RPM-based systems</dt> <dd class="Description-3 Description"> <pre class="verbatim"> rpm -e VirtualGL rpm -i VirtualGL*.rpm </pre> <div class="important"><p class="important"> You may need to add <code>--all-matches</code> to the <code>rpm -e</code> command line if you have installed both the 32-bit and 64-bit VirtualGL RPMs. </p></div> </dd> <dt class="Description-3 Description">Debian-based systems</dt> <dd class="Description-3 Description"> <pre class="verbatim"> dpkg -i virtualgl*.deb </pre> </dd> </dl> </li> </ol> <h2 id="hd005002">5.2 Installing the VirtualGL Client on OS X</h2> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Download the VirtualGL Mac disk image (<code>VirtualGL-</code><em><code>{version}</code></em><code>.dmg</code>) from the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl/files/VirtualGL/" class="remote">Files area</a></span><a name="idx0025"></a> of the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl" class="remote">VirtualGL SourceForge project page</a></span><a name="idx0026"></a>. </li> <li class="Ordered-1 Ordered"> Open the disk image, then open <code>VirtualGL-</code><em><code>{version}</code></em><code>.pkg</code> inside the disk image. Follow the instructions to install the Mac client. </li> </ol> <h2 id="hd005003">5.3 Installing the VirtualGL Client on Windows (Exceed)</h2> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Download the VirtualGL Client for Exceed installer package (<code>VirtualGL[64]-</code><em><code>{version}</code></em><code>-exceed.exe</code>) from the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl/files/VirtualGL/" class="remote">Files area</a></span><a name="idx0027"></a> of the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl" class="remote">VirtualGL SourceForge project page</a></span><a name="idx0028"></a>. </li> <li class="Ordered-1 Ordered"> Run the installer. The installation of the VirtualGL Client should be self-explanatory. The only configuration option is the directory into which you want the files to be installed. </li> </ol> <div class="important"><p class="important"> NOTE: The VirtualGL Client for Exceed installer does not remove any previous versions of the VirtualGL Client that may be installed on your machine. If you wish, you can remove these older versions manually by using the “Add or Remove Programs” applet in the Control Panel (or the “Programs and Features” applet if you are running Windows Vista.) </p></div> <h2 id="hd005004">5.4 Installing the VirtualGL Client on Windows (Cygwin/X)</h2> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Make sure that the following Cygwin packages are installed:<br /><br /> libGL<br /> libGLU<br /> libstdc++<br /> libX11<br /> libXext<br /> openssh<br /> xorg-xserver<br /> xauth<br /> </li> <li class="Ordered-1 Ordered"> Download the VirtualGL Cygwin package (<code>VirtualGL-</code><em><code>{version}</code></em><code>-cygwin.tar.bz2</code>, or <code>VirtualGL-</code><em><code>{version}</code></em><code>-cygwin64.tar.bz2</code> if you are running Cygwin64) from the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl/files/VirtualGL/" class="remote">Files area</a></span><a name="idx0029"></a> of the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl" class="remote">VirtualGL SourceForge project page</a></span><a name="idx0030"></a>. </li> <li class="Ordered-1 Ordered"> Run the Cygwin Setup application (the same application you used to install Cygwin.) </li> <li class="Ordered-1 Ordered"> On the “Choose a Download Source” page, select “Install from Local Directory”. </li> <li class="Ordered-1 Ordered"> On the “Select Root Install Directory” page, use the same options that you used when installing Cygwin. </li> <li class="Ordered-1 Ordered"> On the “Select Local Package Directory” page, enter the directory containing the VirtualGL Cygwin package. </li> <li class="Ordered-1 Ordered"> On the “Select Packages” page, change “View” to “Pending” and verify that the VirtualGL package with the correct version is selected for install. Click “Next>” to install. </li> </ol> <h2 id="hd005005">5.5 Installing VirtualGL from Source</h2> <p>If you are using a platform for which there is not a pre-built VirtualGL binary package available, then download the VirtualGL source tarball (<code>VirtualGL-</code><em><code>{version}</code></em><code>.tar.gz</code>) from the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl/files/VirtualGL/" class="remote">Files area</a></span><a name="idx0031"></a> of the <span class="remote"><a href="http://sourceforge.net/projects/virtualgl" class="remote">VirtualGL SourceForge project page</a></span><a name="idx0032"></a>, uncompress it, <code>cd VirtualGL-</code><em><code>{version}</code></em>, and read the contents of <code>BUILDING.txt</code> for further instructions on how to build and install VirtualGL from source.</p> <h2 id="hd005006">5.6 Uninstalling VirtualGL</h2> <h3 id="hd005006001">Linux</h3> <p>As root, issue one of the following commands:</p> <dl class="Description"> <dt class="Description-1 Description">RPM-based systems</dt> <dd class="Description-1 Description"> <pre class="verbatim"> rpm -e VirtualGL </pre> <div class="important"><p class="important"> You may need to add <code>--all-matches</code> to the RPM command line if you have installed both the 32-bit and 64-bit VirtualGL RPMs. </p></div> </dd> <dt class="Description-1 Description">Debian-based systems</dt> <dd class="Description-1 Description"> <pre class="verbatim"> dpkg -r virtualgl </pre> If you have also installed the 32-bit supplementary package: <pre class="verbatim"> dpkg -r virtualgl32 </pre> </dd> </dl> <h3 id="hd005006002">OS X</h3> <p>Use the “Uninstall VirtualGL” application provided in the VirtualGL disk image, or issue the following command from the Terminal:</p> <pre class="verbatim"> sudo /opt/VirtualGL/bin/uninstall </pre> <h3 id="hd005006003">Windows (Exceed)</h3> <p>Use the “Programs and Features” applet in the Control Panel (or the “Add or Remove Programs” applet if you are running Windows XP), or select “Uninstall VirtualGL Client” in the “VirtualGL Client” Start Menu group.</p> <h3 id="hd005006004">Windows (Cygwin/X)</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Run the Cygwin Setup application (the same application you used to install Cygwin.) </li> <li class="Ordered-1 Ordered"> On the “Choose a Download Source” page, select “Install from Local Directory”. </li> <li class="Ordered-1 Ordered"> On the “Select Root Install Directory” page, use the same options that you used when installing Cygwin. </li> <li class="Ordered-1 Ordered"> On the “Select Local Package Directory” page, enter the directory containing the VirtualGL Cygwin package. </li> <li class="Ordered-1 Ordered"> On the “Select Packages” page, change “View” to “Full”, find the “VirtualGL” package in the list, and change its status from “Keep” to “Uninstall”. Click “Next>” to uninstall. </li> </ol> <p><br /></p> <hr class="break" /> <h1 id="hd006"><a name="file006"></a>6 Configuring a Linux or Unix Machine as a VirtualGL Server</h1> <p><a name="Unix_Config"></a></p> <h2 id="hd006001">6.1 Granting Access to the 3D X Server</h2> <p>VirtualGL requires access to a GPU in the application server so that it can create off-screen pixel buffers (Pbuffers) and redirect the 3D rendering from X windows into these Pbuffers. Unfortunately, accessing a GPU on Linux and Unix systems requires going through an X server. On such systems, the only way to share the application server’s GPU(s) among multiple users is to grant those users access to the 3D X server (the X server attached to the GPU(s). Refer to the figures in Chapter <a href="#Overview" class="ref">3</a>.)</p> <p>It is important to understand the security risks associated with this. Once a user has access to the 3D X server, there is nothing that would prevent the user from logging keystrokes or reading back images from that X server. Using <code>xauth</code>, one can obtain “untrusted” X authentication keys that prevent such exploits, but unfortunately, those untrusted keys also disallow access to the 3D hardware. Thus, it is necessary to grant full, trusted access to the 3D X server for any users that will need to run VirtualGL. Unless you fully trust the users to whom you are granting this access, then you should avoid logging in locally to the 3D X server (particularly as root) unless absolutely necessary.</p> <p>This section will explain how to configure a VirtualGL server such that selected users can run VirtualGL, even if the server is sitting at the login prompt.</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Shut down the display manager: <dl class="Description"> <dt class="Description-3 Description">Ubuntu Linux servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/gdm stop </pre> </dd> <dt class="Description-3 Description">Ubuntu Linux servers running LightDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/lightdm stop </pre> </dd> <dt class="Description-3 Description">SuSE Linux servers</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/xdm stop </pre> </dd> <dt class="Description-3 Description">Red Hat/Fedora Linux servers</dt> <dd class="Description-3 Description"> <pre class="verbatim"> init 3 </pre> </dd> <dt class="Description-3 Description">Solaris 10 servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> svcadm disable gdm2-login </pre> </dd> <dt class="Description-3 Description">Solaris 11/OpenSolaris servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> svcadm disable gdm </pre> </dd> <dt class="Description-3 Description">Solaris servers running dtlogin</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/dtlogin stop </pre> </dd> <dt class="Description-3 Description">FreeBSD servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /usr/local/etc/rc.d/gdm stop </pre> </dd> <dt class="Description-3 Description">FreeBSD servers running KDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /usr/local/kde4/etc/rc.d/kdm4 stop </pre> </dd> </dl> </li> <li class="Ordered-1 Ordered"> Log in as root from the text console (or remotely using SSH.) </li> <li class="Ordered-1 Ordered"> Run <pre class="verbatim"> /opt/VirtualGL/bin/vglserver_config </pre> </li> <li class="Ordered-1 Ordered"> Select option 1 (<code>Configure server for use with VirtualGL in GLX mode</code>.) </li> <li class="Ordered-1 Ordered"> <pre class="verbatim"> Restrict 3D X server access to vglusers group (recommended)? [Y/n] </pre> <dl class="Description"> <dt class="Description-3 Description">Yes</dt> <dd class="Description-3 Description"> Only users in the <code>vglusers</code> group can use VirtualGL (the configuration script will create the <code>vglusers</code> group if it doesn’t already exist.) This is the most secure option, since it prevents any users outside of the <code>vglusers</code> group from accessing (and thus exploiting) the 3D X server. </dd> <dt class="Description-3 Description">No</dt> <dd class="Description-3 Description"> VirtualGL can be used by any user that successfully logs into the VirtualGL server. The 3D X server can also be accessed (and potentially exploited) by any user who is logged into the VirtualGL server. If you choose this option, it is recommended that you also disable the XTEST extension (see below.) </dd> </dl> </li> <li class="Ordered-1 Ordered"> <pre class="verbatim"> Restrict framebuffer device access to vglusers group (recommended)? [Y/n] </pre> <dl class="Description"> <dt class="Description-3 Description">Yes</dt> <dd class="Description-3 Description"> Only users in the <code>vglusers</code> group can run OpenGL applications on the VirtualGL server (the configuration script will create the <code>vglusers</code> group if it doesn’t already exist.) This limits the possibility that an unauthorized user could snoop the 3D framebuffer device(s) and thus see (or alter) the output of a 3D application that is being used with VirtualGL. </dd> <dt class="Description-3 Description">No</dt> <dd class="Description-3 Description"> Any authenticated user can run OpenGL applications on the VirtualGL server. If it is necessary for users outside of the <code>vglusers</code> group to log in locally to this server and run OpenGL applications, then this option must be selected. </dd> </dl> </li> <li class="Ordered-1 Ordered"> <pre class="verbatim"> Disable XTEST extension (recommended)? [Y/n] </pre> <dl class="Description"> <dt class="Description-3 Description">Yes</dt> <dd class="Description-3 Description"> Disabling XTEST will not prevent a user from logging keystrokes or reading images from the 3D X server, but if a user has access to the 3D X server, disabling XTEST will prevent them from inserting keystrokes or mouse events and thus hijacking local X sessions on that X server. <div class="important"><p class="important"> Certain Linux distributions do not have the X server command-line entries in their GDM configuration files. On these distributions, it will be necessary to run <code>gdmsetup</code> and manually add an argument of <code>-tst</code> to the X server command line to disable XTEST for the first time. After this, <code>vglserver_config</code> should be able to disable and enable XTEST properly. This is known to be necessary for openSUSE 10 and Red Hat Enterprise Linux 5. </p></div> </dd> <dt class="Description-3 Description">No</dt> <dd class="Description-3 Description"> <code>x11vnc</code> and <code>x0vncserver</code> both require XTEST, so if you need to attach a VNC server to the 3D X server, then it is necessary to answer “No” (and thus leave XTEST enabled.) </dd> </dl> </li> <li class="Ordered-1 Ordered"> If you chose to restrict X server or framebuffer device access to the <code>vglusers</code> group, then edit <code>/etc/group</code> and add <code>root</code> to the <code>vglusers</code> group. If you choose, you can also add additional users to the group at this time. Note that any user you add to <code>vglusers</code> must log out and back in again before their new group permissions will take effect. </li> <li class="Ordered-1 Ordered"> Restart the display manager: <dl class="Description"> <dt class="Description-3 Description">Ubuntu Linux servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/gdm start </pre> </dd> <dt class="Description-3 Description">Ubuntu Linux servers running LightDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/lightdm start </pre> </dd> <dt class="Description-3 Description">SuSE Linux servers</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/xdm start </pre> </dd> <dt class="Description-3 Description">Red Hat/Fedora Linux servers</dt> <dd class="Description-3 Description"> <pre class="verbatim"> init 5 </pre> </dd> <dt class="Description-3 Description">Solaris 10 servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> svcadm enable gdm2-login </pre> </dd> <dt class="Description-3 Description">Solaris 11/OpenSolaris servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> svcadm enable gdm </pre> </dd> <dt class="Description-3 Description">Solaris servers running dtlogin</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /etc/init.d/dtlogin start </pre> </dd> <dt class="Description-3 Description">FreeBSD servers running GDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /usr/local/etc/rc.d/gdm start </pre> </dd> <dt class="Description-3 Description">FreeBSD servers running KDM</dt> <dd class="Description-3 Description"> <pre class="verbatim"> /usr/local/kde4/etc/rc.d/kdm4 start </pre> </dd> </dl> </li> </ol> <h3 id="hd006001001">Sanity Check</h3> <p>To verify that the application server is ready to run VirtualGL, log out of the server, log back into the server using SSH, and execute the following commands in the SSH session:</p> <dl class="Description"> <dt class="Description-1 Description">If you restricted 3D X server access to <code>vglusers</code></dt> <dd class="Description-1 Description"> <pre class="verbatim"> xauth merge /etc/opt/VirtualGL/vgl_xauth_key xdpyinfo -display :0 /opt/VirtualGL/bin/glxinfo -display :0 -c </pre> <div class="important"><p class="important"> NOTE: <code>xauth</code> and <code>xdpyinfo</code> are in <code>/usr/openwin/bin</code> on Solaris systems. </p></div> </dd> <dt class="Description-1 Description">If you did not restrict 3D X server access</dt> <dd class="Description-1 Description"> <pre class="verbatim"> xdpyinfo -display :0 /opt/VirtualGL/bin/glxinfo -display :0 -c </pre> </dd> </dl> <p>Both commands should output a list of visuals and should complete with no errors. If you chose to disable the XTEST extension, then check the output of <code>xdpyinfo</code> to verify that <code>XTEST</code> does not show up in the list of extensions.</p> <p>You should also examine the output of <code>glxinfo</code> to ensure that at least one of the visuals is 24-bit or 32-bit TrueColor and has Pbuffer support (the latter is indicated by a “P” in the last column.) Example:</p> <pre class="verbatim"> visual x bf lv rg d st colorbuffer ax dp st accumbuffer ms cav drw id dep cl sp sz l ci b ro r g b a F bf th cl r g b a ns b eat typ ------------------------------------------------------------------------------ 0x151 24 tc 0 32 0 r y . 8 8 8 0 . 4 24 8 16 16 16 16 0 0 None PXW </pre> <p>If none of the visuals has Pbuffer support, then this is most likely because there is no 3D acceleration, which is most likely because the correct 3D drivers are not installed (or are misconfigured.) Lack of 3D acceleration is also typically indicated by the word “Mesa” in the client GLX vendor string and/or the OpenGL vendor string, and the words “Software Rasterizer” in the OpenGL renderer string.</p> <h2 id="hd006002">6.2 Using VirtualGL with Multiple GPUs</h2> <p>VirtualGL can redirect the OpenGL commands from a 3D application to any GPU in the server machine. In order for this to work, however, all of the GPUs must be attached to different screens on the same X server or to different X servers. Attaching them to different screens is the easiest and most common approach, and this allows the GPUs to be individually addressed by setting <code>VGL_DISPLAY</code> to (or invoking <code>vglrun -d</code> with) <code>:0.0</code>, <code>:0.1</code>, <code>:0.2</code>, etc. If the GPUs are attached to different X servers, then they can be individually addressed by setting <code>VGL_DISPLAY</code> to (or invoking <code>vglrun -d</code> with) <code>:0.0</code>, <code>:1.0</code>, <code>:2.0</code>, etc.</p> <h2 id="hd006003">6.3 SSH Server Configuration</h2> <p>The application server’s SSH daemon should have the <code>X11Forwarding</code> option enabled and the <code>UseLogin</code> option disabled. This is configured in <code>sshd_config</code>, which is usually located under <code>/etc/ssh</code>.</p> <h2 id="hd006004">6.4 Un-Configuring the Server</h2> <p>You can use the <code>vglserver_config</code> script to restore the security settings that were in place before VirtualGL was installed. Option 2 (<code>Unconfigure server for use with VirtualGL in GLX mode</code>) will remove any shared access to the 3D X server and thus prevent VirtualGL from accessing the 3D hardware in that manner. Additionally, this option will re-enable the XTEST extension on the 3D X server and will restore the framebuffer device permissions to their default (by default, only root or the user that is currently logged into the application server locally can access the framebuffer devices.)</p> <div class="important"><p class="important"> NOTE: Unconfiguring the server does not remove the <code>vglusers</code> group. </p></div> <p>After selecting Option 2, you must restart the display manager before the changes will take effect.</p> <p><br /></p> <hr class="break" /> <h1 id="hd007"><a name="file007"></a>7 Configuring a Windows Machine as a VGL Transport Client</h1> <h2 id="hd007001">7.1 Configuring and Optimizing Exceed</h2> <p>If using the VirtualGL Client for Exceed, then add the Exceed path (example: <code>C:\Program Files\Hummingbird\Connectivity\14.00\Exceed</code>) to the system <code>PATH</code> environment if it isn’t already there.</p> <h3 id="hd007001001">Disabling Pixel Format Conversion (Exceed 2006 and earlier)</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Load Exceed XConfig (right-click on the Exceed taskbar icon, then select <em>Tools–>Configuration</em>.) </li> <li class="Ordered-1 Ordered"> Open the “X Server Protocol” applet in XConfig. <div class="important"><p class="important"> If you are using the “Classic View” mode of XConfig, open the “Protocol” applet instead. </p></div> </li> <li class="Ordered-1 Ordered"> In the “X Server Protocol” applet, select the “Protocol” tab and make sure that “Use 32 bits per pixel for true color” is not checked. <br /><br /> <img src="exceed1.png" alt="exceed1" class="inline" id="imgid_1" name="imgid_1"/> </li> <li class="Ordered-1 Ordered"> Click “Validate and Apply Changes.” If XConfig asks whether you want to perform a server reset, click “Yes.” </li> </ol> <h3 id="hd007001002">Disabling Backing Store (Exceed 2008 and earlier)</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Load Exceed XConfig (right-click on the Exceed taskbar icon, then select <em>Tools–>Configuration</em>.) </li> <li class="Ordered-1 Ordered"> Open the “Other Server Settings” applet in XConfig. <div class="important"><p class="important"> If you are using the “Classic View” mode of XConfig, open the “Performance” applet instead. </p></div> </li> <li class="Ordered-1 Ordered"> Select the “Performance” tab and make sure that “Default Backing Store” is set to “None.” <br /><br /> <img src="exceed3.png" alt="exceed3" class="inline" id="imgid_2" name="imgid_2"/> </li> <li class="Ordered-1 Ordered"> Click “Validate and Apply Changes.” If XConfig asks whether you want to perform a server reset, click “Yes.” </li> </ol> <h3 id="hd007001003">Enabling MIT-SHM</h3> <p>VirtualGL has the ability to take advantage of the MIT-SHM extension in OpenText Exceed to accelerate image drawing on Windows. This can significantly improve the end-to-end performance of the VGL Transport when running over a local-area network.</p> <p>The bad news is that this extension is not consistently implemented across all versions of Exceed. In particular, Exceed 8, Exceed 9, and Exceed 2008 require patches to make it work properly. If you are using one of these versions of Exceed, you will need to obtain the following patches from the OpenText support site:</p> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard">Product</th> <th class="head standard">Patches Required</th> <th class="head standard">How to Obtain</th> </tr> </thead> <tr class="standard"> <td class="standard">Exceed 8.0</td> <td class="standard"><code>hclshm.dll</code> v9.0.0.1 (or higher)<br /> <code>xlib.dll</code> v9.0.0.3 (or higher)<br /> <code>exceed.exe</code> v8.0.0.28 (or higher)</td> <td class="standard">Download all patches from the <span class="remote"><a href="http://support.hummingbird.com/customer/cspatches/patches.asp" class="remote">OpenText support site</a></span><a name="idx0033"></a>. <br /> (<em>OpenText WebSupport account required</em>)</td> </tr> <tr class="standard"> <td class="standard">Exceed 9.0</td> <td class="standard"><code>hclshm.dll</code> v9.0.0.1 (or higher)<br /> <code>xlib.dll</code> v9.0.0.3 (or higher)<br /> <code>exceed.exe</code> v9.0.0.9 (or higher)</td> <td class="standard"><code>exceed.exe</code> can be patched by running Hummingbird Update.<br /> <br /> All other patches must be downloaded from the <span class="remote"><a href="http://support.hummingbird.com/customer/cspatches/patches.asp" class="remote">OpenText support site</a></span><a name="idx0034"></a>. <br /> (<em>OpenText WebSupport account required</em>)</td> </tr> <tr class="standard"> <td class="standard">Exceed 2008</td> <td class="standard"><code>xlib.dll</code> v13.0.1.235 (or higher)<br /> (or install the latest <span class="remote"><a href="http://support.hummingbird.com/customer/cspatches/patches.asp" class="remote">Connectivity 2008 Service Pack</a></span><a name="idx0035"></a>)</td> <td class="standard">Download all patches from the <span class="remote"><a href="http://support.hummingbird.com/customer/cspatches/patches.asp" class="remote">OpenText support site</a></span><a name="idx0036"></a>. <br /> (<em>OpenText WebSupport account required</em>)</td> </tr> </table> </div> <p>No patches should be necessary for Exceed 10, 2006, 2007, or 14.</p> <p>Next, you need to enable the MIT-SHM extension in Exceed:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Load Exceed XConfig (right-click on the Exceed taskbar icon, then select <em>Tools–>Configuration</em>.) </li> <li class="Ordered-1 Ordered"> Open the “X Server Protocol” applet in XConfig. <div class="important"><p class="important"> If you are using the “Classic View” mode of XConfig, open the “Protocol” applet instead. </p></div> </li> <li class="Ordered-1 Ordered"> Select the “Extensions” tab and make sure that “MIT-SHM” is checked. <br /><br /> <img src="exceed2.png" alt="exceed2" class="inline" id="imgid_3" name="imgid_3"/> </li> <li class="Ordered-1 Ordered"> Click “Validate and Apply Changes.” If XConfig asks whether you want to perform a server reset, click “Yes.” </li> </ol> <h2 id="hd007002">7.2 Optimizing Cygwin/X</h2> <p>VirtualGL has the ability to take advantage of the MIT-SHM extension in Cygwin/X to accelerate image drawing on Windows. This can significantly improve the overall performance of the VirtualGL pipeline when running over a local-area network.</p> <p>To enable MIT-SHM in Cygwin/X:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Open a Cygwin Bash shell </li> <li class="Ordered-1 Ordered"> Run <code>cygserver-config</code> </li> <li class="Ordered-1 Ordered"> Answer “yes” when asked “Do you want to install cygserver as service?” </li> <li class="Ordered-1 Ordered"> From a Windows (not Cygwin) command prompt, run <code>net start cygserver</code> </li> <li class="Ordered-1 Ordered"> Add <code>server</code> to the <code>CYGWIN</code> system environment variable (create this environment variable if it doesn’t already exist) </li> <li class="Ordered-1 Ordered"> Start or re-start Cygwin/X </li> <li class="Ordered-1 Ordered"> Run <code>xdpyinfo</code> and verify that <code>MIT-SHM</code> appears in the list of X extensions </li> </ol> <p><br /></p> <hr class="break" /> <h1 id="hd008"><a name="file008"></a>8 Using VirtualGL with the VGL Transport</h1> <p><a name="VGL_Transport_Usage"></a></p> <h3 id="hd008000001">Advantages of the VGL Transport</h3> <ul class="Itemize"> <li class="Itemize-1 Itemize asterisk"> Seamless windows; every application window appears as a separate window on the user’s desktop </li> <li class="Itemize-1 Itemize asterisk"> Supports stereographic rendering (see requirements in Sections <a href="#Stereo_Requirements" class="ref">4.4</a> and <a href="#Overlay_Requirements" class="ref">4.5</a>) </li> <li class="Itemize-1 Itemize asterisk"> Consumes less memory and CPU time on the server, since the 2D (X11) rendering occurs on the client machine </li> </ul> <h3 id="hd008000002">Disadvantages of the VGL Transport</h3> <ul class="Itemize"> <li class="Itemize-1 Itemize asterisk"> The VGL Transport is designed to be used with remote X servers, and thus it relies on the chatty remote X11 protocol to send the 2D elements of the application’s GUI to the user’s desktop. As a result, the VGL Transport is not recommended for use on high-latency or low-bandwidth networks </li> <li class="Itemize-1 Itemize asterisk"> No collaboration features </li> <li class="Itemize-1 Itemize asterisk"> The client is not stateless. As with any remote X11 environment, if the network connection drops, then the application will exit </li> </ul> <h2 id="hd008001">8.1 VGL Transport with X11 Forwarding</h2> <p><a name="X11_Forwarding"></a></p> <p>This mode is recommended for use only on secure local-area networks. The X11 traffic is encrypted, but the VGL Transport is left unencrypted to maximize performance.</p> <h3 id="hd008001001">Procedure for Linux/Unix/Mac/Cygwin Clients</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Start the X server if it isn’t started already.<br /> <em>Mac clients:</em> start the Mac X11 application or XQuartz.<br /> <em>Cygwin clients:</em> start Cygwin/X. </li> <li class="Ordered-1 Ordered"> Open a new terminal window.<br /> <em>Mac clients:</em> in the X11 application, start a new xterm [Command-N] if one isn’t already started.<br /> <em>Cygwin clients:</em> start a new xterm if one isn’t already started (right-click on the Cygwin/X taskbar icon, then select <em>Applications–>xterm</em>.) </li> <li class="Ordered-1 Ordered"> In the same terminal/xterm window, open a Secure Shell (SSH) session into the VirtualGL server: <pre class="verbatim"> /opt/VirtualGL/bin/vglconnect {user}@{server} </pre> Replace <em><code>{user}</code></em> with your username on the VirtualGL server and <em><code>{server}</code></em> with the hostname or IP address of that server. </li> <li class="Ordered-1 Ordered"> In the SSH session, start a 3D application using VirtualGL: <pre class="verbatim"> /opt/VirtualGL/bin/vglrun [vglrun options] {application_executable_or_script} {arguments} </pre> Consult Chapter <a href="#Advanced_Configuration" class="ref">20</a> for more information on <code>vglrun</code> command-line options. </li> </ol> <h3 id="hd008001002">Procedure for Windows Clients Running Exceed</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Start Exceed if it isn’t already started. Hover the mouse pointer over the Exceed taskbar icon and make a note of the display number on which Exceed is listening (<em>Example:</em> “Exceed 0.0 Multiwindow Mode”.) </li> <li class="Ordered-1 Ordered"> Open a new Command Prompt. </li> <li class="Ordered-1 Ordered"> In the same Command Prompt window, set the <code>DISPLAY</code> environment variable to match the display on which Exceed is listening. Example: <pre class="verbatim"> set DISPLAY=:0.0 </pre> <div class="important"><p class="important"> If you only ever plan to use one Exceed session at a time, then you can set the <code>DISPLAY</code> environment variable in your user environment, which is configured through the Windows Control Panel. </p></div> </li> <li class="Ordered-1 Ordered"> Open a Secure Shell (SSH) session into the VirtualGL server: <pre class="verbatim"> cd /d "c:\program files\virtualgl-{version}-{build}" vglconnect {user}@{server} </pre> Replace <em><code>{user}</code></em> with your username on the VirtualGL server and <em><code>{server}</code></em> with the hostname or IP address of that server. </li> <li class="Ordered-1 Ordered"> In the SSH session, start a 3D application using VirtualGL: <pre class="verbatim"> /opt/VirtualGL/bin/vglrun [vglrun options] {application_executable_or_script} {arguments} </pre> Consult Chapter <a href="#Advanced_Configuration" class="ref">20</a> for more information on <code>vglrun</code> command-line options. </li> </ol> <h2 id="hd008002">8.2 VGL Transport with a Direct X11 Connection</h2> <p><a name="Direct_X11_Connection"></a></p> <p>As with the previous mode, this mode performs optimally on local-area networks. However, it is less secure, since both the X11 traffic and the VGL Transport are unencrypted. This mode is primarily useful in grid environments, in which you may not know ahead of time which server will execute a VirtualGL job. It is assumed that the “submit host” (the machine into which you connect with SSH) and the “execute hosts” (the machines that will run VirtualGL jobs) share the same home directories and reside in the same domain.</p> <div class="important"><p class="important"> Most newer Linux and Unix distributions ship with default settings that do not allow TCP connections into the X server. Such systems cannot be used as clients with this procedure unless they are reconfigured to allow X11 TCP connections. </p></div> <h3 id="hd008002001">Procedure</h3> <p>The procedure for this mode is identical to the procedure for the <a href="#X11_Forwarding">VGL Transport with X11 forwarding</a><a name="idx0037"></a>, except that you should pass a <code>-x</code> argument to <code>vglconnect</code> when connecting to the server:</p> <pre class="verbatim"> /opt/VirtualGL/bin/vglconnect -x {user}@{server} </pre> <h2 id="hd008003">8.3 VGL Transport with X11 Forwarding and SSH Tunneling</h2> <p><a name="SSH_Tunneling"></a></p> <p>Both the VGL Transport and the X11 traffic are tunneled through SSH when using this mode, and thus it provides a completely secure solution. It is also useful when either the VirtualGL server or the client machine are behind restrictive firewalls and only SSH connections are allowed through. Using SSH tunneling on wide-area networks should not affect performance significantly. However, using SSH tunneling on a local-area network can reduce the end-to-end performance of the VGL Transport by anywhere from 20-40%.</p> <h3 id="hd008003001">Procedure</h3> <p>The procedure for this mode is identical to the procedure for the <a href="#X11_Forwarding">VGL Transport with X11 forwarding</a><a name="idx0038"></a>, except that you should pass a <code>-s</code> argument to <code>vglconnect</code> when connecting to the server:</p> <pre class="verbatim"> /opt/VirtualGL/bin/vglconnect -s {user}@{server} </pre> <p><code>vglconnect</code> will make two SSH connections into the server, the first to find an open port on the server and the second to create the SSH tunnel for the VGL Transport and open the secure shell. If you are not using an SSH agent to create password-less logins, then this mode will require you to enter your password twice.</p> <p><code>vglconnect -s</code> can be used to create multi-layered SSH tunnels. For instance, if the VirtualGL server is not directly accessible from the Internet, then you can run <code>vglconnect -s</code> on the client machine to connect to an SSH gateway server, then you can run <code>vglconnect -s</code> again on the gateway server to connect to the VirtualGL server (application server.) Both the X11 traffic and the VGL Transport will be forwarded from the VirtualGL server through the gateway and to the client.</p> <div class="figure"> <img src="sshtunnel.png" alt="sshtunnel" class="figure" id="imgid_11" name="imgid_11"/> </div> <h2 id="hd008004">8.4 VGL Transport over Gigabit Networks</h2> <p>When using the VGL Transport over Gigabit Ethernet or faster networks, it may be desirable to disable image compression. This can be accomplished by passing an argument of <code>-c rgb</code> to <code>vglrun</code> or setting the <code>VGL_COMPRESS</code> environment variable to <code>rgb</code> on the VirtualGL server. Disabling image compression will reduce VirtualGL’s server and client CPU usage by 50% or more, but the tradeoff is that it will also increase VirtualGL’s network usage by a factor of 10 or more. Thus, disabling image compression is not recommended unless you are using switched Gigabit Ethernet (or faster) infrastructure and have plenty of bandwidth to spare.</p> <h2 id="hd008005">8.5 VGL Transport with XDMCP</h2> <div class="important"><p class="important"> XDMCP is very insecure and is not recommended as a means of running VirtualGL, in general. This section is provided mainly for completeness and should not be construed as an endorsement of XDMCP. In general, using an X proxy is a much better approach for accessing a remote desktop session on the 3D application server. </p></div> <p>Using the VGL Transport with XDMCP is conceptually similar to using the VGL Transport with a <a href="#Direct_X11_Connection">direct X11 connection</a><a name="idx0039"></a>. The major difference is that, rather than remotely displaying individual X windows to the 2D X server, XDMCP remotely displays a complete desktop session from the application server. X11 applications are launched inside of this remote desktop session rather than in a separate shell, so <code>vglconnect</code> cannot be used in this case. Instead, it is necessary to start <code>vglclient</code> manually on the client machine.</p> <h3 id="hd008005001">Procedure</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Configure the server machine to accept XDMCP connections. This may require opening specific ports in its firewall. </li> <li class="Ordered-1 Ordered"> Configure the client machine to make XDMCP connections. This may require enabling X11 TCP connections and opening specific ports in its firewall. </li> <li class="Ordered-1 Ordered"> Once you have established an XDMCP connection from the client to the server, open a terminal inside the XDMCP session and type: <pre class="verbatim"> xhost +LOCAL: </pre> <div class="important"><p class="important"> This grants access to the 2D X server for any user that is currently logged into the client machine. This is not very secure, but neither is using XDMCP. If you are concerned, then see below for a discussion of how to use <code>xauth</code> to provide 2D X server access in a slightly more secure manner. </p></div> </li> <li class="Ordered-1 Ordered"> If you are using a Mac or Windows client, or if you are using a nested X server (such as Xephyr or XNest) on a Linux/Unix client to make the XDMCP connection, then the next step is easy. Simply open a new terminal/command prompt on the client machine, set the <code>DISPLAY</code> environment variable to the display name of the X server that is running the XDMCP session (usually <code>:0</code> or <code>:1</code>), and type: <pre class="verbatim"> vglclient -detach </pre> You can now close the terminal/command prompt, if you wish. </li> <li class="Ordered-1 Ordered"> If you are running a full-screen XDMCP session on a Linux/Unix client (for instance, using GDM Chooser), then starting <code>vglclient</code> is a bit trickier. In this case, you can’t open up a separate terminal window on the client machine for the purposes of running <code>vglclient</code>. However, from inside of the XDMCP session, you can open an SSH session back into the client machine. In this SSH session, set the <code>DISPLAY</code> environment variable to the display name of the X server that is running the XDMCP session (usually <code>:0</code> or <code>:1</code>), and type: <pre class="verbatim"> vglclient -detach </pre> You can now close the SSH session, if you wish. </li> </ol> <h3 id="hd008005002">Security</h3> <p>Typing <code>xhost +LOCAL:</code> in step 3 above opens the 2D X server to all current users of the client machine. This shouldn’t pose any significant risk if the client is a Windows or a Mac machine. However, Linux/Unix clients might have multiple simultaneous users, so in these cases, it may be desirable to use a more secure method of granting access to the 2D X server.</p> <p>Instead of typing <code>xhost +LOCAL:</code>, you can type the following:</p> <pre class="verbatim"> xauth nextract - $DISPLAY | sed "s/.*[ ]//g" | xargs ssh {client} xauth add {display} . </pre> <p>where <em><code>{client}</code></em> is the hostname or IP address of the client machine and <em><code>{display}</code></em> is the display name of the 2D X server, from the point of view of the client machine (usually <code>:0</code> or <code>:1</code>).</p> <p>This extracts the XAuth key for the 2D X server, then remotely adds it to the XAuth keyring on the client machine.</p> <h2 id="hd008006">8.6 The VirtualGL Client Application: Nuts and Bolts</h2> <p>The VirtualGL Client application (<code>vglclient</code>) receives encoded and/or compressed images on a dedicated TCP socket, decodes and/or decompresses the images, and draws the images into the appropriate X window. The <code>vglconnect</code> script wraps both <code>vglclient</code> and SSH to greatly simplify the process of creating VGL Transport connections.</p> <p><code>vglconnect</code> invokes <code>vglclient</code> with an argument of <code>-detach</code>, which causes <code>vglclient</code> to completely detach from the console and run as a background daemon. It will remain running silently in the background, accepting VGL Transport connections for the X server on which it was started, until that X server is reset or until the <code>vglclient</code> process is explicitly killed. Logging out of the X server will reset the X server and thus kill all <code>vglclient</code> instances that are attached to it. You can also explicitly kill all instances of <code>vglclient</code> running under your user account by invoking</p> <pre class="verbatim"> vglclient -kill </pre> <p>(<code>vglclient</code> for Linux/Unix/Mac/Cygwin installs in <code>/opt/VirtualGL/bin</code> by default, and <code>vglclient</code> for Exceed is in <code>c:\program files\virtualgl-</code><em><code>{version}</code></em><code>-</code><em><code>{build}</code></em>).</p> <p><code>vglconnect</code> instructs <code>vglclient</code> to redirect all of its console output to a log file named <em><code>{home}</code></em><code>/.vgl/vglconnect-</code><em><code>{hostname}</code></em><code>-</code><em><code>{display}</code></em><code>.log</code>, where <em><code>{home}</code></em> is the path of the current user’s home directory (<code>%USERPROFILE%</code> if using the VirtualGL Client for Exceed), <em><code>{hostname}</code></em> is the name of the computer running <code>vglconnect</code>, and <em><code>{display}</code></em> is the name of the current X display (read from the <code>DISPLAY</code> environment or passed to <code>vglconnect</code> using the <code>-display</code> argument.) In the event that something goes wrong, this log file is the first place to check.</p> <p>When <code>vglclient</code> successfully starts on a given X display, it stores its listener port number in a root window property on the X display. If other <code>vglclient</code> instances attempt to start on the same X display, they read the X window property, determine that another <code>vglclient</code> instance is already running, and exit to allow the first instance to retain control. <code>vglclient</code> will clean up the X property under most circumstances, even if it is explicitly killed. However, under rare circumstances (if sent a SIGKILL signal on Unix, for instance), a <code>vglclient</code> instance may exit uncleanly and leave the X property set. In these cases, it may be necessary to add an argument of <code>-force</code> to <code>vglconnect</code> the next time you use it. This tells <code>vglconnect</code> to start a new <code>vglclient</code> instance, regardless of whether <code>vglclient</code> thinks that there is already an instance running on this X display. Alternately, you can simply reset your X server to clear the orphaned X window property.</p> <h3 id="hd008006001">8.6.1 The VirtualGL Client and Firewalls</h3> <p>To retain compatibility with previous versions of VirtualGL, the first <code>vglclient</code> instance on a given machine will attempt to listen on port 4242 for unencrypted connections and 4243 for SSL connections (if VirtualGL was built with OpenSSL support.) If it fails to obtain one of those ports, because another application or another <code>vglclient</code> instance is already using them, then <code>vglclient</code> will try to obtain a free port in the range of 4200-4299. Failing that, it will request a free port from the operating system.</p> <p>In a nutshell: if you only ever plan to run one X server at a time on your client machine, which means that you’ll only ever need one instance of <code>vglclient</code> at a time, then it is sufficient to open inbound port 4242 (and 4243 if you plan to use SSL) in your client machine’s firewall. If you plan to run multiple X servers on your client machine, which means that you will need to run multiple <code>vglclient</code> instances, then you may wish to open ports 4200-4299. Similarly, if you are running <code>vglclient</code> on a multi-user X proxy server that has a firewall, then you may wish to open ports 4200-4299 in the server’s firewall. Opening ports 4200-4299 will accommodate up to 100 separate <code>vglclient</code> instances (50 if OpenSSL support is enabled.) More instances than that cannot be accommodated on a firewalled machine, unless the firewall is able to create rules based on application executables instead of listening ports.</p> <p>Note that it is not necessary to open any inbound ports in the firewall to use the VGL Transport with <a href="#SSH_Tunneling">SSH Tunneling</a><a name="idx0040"></a>.</p> <p><br /></p> <hr class="break" /> <h1 id="hd009"><a name="file009"></a>9 Using VirtualGL with X Proxies Such as VNC</h1> <p><a name="X11_Proxy_Usage"></a></p> <p>The VGL Transport is a good solution for using VirtualGL over a fast network. However, the VGL Transport is not generally suitable for high-latency or low-bandwidth networks, due to its reliance on the X11 protocol to send the non-3D elements of the 3D application’s GUI. The VGL Transport also requires an X server to be running on the client machine, which makes it generally more difficult to deploy and use on Windows clients. VirtualGL can be used with an “X proxy” to overcome these limitations. An X proxy acts as a virtual X server, receiving X11 commands from the application (and from VirtualGL), rendering the X11 commands into images, compressing the resulting images, and sending the compressed images over the network to a client or clients. X proxies perform well on all types of networks, including high-latency and low-bandwidth networks. They often provide rudimentary collaboration capabilities, allowing multiple clients to simultaneously view the same X session and pass around control of the keyboard and mouse. X proxies are also stateless, meaning that the client can disconnect and reconnect at will from any machine on the network, and the 3D application will remain running on the server.</p> <p>Since VirtualGL is sending rendered 3D images to the X proxy at a very fast rate, the proxy must be able to compress the images very quickly in order to keep up. Unfortunately, however, most X proxies can’t. They simply aren’t designed to compress, with any degree of performance, the large and complex images generated by 3D applications. Therefore, the VirtualGL Project provides an optimized X proxy called “TurboVNC”, a variant of <span class="remote"><a href="http://www.tightvnc.com" class="remote">TightVNC</a></span><a name="idx0041"></a> that is designed specifically to achieve high levels of performance with 3D applications running in VirtualGL. More information about TurboVNC, including instructions for using it with VirtualGL, can be found in the TurboVNC User’s Guide.</p> <p><span class="remote"><a href="http://www.tigervnc.com" class="remote">TigerVNC</a></span><a name="idx0042"></a> is a next-generation VNC project based on the RealVNC and Xorg code bases. TigerVNC spun off from the TightVNC project in early 2009, and it uses the same high-speed JPEG codec as VirtualGL and TurboVNC (<span class="remote"><a href="http://www.libjpeg-turbo.org" class="remote">libjpeg-turbo</a></span><a name="idx0043"></a>). Unlike TurboVNC, TigerVNC is not specifically designed for 3D applications. However, TigerVNC can, if properly configured, produce performance that is sufficient for use with VirtualGL. As of this writing, TigerVNC is available in Fedora 11 or later and in Red Hat Enterprise Linux 6.</p> <p>Many other X proxy solutions work well with VirtualGL, and some of these solutions provide compelling features (seamless windows, for instance, or session management), but none of these X proxies matches the performance of TurboVNC, as of this writing.</p> <h2 id="hd009001">9.1 Using VirtualGL with an X Proxy on the Same Server</h2> <p><a name="X11_Proxy_Usage_Local"></a></p> <p>The most common (and optimal) way to use VirtualGL with an X proxy is to set up both on the same server. This allows VirtualGL to send its rendered 3D images to the X proxy through shared memory rather than sending them over a network.</p> <div class="figure"> <img src="x11transport.png" alt="x11transport" class="figure" id="imgid_12" name="imgid_12"/> </div> <p>With this configuration, you can usually invoke</p> <pre class="verbatim"> /opt/VirtualGL/bin/vglrun {application_executable_or_script} </pre> <p>from a terminal inside of the X proxy session, and it will “just work.” VirtualGL reads the value of the <code>DISPLAY</code> environment variable to determine whether to enable the X11 Transport by default. If <code>DISPLAY</code> begins with a colon (“<code>:</code>”) or with “<code>unix:</code>”, then VirtualGL will assume that the X connection is local and will enable the X11 Transport as the default. In some cases, however, the <code>DISPLAY</code> environment variable in the X proxy session may not begin with a colon or “<code>unix:</code>”. In these cases, it is necessary to manually enable the X11 Transport by setting the <code>VGL_COMPRESS</code> environment variable to <code>proxy</code> or by passing an argument of <code>-c proxy</code> to <code>vglrun</code>.</p> <h2 id="hd009002">9.2 Using VirtualGL with an X Proxy on a Different Machine</h2> <p><a name="X11_Proxy_Usage_Remote"></a></p> <div class="figure"> <img src="vgltransportservernetwork.png" alt="vgltransportservernetwork" class="figure" id="imgid_13" name="imgid_13"/> </div> <p>If the X proxy and VirtualGL are running on different servers, then it is desirable to use the VGL Transport to send images from the VirtualGL server to the X proxy. It is also desirable to disable image compression in the VGL Transport. Otherwise, the images would have to be compressed by the VirtualGL server, decompressed by the VirtualGL Client, then recompressed by the X proxy, which is a waste of CPU resources. However, sending images uncompressed over a network requires a fast network (generally, Gigabit Ethernet or faster), so there needs to be a fast link between the VirtualGL server and the X proxy server for this procedure to perform well.</p> <p>The procedure for using the VGL Transport to display 3D applications from a VirtualGL server to a remote X proxy is the same as the <a href="#X11_Forwarding">procedure</a><a name="idx0044"></a> for using the VGL Transport to display 3D applications from a VirtualGL server to a remote 2D X server, with the following exceptions:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> The “client” in this case is really the X proxy server. </li> <li class="Ordered-1 Ordered"> The “X server” is really the X proxy. </li> <li class="Ordered-1 Ordered"> It is recommended that you disable image compression in the VGL Transport by either setting the <code>VGL_COMPRESS</code> environment variable to <code>rgb</code> or passing an argument of <code>-c rgb</code> to <code>vglrun</code> when launching VirtualGL. Otherwise, VirtualGL will detect that the connection to the X server is remote, and it will automatically try to enable JPEG compression. </li> </ol> <p><br /></p> <hr class="break" /> <h1 id="hd0010"><a name="file010"></a>10 Support for the X Video Extension</h1> <p><a name="X_Video_Support"></a></p> <p>The X Video extension allows applications to pre-encode or pre-compress images and send them through the X server to the graphics adapter, which presumably has on-board video decoding capabilities. This approach greatly reduces the amount of CPU resources used by the X server, which can be beneficial if the X server is running on a different machine than the application.</p> <p>In the case of VirtualGL, what this means is that the VirtualGL client machine no longer has to decode or decompress images from the 3D application server. It can simply pass the images along to the graphics adapter for decoding.</p> <p>VirtualGL supports the X Video extension in two ways:</p> <h2 id="hd0010001">10.1 The VGL Transport with YUV Encoding</h2> <p>Setting the <code>VGL_COMPRESS</code> environment variable to <code>yuv</code> or passing an argument of <code>-c yuv</code> to <code>vglrun</code> enables the VGL Transport with YUV encoding. When this mode is enabled, VirtualGL encodes images as YUV420P (a form of YUV encoding that uses 4X chrominance subsampling and separates Y, U, and V pixel components into separate image planes) instead of RGB or JPEG. The YUV420P images are sent to the VirtualGL Client, which draws them using the X Video extension.</p> <p>On a per-frame basis, YUV encoding uses about half the server CPU time as JPEG compression and only slightly more server CPU time than RGB encoding. On a per-frame basis, YUV encoding uses about 1/3 the client CPU time as JPEG compression and about half the client CPU time as RGB encoding. YUV encoding also uses about half the network bandwidth (per frame) as RGB.</p> <p>However, since YUV encoding uses 4X chrominance subsampling, the resulting images may contain some visible artifacts. In particular, narrow, aliased lines and other sharp features may appear “soft”.</p> <h2 id="hd0010002">10.2 The XV Transport</h2> <p>Setting the <code>VGL_COMPRESS</code> environment variable to <code>xv</code> or passing an argument of <code>-c xv</code> to <code>vglrun</code> enables the XV Transport. The XV Transport is a special version of the X11 Transport that encodes images as YUV420P and draws them directly to the 2D X server using the X Video extension. This is mainly useful in conjunction with X proxies that support the X Video extension. The idea is that, if the X proxy is going to have to transcode the image to YUV anyhow, VirtualGL may be faster at doing this, since it has a SIMD-accelerated YUV encoder.</p> <p><br /></p> <hr class="break" /> <h1 id="hd0011"><a name="file011"></a>11 Transport Plugins</h1> <p><a name="Transport_Plugins"></a></p> <p>VirtualGL 2.2 (and later) includes an API that allows you to write your own image transports. Thus, you can use VirtualGL for doing split rendering and pixel readback but then use your own library for delivering the pixels to the client.</p> <p>When the <code>VGL_TRANSPORT</code> environment variable (or the <code>-trans</code> option to <code>vglrun</code>) is set to <em><code>{t}</code></em>, then VirtualGL will look for a DSO (dynamic shared object) with the name <code>libtransvgl_</code><em><code>{t}</code></em><code>.so</code> in the dynamic linker path and will attempt to access a set of API functions from this library. The functions that the plugin library must export are defined in <code>/opt/VirtualGL/include/rrtransport.h</code>, and an example of their usage can be found in <code>server/testplugin.cpp</code> and <code>server/testplugin2.cpp</code> in the VirtualGL source distribution. The former wraps the VGL Transport as an image transport plugin, and the latter does the same for the X11 Transport.</p> <p><br /></p> <hr class="break" /> <h1 id="hd0012"><a name="file012"></a>12 Using VirtualGL with setuid/setgid Executables</h1> <p><code>vglrun</code> can be used to launch either binary executables or shell scripts, but there are a few things to keep in mind when using <code>vglrun</code> to launch a shell script. When you launch a shell script with <code>vglrun</code>, the VirtualGL faker library will be preloaded into every executable that the script launches. Normally this is innocuous, but if the script calls any executables that have the setuid and/or setgid permission bits set, then the dynamic linker will refuse to preload the VirtualGL faker library into those executables. One of the following warnings will be printed for each setuid/setgid executable that the script tries to launch:</p> <dl class="Description"> <dt class="Description-1 Description">Linux</dt> <dd class="Description-1 Description"> </dd> </dl> <pre class="verbatim"> ERROR: ld.so: object 'librrfaker.so' from LD_PRELOAD cannot be preloaded: ignored. </pre> <pre class="verbatim"> ERROR: ld.so: object 'libdlfaker.so' from LD_PRELOAD cannot be preloaded: ignored. </pre> <dl class="Description"> <dt class="Description-1 Description">Solaris</dt> <dd class="Description-1 Description"> </dd> </dl> <pre class="verbatim"> ld.so.1: warning: librrfaker.so: open failed: No such file in secure directories </pre> <pre class="verbatim"> ld.so.1: warning: libdlfaker.so: open failed: No such file in secure directories </pre> <p>On Solaris and versions of Linux with GLIBC 2.3 and later, these are just warnings, and the setuid/setgid executables will continue to run (without VirtualGL preloaded into them.) However, if you want to get rid of the warnings, an easy way to do so is to simply edit the application script and make it store the value of the <code>LD_PRELOAD</code> environment variable until right before the application executable is launched. For instance, consider the following application script:</p> <p>Initial contents of <code>application.sh</code>:</p> <pre class="verbatim"> #!/bin/sh some_setuid_executable some_3D_application_executable </pre> <p>You could modify the script as follows:</p> <pre class="verbatim"> #!/bin/sh LD_PRELOAD_SAVE=$LD_PRELOAD LD_PRELOAD= export LD_PRELOAD some_setuid_executable LD_PRELOAD=$LD_PRELOAD_SAVE export LD_PRELOAD some_3D_application_executable </pre> <p>This procedure may be necessary to work around certain other interaction issues between VirtualGL and the launch scripts of specific applications. See <a href="#Application_Recipes">Application Recipes</a><a name="idx0045"></a> for more details.</p> <p>If the 3D application that you are intending to run in VirtualGL is itself a setuid/setgid executable, then further steps are required. Otherwise, the 3D application will launch without VirtualGL preloaded into it. Forcing VirtualGL to be preloaded into setuid/setgid executables has security ramifications, so please be aware of these before you do it. By applying one of the following workarounds, you are essentially telling the operating system that you trust the security and stability of VirtualGL as much as you trust the security and stability of the operating system. While we’re flattered, we’re not sure that we’re necessarily deserving of that accolade, so if you are in a security-critical environment, apply the appropriate level of paranoia here.</p> <p><a name="setuid_linux"></a> To force VirtualGL to be preloaded into setuid/setgid executables on Linux, you have to first make sure that the faker libraries (<code>librrfaker.so</code> and <code>libdlfaker.so</code>) are installed in the “system” library path (usually <code>/usr/lib</code>, <code>/usr/lib64</code>, <code>/usr/lib32</code>, or <code>/usr/lib/i386-linux-gnu</code>). Next, make <code>librrfaker.so</code> and <code>libdlfaker.so</code> setuid executables. To do this, run the following commands as root:</p> <pre class="verbatim"> chmod u+s /usr/{lib}/librrfaker.so chmod u+s /usr/{lib}/libdlfaker.so </pre> <p>where <em><code>{lib}</code></em> is <code>lib</code>, <code>lib64</code>, <code>lib32</code>, or <code>lib/i386-linux-gnu</code>, depending on your system.</p> <p>On Solaris, you can force VirtualGL to be preloaded into setuid/setgid executables by adding the VirtualGL library directories to the Solaris “secure path.” Solaris keeps a tight lid on what goes into <code>/usr/lib</code> and <code>/lib</code>, and by default, it will only allow libraries in those paths to be preloaded into an executable that is setuid and/or setgid. Generally, 3rd party packages are forbidden from installing anything into <code>/usr/lib</code> or <code>/lib</code>, but you can use the <code>crle</code> utility to add other directories to the operating system’s list of secure paths. In the case of VirtualGL, you would execute one of the following commands (as root):</p> <dl class="Description"> <dt class="Description-1 Description">32-bit VirtualGL:</dt> <dd class="Description-1 Description"> <pre class="verbatim"> crle -u -s /opt/VirtualGL/lib32 </pre> </dd> <dt class="Description-1 Description">64-bit VirtualGL:</dt> <dd class="Description-1 Description"> <pre class="verbatim"> crle -64 -u -s /opt/VirtualGL/lib64 </pre> </dd> </dl> <p><br /></p> <hr class="break" /> <h1 id="hd0013"><a name="file013"></a>13 Using VirtualGL with Chromium</h1> <p><a name="Chromium"></a></p> <p>Chromium is a powerful framework for performing various types of parallel OpenGL rendering. It is usually used on clusters of commodity Linux PCs to divide up the task of rendering scenes with large geometries or large pixel counts (such as when driving a display wall.) Chromium is most often used in one of three configurations:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered" value="1"> Sort-First Rendering (Image-Space Decomposition) </li> <li class="Ordered-1 Ordered" value="2"> Sort-First Rendering (Image-Space Decomposition) with Readback </li> <li class="Ordered-1 Ordered" value="3"> Sort-Last Rendering (Object-Space Decomposition) </li> </ol> <h2 id="hd0013001">13.1 Configuration 1: Sort-First Rendering (Image-Space Decomposition)</h2> <div class="figure"> <img src="chromium-displaywall.png" alt="chromium-displaywall" class="figure" id="imgid_14" name="imgid_14"/> </div> <p>Sort-First Rendering (Image-Space Decomposition) is used to overcome the fill rate limitations of individual GPUs. When configured to use sort-first rendering, Chromium divides up the scene based on which polygons will be visible in a particular section of the final image. It then instructs each node of the cluster to render only the polygons that are necessary to generate the image section (“tile”) for that node. This is primarily used to drive high-resolution displays that would be impractical to drive from a single GPU due to limitations in the GPU’s framebuffer memory, processing power, or both. Configuration 1 could be used, for instance, to drive a CAVE, video wall, or even an extremely high-resolution monitor. In this configuration, each Chromium node generally uses all of its screen real estate to render a section of the multi-screen image.</p> <p>VirtualGL is generally not very useful with Configuration 1. You could theoretically install a separate copy of VirtualGL on each display node and use it to redirect the output of each <code>crserver</code> instance to a separate VirtualGL Client instance running on a multi-screen 2D X server elsewhere on the network. However, synchronizing the frames on the remote end would require extensive modifications to VirtualGL and perhaps to Chromium as well. Such is left as an exercise for the reader.</p> <h2 id="hd0013002">13.2 Configuration 2: Sort-First Rendering (Image-Space Decomposition) with Readback</h2> <div class="figure"> <img src="chromium-sortfirst.png" alt="chromium-sortfirst" class="figure" id="imgid_15" name="imgid_15"/> </div> <p>Configuration 2 uses the same sort-first principle as Configuration 1, except that each tile is only a fraction of a single screen, and the tiles are recombined into a single window on Node 0. This configuration is perhaps the least often used of the three, but it is useful in cases where the scene contains a large amount of textures (such as in volume rendering) and thus rendering the whole scene on a single node would be prohibitively slow due to fill rate limitations.</p> <p>In this configuration, the application is allowed to choose a visual, create an X window, and manage the window as it would normally do. However, all other OpenGL and GLX activity is intercepted by the Chromium App Faker (CrAppFaker) so that the 3D rendering can be split up among the rendering nodes. Once each node has rendered its section of the final image, the image tiles are passed back to a Chromium Server (CrServer) process running on Node 0. This CrServer process attaches to the previously-created application window and draws the pixels into the window using <code>glDrawPixels()</code>.</p> <p>The general strategy for making this work with VirtualGL is to first make it work without VirtualGL and then insert VirtualGL only into the processes that run on Node 0. VirtualGL must be inserted into the CrAppFaker process to prevent CrAppFaker from sending <code>glXChooseVisual()</code> calls to the 2D X server (which would fail if this X server did not support GLX.) VirtualGL must be inserted into the CrServer process on Node 0 to prevent it from sending <code>glDrawPixels()</code> calls to the 2D X server (which would similarly fail if the 2D X server did not support GLX, and which would perform very poorly if the 2D X server was remote.) Instead, VirtualGL forces CrServer to draw into a Pbuffer, and VGL then takes charge of transmitting the pixels from the Pbuffer to the 2D X server in the most efficient way possible.</p> <p>As with any normal OpenGL application, CrServer can be launched using <code>vglrun</code>. However, because CrAppFaker also interposes OpenGL and GLX functions, it must be handled differently in order to avoid interference with VirtualGL. Chromium provides an environment variable, <code>CR_SYSTEM_GL_PATH</code>, which allows one to specify an alternate path to be searched for <code>libGL.so</code>. On Linux and Unix systems, VirtualGL installs a symbolic link named <code>libGL.so</code>, which points to the VirtualGL faker library (<code>librrfaker.so</code>). This symbolic link is located in its own isolated directory, so that directory can be passed to Chromium in the <code>CR_SYSTEM_GL_PATH</code> environment variable, and this will cause Chromium to load VirtualGL rather than the “real” OpenGL library. Refer to the following table:</p> <p><a name="CR_SYSTEM_GL_PATH_Table"></a></p> <div class="table"> <table class="standard" align="center"> <thead class="standard"> <tr class="head "> <th class="head standard">32-bit Applications</th> <th class="head standard">64-bit Applications</th> </tr> </thead> <tr class="standard"> <td class="standard"><code>/opt/VirtualGL/fakelib32</code></td> <td class="standard"><code>/opt/VirtualGL/fakelib64</code></td> </tr> </table> <div class="tableNote" style="text-align=center;"><code>CR_SYSTEM_GL_PATH</code> setting required to use VirtualGL with Chromium</div> </div> <p>To run CrAppFaker, it is necessary to set this environment variable to the appropriate value so that Chromium will load the interposed versions of OpenGL and GLX functions from VirtualGL. It is also necessary to set <code>VGL_GLLIB</code> to the location of the “real” OpenGL library (example: <code>/usr/lib64/libGL.so.1</code>). CrAppFaker creates its own fake version of <code>libGL.so</code>, which is really just a copy of Chromium’s <code>libcrfaker.so</code>. Thus, if left to its own devices, VirtualGL will unwittingly try to load <code>libcrfaker.so</code> instead of the “real” OpenGL library. Chromium’s <code>libcrfaker.so</code> will, in turn, try to load VirtualGL, and an endless loop will occur.</p> <p>Therefore, we must use the <code>CR_SYSTEM_GL_PATH</code> environment variable to tell Chromium to pass OpenGL commands into VirtualGL, then we must use the <code>VGL_GLLIB</code> environment variable to tell VirtualGL <em>not</em> to pass OpenGL commands into Chromium. For example:</p> <pre class="verbatim"> export CR_SYSTEM_GL_PATH=/opt/VirtualGL/fakelib64 export VGL_GLLIB=/usr/lib64/libGL.so.1 crappfaker </pre> <p>CrAppFaker will copy the application into a temporary directory and then copy <code>libcrfaker.so</code> to that same directory, renaming it as <code>libGL.so</code>. This causes the application to load <code>libcrfaker.so</code> instead of <code>libGL.so</code>. <code>libcrfaker.so</code> will then load VirtualGL instead of the “real” OpenGL library, because we’ve set <code>CR_SYSTEM_GL_PATH</code> to point to the directory containing VirtualGL’s fake <code>libGL.so</code>. VirtualGL will then use the library specified in <code>VGL_GLLIB</code> to make any “real” OpenGL calls that it needs to make.</p> <div class="important"><p class="important"> NOTE: <code>crappfaker</code> should not be invoked with <code>vglrun</code>. </p></div> <p>So, putting this all together, here is an example of how you might start a sort-first rendering job using Chromium and VirtualGL:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Start the mothership on Node 0 with an appropriate configuration for performing sort-first rendering with readback </li> <li class="Ordered-1 Ordered"> Start <code>crserver</code> on each of the rendering nodes <div class="important"><p class="important"> NOTE: <code>crserver</code> should be run on display :0 (or whichever display is attached to the 3D hardware.) </p></div> </li> <li class="Ordered-1 Ordered"> On Node 0, execute <code>vglrun crserver &</code> </li> <li class="Ordered-1 Ordered"> On Node 0, set the <code>CR_SYSTEM_GL_PATH</code> environment variable to the appropriate value based on whether <code>crappfaker</code> was compiled as a 32-bit or a 64-bit app (see table above) </li> <li class="Ordered-1 Ordered"> On Node 0, set <code>VGL_GLLIB</code> to the location of the “real” OpenGL library (example: <code>/usr/lib64/libGL.so.1</code>). </li> <li class="Ordered-1 Ordered"> On Node 0, launch <code>crappfaker</code> (do not use <code>vglrun</code> here) </li> </ol> <p>Again, it’s always a good idea to make sure this works without VirtualGL before adding VirtualGL into the mix.</p> <p><a name="Force_Pbuffer"></a></p> <h3 id="hd0013002001">Using VirtualGL to Force Pbuffer Rendering</h3> <p>In the procedure above, VirtualGL can also be used on the rendering nodes to redirect the rendering commands from <code>crserver</code> into a Pbuffer instead of a window. If you want to do this, then perform the following procedure in place of step 2 above:</p> <p>On each of the rendering nodes,</p> <ul class="Itemize"> <li class="Itemize-1 Itemize asterisk"> set the <code>VGL_READBACK</code> environment variable to <code>0</code> </li> <li class="Itemize-1 Itemize asterisk"> <code>vglrun crserver</code> <div class="important"><p class="important"> NOTE: You must configure each of the rendering nodes as a VirtualGL server (that is, run <code>'vglserver_config</code>’ as instructed in Chapter <a href="#Unix_Config" class="ref">6</a>) in order for this to work when the rendering nodes are sitting at the login prompt. </p></div> </li> </ul> <h2 id="hd0013003">13.3 Configuration 3: Sort-Last Rendering (Object-Space Decomposition)</h2> <div class="figure"> <img src="chromium-sortlast.png" alt="chromium-sortlast" class="figure" id="imgid_16" name="imgid_16"/> </div> <p>Sort-Last Rendering is used when the scene contains a huge number of polygons and the rendering bottleneck is processing all of that geometry on a single GPU. In this case, each node runs a separate copy of the application, and for best results, the application needs to be aware that it is running in a parallel environment so that it can give Chromium hints as to how to distribute the various objects to be rendered. Each node generates an image of a particular portion of the object space, and these images must be composited in such a way that the front-to-back ordering of pixels is maintained. This is generally done by collecting Z buffer data from each node to determine whether a particular pixel on a particular node is visible in the final image. The rendered images from each node are often composited using a “binary swap”, whereby the nodes combine their images in a cascading tree so that the overall compositing time is proportional to log<sub>2</sub>(N) rather than N.</p> <p>To make this configuration work with VirtualGL:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Start the mothership on Node 0 with an appropriate configuration for performing sort-last rendering </li> <li class="Ordered-1 Ordered"> Start <code>crappfaker</code> on each of the rendering nodes <div class="important"><p class="important"> NOTE: <code>crappfaker</code> should be run on display :0 (or whichever display is attached to the 3D hardware.) </p></div> </li> <li class="Ordered-1 Ordered"> On Node 0, execute <code>vglrun crserver</code> </li> </ol> <h3 id="hd0013003001">CRUT</h3> <p>The Chromium Utility Toolkit provides a convenient way for graphics applications to specifically take advantage of Chromium’s sort-last rendering capabilities. Such applications can use CRUT to explicitly specify how their object space should be decomposed. CRUT applications require an additional piece of software, <code>crutserver</code>, to be run on Node 0. Therefore, the following procedure should be used to make these applications work with VirtualGL:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Start the mothership on Node 0 with an appropriate configuration for performing sort-last rendering </li> <li class="Ordered-1 Ordered"> Start <code>crappfaker</code> on each of the rendering nodes <div class="important"><p class="important"> NOTE: <code>crappfaker</code> should be run on display :0 (or whichever display is attached to the 3D hardware.) </p></div> </li> <li class="Ordered-1 Ordered"> On Node 0, execute <code>vglrun crutserver &</code> </li> <li class="Ordered-1 Ordered"> On Node 0, execute <code>vglrun crserver</code> </li> </ol> <h2 id="hd0013004">13.4 A Note About Performance</h2> <p>Chromium’s use of X11 is generally not very optimal. It assumes a very fast connection between the 2D X server and the Chromium Server. In certain modes, Chromium polls the 2D X server on every frame to determine whether windows have been resized, etc. Thus, we have observed that, even on a fast network, Chromium tends to perform much better with VirtualGL running in an X proxy as opposed to using the VGL Transport.</p> <p><br /></p> <hr class="break" /> <h1 id="hd0014"><a name="file014"></a>14 Using VirtualGL with VirtualBox</h1> <p>VirtualBox is an enterprise-class, open source virtualization solution provided by Oracle. Since version 2.1.0, VirtualBox has provided support for hardware-accelerated OpenGL in Windows and Linux guests running on Windows, Mac/Intel, Linux, and Solaris/x86 hosts. 3D acceleration in VirtualBox is accomplished by installing a special driver in the guest that uses a subset of Chromium to transmit OpenGL calls through a local connection to the VirtualBox process running on the host. When used in conjunction with VirtualGL on a Linux or Solaris/x86 host, this solution provides a means of displaying Windows 3D applications remotely.</p> <p>To use VirtualGL with VirtualBox, perform the following procedures:</p> <h3 id="hd0014000001">Configuring the System</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Launch VirtualBox and use the VirtualBox GUI to create and test your virtual machine. </li> <li class="Ordered-1 Ordered"> Follow the procedures outlined in the VirtualBox User’s Manual to enable 3D acceleration in the virtual machine. Verify that 3D acceleration works <em>without</em> VirtualGL before adding VirtualGL to the mix. </li> <li class="Ordered-1 Ordered"> Follow the procedure described in Chapter <a href="#setuid_linux" class="ref">12</a> to make <code>librrfaker.so</code> and <code>libdlfaker.so</code> setuid executables (Linux) or to add the VirtualGL library directory to the list of secure paths (Solaris). </li> </ol> <h3 id="hd0014000002">Launching VirtualBox</h3> <pre class="verbatim"> vglrun VirtualBox -startvm {VM name or UUID} </pre> <p>This should work on most systems. It is known to work with VirtualBox 4.1.8 and prior and with VirtualBox 4.2 and later on Linux.</p> <div class="important"><p class="important"> With VirtualBox 4.1.10 and later 4.1.x versions, it is necessary to rename <code>/usr/lib/virtualbox/VBoxTestOGL</code> and execute <code>ln -fs /bin/true /usr/lib/virtualbox/VBoxTestOGL</code> in order to use those versions of VirtualBox with VirtualGL. </p></div> <h3 id="hd0014000003">NOTES</h3> <ul class="Itemize"> <li class="Itemize-1 Itemize asterisk"> You might want to temporarily enable profiling (add an argument of <code>+pr</code> to <code>vglrun</code> above) or set the <code>VGL_LOGO</code> environment variable to <code>1</code> in order to verify that VirtualGL is loaded and working. </li> <li class="Itemize-1 Itemize asterisk"> It is necessary to start the virtual machine directly as described above. Simply executing <code>vglrun VirtualBox</code> and using the GUI to launch the VM does not work. VirtualBox forks a separate process for each VM, and the value of the <code>LD_PRELOAD</code> environment variable from <code>vglrun</code> does not get propagated to the VM process unless you start it directly. </li> <li class="Itemize-1 Itemize asterisk"> VirtualBox sends mainly uncompressed image updates to the X display, so the 2D elements of the virtual machine’s display will not perform well over a remote X11 connection unless gigabit Ethernet (or faster) and a <a href="#Direct_X11_Connection">direct X11 connection</a><a name="idx0046"></a> (<code>vglconnect -x</code>) are used. Using a high-performance X proxy (such as TurboVNC) is generally preferred when remotely displaying VirtualBox using VirtualGL, particularly on 100 Megabit and slower networks. </li> </ul> <p><br /></p> <hr class="break" /> <h1 id="hd0015"><a name="file015"></a>15 Using VirtualGL with VMWare Workstation</h1> <p>VirtualGL can also be used with VMWare Workstation, and the concept is basically the same as that of VirtualBox. As with VirtualBox, VMWare uses a special driver in the guest O/S to intercept the OpenGL commands and marshal them to the host O/S, where VirtualGL picks them up.</p> <p>To use VirtualGL with VMWare Workstation, perform the following procedures:</p> <h3 id="hd0015000001">Configuring the System</h3> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Launch VMWare and use the VMWare GUI to create and test your virtual machine. </li> <li class="Ordered-1 Ordered"> Follow the procedures outlined in the VMWare User’s Manual to enable 3D acceleration in the virtual machine. Verify that 3D acceleration works <em>without</em> VirtualGL before adding VirtualGL to the mix. </li> <li class="Ordered-1 Ordered"> Follow the procedure described in Chapter <a href="#setuid_linux" class="ref">12</a> to make <code>librrfaker.so</code> and <code>libdlfaker.so</code> setuid executables. </li> </ol> <h3 id="hd0015000002">Launching VMWare</h3> <p>This has been tested with VMWare Workstation 9.</p> <pre class="verbatim"> vglrun vmware -X {VM path}/{VM name}.vmx </pre> <h3 id="hd0015000003">NOTES</h3> <p>The notes from the previous chapter apply to VMWare Workstation as well.</p> <p><br /></p> <hr class="break" /> <h1 id="hd0016"><a name="file016"></a>16 Other Application Recipes</h1> <p><a name="Application_Recipes"></a></p> <div class="table"> <table class="standard"> <thead class="standard"> <tr class="head "> <th class="head standard">Application</th> <th class="head standard">Platform</th> <th class="head standard">Recipe</th> <th class="head standard">Notes</th> </tr> </thead> <tr class="standard"> <td class="standard">Abaqus v6</td> <td class="standard">Linux</td> <td class="standard">It is necessary to add <br /><br /> <code>import os</code><br /> <code>os.environ['ABAQUS_EMULATE_OVERLAYS'] = "1"</code> <br /><br /> to <code>/{abaqus_install_dir}/{abaqus_version}/site/abaqus_v6.env</code> to make Abaqus v6 work properly with VirtualGL if the 2D X server does not support transparent overlays. If this is not done, then the application may fail to launch, fail to display the 3D pixels, or the 3D pixels may become corrupted whenever other windows obscure them.</td> <td class="standard">VirtualGL does not redirect the rendering of transparent overlays, since those cannot be rendered in a Pbuffer. Thus, in order to use transparent overlays, the 2D X server must be able to render them, which is rarely the case for X proxies (see Section <a href="#overlays" class="ref">17.2</a> for more details.) Setting <code>ABAQUS_EMULATE_OVERLAYS</code> to <code>1</code> causes the application to emulate overlay rendering instead of using actual transparent overlays. This workaround is known to be necessary when running Abaqus 6.9 and 6.10 in VNC.</td> </tr> <tr class="standard"> <td class="standard">Abaqus v6</td> <td class="standard">Linux</td> <td class="standard"><code>vglrun -nodl {abaqus_path}/abaqus</code></td> <td class="standard">User reports indicate that Abaqus 6.9 will not work properly if <code>libdlfaker.so</code> from VirtualGL is preloaded into it. This may be true for other versions of Abaqus as well.</td> </tr> <tr class="standard"> <td class="standard">Abaqus v6</td> <td class="standard">Linux</td> <td class="standard">Set the <code>VGL_DEFAULTFBCONFIG</code> environment variable to <code>GLX_STENCIL_SIZE,8</code> prior to launching the application with <code>vglrun</code></td> <td class="standard">Abaqus 6.10 requires a visual with a stencil buffer, but it does not call <code>glXChooseVisual()</code> to specify its desire for such a visual. Thus, VirtualGL has no idea of the application’s preference, and it is necessary to use <code>VGL_DEFAULTFBCONFIG</code> to give VirtualGL a hint as to what the application wants. Otherwise, VirtualGL will create a Pbuffer without a stencil buffer, and this will cause incorrect rendering when using certain functions within Abaqus (such as View Cut.) See Section <a href="#VGL_DEFAULTFBCONFIG" class="ref">20.1</a> for further information.</td> </tr> <tr class="standard"> <td class="standard">Animator 4</td> <td class="standard">Linux</td> <td class="standard">Comment out the line that reads <br /><br /> <code>unsetenv LD_PRELOAD</code> <br /><br /> in the <code>a4</code> script, then launch Animator 4 using <br /><br /> <code>vglrun -ge a4</code> <br /><br /></td> <td class="standard">When the <code>a4</code> script unsets <code>LD_PRELOAD</code>, this prevents VirtualGL from being loaded into the application. Animator 4 additionally checks the value of <code>LD_PRELOAD</code> and attempts to unset it from inside the application. Using <code>vglrun -ge</code> to launch the application fools Animator 4 into thinking that <code>LD_PRELOAD</code> is unset.</td> </tr> <tr class="standard"> <td class="standard">ANSA v12.1.0</td> <td class="standard">Linux</td> <td class="standard">Add <br /><br /> <code>LD_PRELOAD_SAVE=$LD_PRELOAD</code><br /> <code>export LD_PRELOAD=</code> <br /><br /> to the top of the <code>ansa.sh</code> script, then add <br /><br /> <code>export LD_PRELOAD=$LD_PRELOAD_SAVE</code> <br /><br /> just prior to the <code>${ANSA_EXEC_DIR}bin/ansa_linux${ext2}</code> line.</td> <td class="standard">The ANSA startup script directly invokes <code>/lib/libc.so.6</code> to query the glibc version. Since the VirtualGL faker depends on libc, preloading VirtualGL when directly invoking <code>libc.so.6</code> creates an infinite loop. Thus, it is necessary to disable the preloading of VirtualGL in the application script and then re-enable it prior to launching the actual application.</td> </tr> <tr class="standard"> <td class="standard">ANSYS HFSS, ANSYS ICEM CFD, Roxar RMS</td> <td class="standard">Linux</td> <td class="standard">Set the <code>VGL_SPOILLAST</code> environment variable to <code>0</code> prior to launching the application with <code>vglrun</code></td> <td class="standard">These applications draw node highlighting and/or rubber banding directly to the front buffer. In order for these front buffer operations to be displayed properly, it is necessary to use the “spoil first” frame spoiling algorithm whenever the application calls <code>glFlush()</code>. See Section <a href="#VGL_SPOILLAST" class="ref">20.1</a> for further information.</td> </tr> <tr class="standard"> <td class="standard">AutoForm v4.0x</td> <td class="standard">All</td> <td class="standard"><code>vglrun +sync xaf_</code><em><code>{version}</code></em></td> <td class="standard">AutoForm relies on mixed X11/OpenGL rendering, and thus certain features (particularly the “Dynamic Section” dialog and “Export Image” feature) do not work properly unless <code>VGL_SYNC</code> is enabled. Since <code>VGL_SYNC</code> automatically enables the X11 transport and disables frame spoiling, it is highly recommended that you use an X proxy when <code>VGL_SYNC</code> is enabled. See Section <a href="#VGL_SYNC" class="ref">20.1</a> for further information.</td> </tr> <tr class="standard"> <td class="standard">Cedega v6.0.x</td> <td class="standard">Linux</td> <td class="standard">Add <br /><br /> <code>export LD_PRELOAD=librrfaker.so</code> <br /><br /> to the top of <code>~/.cedega/.winex_ver/winex-{version}/bin/winex3</code>, then run Cedega as you would normally (without <code>vglrun</code>.) Since <code>vglrun</code> is not being used, it is necessary to use environment variables or the VirtualGL Configuration dialog to modify VirtualGL’s configuration.</td> <td class="standard">The actual binary (WineX) that uses OpenGL is buried beneath several layers of Python and shell scripts. The <code>LD_PRELOAD</code> variable does not get propagated down from the initial shell that invoked <code>vglrun</code>.</td> </tr> <tr class="standard"> <td class="standard">Compiz</td> <td class="standard">Linux</td> <td class="standard">Set the <code>VGL_WM</code> environment variable to <code>1</code> prior to launching the window manager with <code>vglrun</code>, or pass an argument of <code>+wm</code> to <code>vglrun</code>.</td> <td class="standard">See Section <a href="#VGL_WM" class="ref">20.1</a> for further information.</td> </tr> <tr class="standard"> <td class="standard">Heretic II</td> <td class="standard">Linux</td> <td class="standard"><code>vglrun heretic2 +set vid_ref glx</code></td> <td class="standard"></td> </tr> <tr class="standard"> <td class="standard">MAGMA5</td> <td class="standard">Linux</td> <td class="standard">Set the <code>VGL_DEFAULTFBCONFIG</code> environment variable to <code>GLX_STENCIL_SIZE,8</code> prior to launching the application with <code>vglrun</code></td> <td class="standard">MAGMA5 requires a visual with a stencil buffer, but it does not call <code>glXChooseVisual()</code> to specify its desire for such a visual. Thus, VirtualGL has no idea of the application’s preference, and it is necessary to use <code>VGL_DEFAULTFBCONFIG</code> to give VirtualGL a hint as to what the application wants. Otherwise, VirtualGL will create a Pbuffer without a stencil buffer, and this will cause incorrect rendering when using certain functions within MAGMA5. See Section <a href="#VGL_DEFAULTFBCONFIG" class="ref">20.1</a> for further information.</td> </tr> <tr class="standard"> <td class="standard">Mathematica 7</td> <td class="standard">Linux</td> <td class="standard">Set the <code>VGL_ALLOWINDIRECT</code> environment variable to <code>1</code> prior to launching the application with <code>vglrun</code></td> <td class="standard">Mathematica 7 will not draw the axis numbers on 3D charts correctly unless it is allowed to create an indirect OpenGL context. See Section <a href="#VGL_ALLOWINDIRECT" class="ref">20.1</a> for further information.</td> </tr> <tr class="standard"> <td class="standard">Tecplot 360 2011 and earlier</td> <td class="standard">Linux</td> <td class="standard">Set the <code>VGL_GLFLUSHTRIGGER</code> environment variable to <code>0</code> prior to launching the application with <code>vglrun</code></td> <td class="standard">When running in TurboVNC (using VirtualGL), flashing artifacts will be produced when the user zooms/pans/rotates the scene in Tecplot 360, unless VirtualGL is instructed not to use <code>glFlush()</code> as an end-of-frame trigger. This has been fixed in Tecplot 2012 and later. See Section <a href="#VGL_GLFLUSHTRIGGER" class="ref">20.1</a> for further information.</td> </tr> </table> </div> <p><br /></p> <hr class="break" /> <h1 id="hd0017"><a name="file017"></a>17 Advanced OpenGL Features</h1> <p><a name="Advanced_OpenGL"></a></p> <h2 id="hd0017001">17.1 Stereographic Rendering</h2> <p>Stereographic rendering is a feature of OpenGL that creates separate rendering buffers for the left and right eyes and allows the application to render a different image into each buffer. How the stereo images are subsequently displayed depends on the particulars of the 3D hardware and the user’s environment. VirtualGL can support stereographic applications in one of two ways: (1) by sending the stereo image pairs to the client to be displayed in stereo by the client’s GPU, or (2) by combining each stereo image pair into a single image that can be viewed with traditional anaglyphic 3D glasses or a passive stereo system, such as a 3D TV.</p> <h3 id="hd0017001001">17.1.1 Quad-Buffered Stereo</h3> <p>The name “quad-buffered” stereo refers to the fact that OpenGL uses four buffers (left front, right front, left back, and right back) to support stereographic rendering with double buffering. GPUs with quad-buffered stereo capabilities generally provide some sort of synchronization signal that can be used to control various types of active stereo 3D glasses. Some also support “passive stereo”, which requires displaying the left and right eye buffers to different monitor outputs. VirtualGL supports quad-buffered stereo by rendering the stereo images on the server and sending the image pairs across the network to be displayed on the client.</p> <p>In most cases, VirtualGL does not require that a GPU be present in the client machine. However, a GPU is required to display stereo image pairs, so one must be present in any client machine that will use VirtualGL’s quad-buffered stereo feature. Since the GPU is only being used to draw images, it need not necessarily be a high-end GPU. Generally, the least expensive GPU that has stereo capabilities will work fine in a VirtualGL client machine. The VirtualGL server must also have a GPU that supports stereo, since this is the only way that VirtualGL can obtain a stereo Pbuffer.</p> <p>When an application tries to render something in stereo, VirtualGL will default to using quad-buffered stereo rendering if the 2D X server supports OpenGL and has stereo visuals available (Exceed 3D is required for Windows clients.) Otherwise, VirtualGL will fall back to using anaglyphic stereo (see below.) It is usually necessary to explicitly enable stereo in the graphics driver configuration for both the client and server machines. The <a href="#Troubleshooting">Troubleshooting</a><a name="idx0047"></a> section below lists a way to verify that both the 3D X server and the 2D X server have stereo visuals available.</p> <p>In quad-buffered mode, VirtualGL reads back both the left and right eye buffers on the server and sends the contents as a pair of compressed images to the VirtualGL Client. The VirtualGL Client then decompresses both images and draws them as a single stereo frame to the client machine’s X display using <code>glDrawPixels()</code>. It should thus be no surprise that enabling quad-buffered stereo in VirtualGL decreases performance by 50% or more and uses twice the network bandwidth to maintain the same frame rate as mono.</p> <p>Quad-buffered stereo requires the VGL Transport. Attempting to enable it with any other image transport will cause VGL to fall back to anaglyphic stereo mode.</p> <h3 id="hd0017001002">17.1.2 Anaglyphic Stereo</h3> <p>Anaglyphic stereo is the type of stereographic display used by old 3D movies. It typically relies on a set of 3D glasses consisting of red transparency film over the left eye and cyan transparency film over the right eye, although green/magenta and blue/yellow schemes can be used as well. To generate a 3D anaglyph, one color channel from the left eye buffer is combined with the other two color channels from the right eye buffer, thus allowing a single monographic image to contain stereo data. For instance, in the case of red/cyan, the red channel is taken from the left eye buffer, and the green and blue channels are taken from the right eye buffer. From the point of view of VirtualGL, an anaglyphic image is the same as a monographic image, so anaglyphic stereo images can be sent using any image transport to any type of client, regardless of the client’s capabilities.</p> <p>VirtualGL uses anaglyphic stereo if it detects that an application has rendered something in stereo but quad-buffered stereo is not available, either because the client doesn’t support it or because a transport other than the VGL Transport is being used. Anaglyphic stereo provides a cheap and easy way to view stereographic applications in X proxies and on clients that do not support quad-buffered stereo. Additionally, anaglyphic stereo performs much faster than quad-buffered stereo, since it does not require sending twice the data to the client.</p> <p>As with quad-buffered stereo, anaglyphic stereo requires that the VirtualGL server have stereo rendering capabilities. However, anaglyphic stereo does not require any 3D rendering capabilities (stereo or otherwise) on the client machine.</p> <h3 id="hd0017001003">17.1.3 Passive Stereo</h3> <p>As with anaglyphic stereo, passive stereo combines a stereographic image pair into a single image (a “stereogram”), and thus it can be used with any image transport. However, unlike anaglyphic stereo, passive stereo must be used with specific display hardware, such as a 3D TV or monitor, that decodes the left and right eye images from the stereogram and sends them separately to a pair of 3D glasses (typically, this is accomplished by way of polarization.)</p> <p>VirtualGL supports three methods of encoding stereograms:</p> <dl class="Description"> <dt class="Description-1 Description">Interleaved</dt> <dd class="Description-1 Description"> The even rows of the stereogram are taken from the left eye image, and the odd rows are taken from the right eye image. </dd> <dt class="Description-1 Description">Top/Bottom</dt> <dd class="Description-1 Description"> The top half of the stereogram is taken from the left eye image, and the bottom half is taken from the right eye image. Both halves are subsampled 2X vertically. </dd> <dt class="Description-1 Description">Side-by-Side</dt> <dd class="Description-1 Description"> The left half of the stereogram is taken from the left eye image, and the right half is taken from the right eye image. Both halves are subsampled 2X horizontally. </dd> </dl> <p>Most 3D TVs/monitors can be configured to decode at least one of these types of stereograms. In order for this to work, however, the 3D drawing area must be full-screen.</p> <h3 id="hd0017001004">17.1.4 Selecting a Stereo Mode</h3> <p>A particular stereo mode can be selected by setting the <code>VGL_STEREO</code> environment variable or by using the <code>-st</code> argument to <code>vglrun</code>. See Section <a href="#VGL_STEREO" class="ref">20.1</a> for more details.</p> <h2 id="hd0017002">17.2 Transparent Overlays</h2> <p><a name="overlays"></a></p> <p>In the case of transparent overlays, VirtualGL completely bypasses its own GLX faker and uses indirect OpenGL rendering to draw to the transparent overlay using the 2D X server. The underlay is still rendered on the 3D X server, read back, and sent to the 2D X server, as always. Using indirect rendering to render the overlay is unfortunately necessary, because there is no reliable way to draw to an overlay using 2D (X11) functions, there are severe performance issues (on some cards) with using <code>glDrawPixels()</code> to draw to the overlay, and there is no reasonable way to composite the overlay and underlay in a Pbuffer on the VirtualGL server.</p> <p>The use of overlays is becoming more and more infrequent, and when they are used, it is generally only for drawing small, simple, static shapes and text. We have found that it is often faster to ship the overlay geometry over to the 2D X server rather than to render it as an image and send the image. Thus, even if it were possible to implement overlays without using indirect rendering, it is likely that indirect rendering of overlays would still be the fastest approach for most applications.</p> <p>As with quad-buffered stereo, overlays must be explicitly enabled in the graphics driver and X server configurations. In the case of overlays, however, they need only be supported and enabled on the client machine and in the 2D X server. Some graphics drivers are known to disallow using both quad-buffered stereo and overlays at the same time.</p> <p>Indexed color (8-bit) overlays have been tested and are known to work with VirtualGL. True color (24-bit) overlays work, in theory, but have not been tested. Use <code>glxinfo</code> (see <a href="#Troubleshooting">Troubleshooting</a><a name="idx0048"></a> below) to verify whether your client’s X display supports overlays and whether they are enabled. In Exceed 3D, make sure that the “Overlay Support” option is checked in the “Exceed 3D and GLX” applet:</p> <p><img src="exceed6.png" alt="exceed6" class="inline" id="imgid_4" name="imgid_4"/></p> <h2 id="hd0017003">17.3 Color Index (PseudoColor) Rendering</h2> <p>In a PseudoColor visual, each pixel is represented by an index that refers to a location in a color table. The color table stores the actual color values (256 of them in the case of 8-bit PseudoColor) that correspond to each index. An application merely tells the X server which color index to use when drawing, and the X server takes care of mapping that index to an actual color from the color table. OpenGL allows for rendering to PseudoColor visuals, and it does so by being intentionally ignorant of the relationship between indices and actual colors. As far as OpenGL is concerned, each color index value is just a meaningless number, and it is only when the final image is drawn by the X server that these numbers take on meaning. As a result, many pieces of OpenGL’s core functionality either have undefined behavior or do not work at all with PseudoColor rendering. PseudoColor rendering used to be a common technique for visualizing scientific data, because such data often only contained 8 bits per sample to begin with. Applications could manipulate the color table to allow the user to dynamically control the relationship between sample values and colors. However, since most modern graphics cards have dropped support for PseudoColor rendering (it was removed as a feature in OpenGL 3.1), the applications that use it have become a vanishing breed.</p> <p>VirtualGL supports PseudoColor rendering if a PseudoColor visual is available on the 2D X server or X proxy. A PseudoColor visual need not be present on the 3D X server. On the 3D X server, VirtualGL uses the red channel of a standard RGB Pbuffer to store the color index. Upon receiving an end-of-frame trigger, VirtualGL reads back the red channel of the Pbuffer and uses <code>XPutImage()</code> to draw the color indices into the appropriate X window. To put this another way, PseudoColor rendering in VirtualGL always uses the X11 Transport. However, since there is only 1 byte per pixel in a PseudoColor “image”, reasonable performance can still be obtained when displaying to a remote X server on the same local-area network.</p> <p>VirtualGL’s PseudoColor rendering mode works with X proxies, provided that the X proxy provides a PseudoColor visual. Note, however, that some X proxies, such as VNC, cannot provide both PseudoColor and TrueColor visuals at the same time.</p> <h2 id="hd0017004">17.4 Troubleshooting</h2> <p><a name="Troubleshooting"></a></p> <p>VirtualGL includes a modified version of <code>glxinfo</code> that can be used to determine whether or not the 2D and 3D X servers have stereo, overlay, or PseudoColor visuals enabled.</p> <p>Run the following command sequence on the VirtualGL server to determine whether the 3D X server has a suitable visual for stereographic rendering:</p> <pre class="verbatim"> xauth merge /etc/opt/VirtualGL/vgl_xauth_key /opt/VirtualGL/bin/glxinfo -display :{n} -c -v </pre> <p>(where <code>{n}</code> is the display number of the 3D X server.) One or more of the visuals should say “stereo=1” and should list “Pbuffer” as one of the “Drawable Types.”</p> <p>Run the following command sequence on the VirtualGL server to determine whether the 2D X server has a suitable visual for stereographic rendering, transparent overlays, or PseudoColor.</p> <pre class="verbatim"> /opt/VirtualGL/bin/glxinfo -v </pre> <p>In order to use stereo, one or more of the visuals should say “stereo=1”. In order to use transparent overlays, one or more of the visuals should say “level=1”, should list a “Transparent Index” (non-transparent visuals will say “Opaque” instead), and should have a class of “PseudoColor.” In order to use PseudoColor (color index) rendering, one of the visuals should have a class of “PseudoColor.”</p> <p><br /></p> <hr class="break" /> <h1 id="hd0018"><a name="file018"></a>18 Performance Measurement</h1> <p><a name="Perf_Measurement"></a></p> <h2 id="hd0018001">18.1 VirtualGL’s Built-In Profiling System</h2> <p>The easiest way to uncover bottlenecks in VirtualGL’s image pipeline is to set the <code>VGL_PROFILE</code> environment variable to <code>1</code> on both server and client (passing an argument of <code>+pr</code> to <code>vglrun</code> on the server has the same effect.) This will cause VirtualGL to measure and report the throughput of various stages in the pipeline. For example, here are some measurements from a dual Pentium 4 server communicating with a Pentium III client on a 100 Megabit LAN:</p> <dl class="Description"> <dt class="Description-1 Description">Server</dt> <dd class="Description-1 Description"> <pre class="verbatim"> Readback - 43.27 Mpixels/sec - 34.60 fps Compress 0 - 33.56 Mpixels/sec - 26.84 fps Total - 8.02 Mpixels/sec - 6.41 fps - 10.19 Mbits/sec (18.9:1) </pre> </dd> <dt class="Description-1 Description">Client</dt> <dd class="Description-1 Description"> <pre class="verbatim"> Decompress - 10.35 Mpixels/sec - 8.28 fps Blit - 35.75 Mpixels/sec - 28.59 fps Total - 8.00 Mpixels/sec - 6.40 fps - 10.18 Mbits/sec (18.9:1) </pre> </dd> </dl> <p>The total throughput of the pipeline is 8.0 Megapixels/sec, or 6.4 frames/sec, indicating that our frame is 8.0 / 6.4 = 1.25 Megapixels in size (a little less than 1280 x 1024 pixels.) The readback and compress stages, which occur in parallel on the server, are obviously not slowing things down, and we’re only using 1/10 of our available network bandwidth. Looking at the client, however, we discover that its slow decompression speed (10.35 Megapixels/second) is the primary bottleneck. Decompression and blitting on the client cannot be done in parallel, so the aggregate performance is the harmonic mean of the decompression and blitting rates: <em>[1/ (1/10.35 + 1/35.75)] = 8.0 Mpixels/sec</em>. In this case, we could improve the performance of the whole system by simply using a client with a faster CPU.</p> <div class="important"><p class="important"> This example is meant to demonstrate how the client can sometimes be the primary impediment to VirtualGL’s end-to-end performance. Using “modern” hardware on both ends of the connection, VirtualGL can easily stream 50+ Megapixels/sec across a LAN, as of this writing. </p></div> <h2 id="hd0018002">18.2 Frame Spoiling</h2> <p><a name="Frame_Spoiling"></a></p> <p>By default, VirtualGL will only send a frame to the client if the client is ready to receive it. If VirtualGL detects that the application has finished rendering a new frame but there are already frames waiting in the queue to be processed, then those unprocessed frames are dropped (“spoiled”) and the new frame is promoted to the head of the queue. This prevents a backlog of frames on the server, which would cause a perceptible delay in the responsiveness of interactive applications. However, when running non-interactive applications, particularly benchmarks, frame spoiling should always be disabled. With frame spoiling disabled, the server will render frames only as quickly as VirtualGL can send those frames to the client, which will conserve server resources as well as allow OpenGL benchmarks to accurately measure the end-to-end performance of VirtualGL (assuming that the VGL Transport is used.) With frame spoiling enabled, OpenGL benchmarks will report meaningless data, since the rate at which the server can render frames is decoupled from the rate at which VirtualGL can send those frames to the client.</p> <p>In most X proxies (including VNC), there is effectively another layer of frame spoiling, since the rate at which the X proxy can send frames to the client is decoupled from the rate at which VirtualGL can draw images into the X proxy. Thus, even if frame spoiling is disabled in VirtualGL, OpenGL benchmarks will still report inaccurate data if they are run in such X proxies. TCBench, described below, provides a limited solution to this problem.</p> <p>To disable frame spoiling, set the <code>VGL_SPOIL</code> environment variable to <code>0</code> on the VirtualGL server or pass an argument of <code>-sp</code> to <code>vglrun</code>. See Section <a href="#VGL_SPOIL" class="ref">20.1</a> for further information.</p> <h2 id="hd0018003">18.3 VirtualGL Diagnostic Tools</h2> <p>VirtualGL includes several tools that can be useful in diagnosing performance problems with the system.</p> <h3 id="hd0018003001">NetTest</h3> <p>NetTest is a network benchmark that uses the same network I/O classes as VirtualGL. It can be used to test the latency and throughput of any TCP/IP connection, with or without SSL encryption. <code>nettest</code> can be found in <code>/opt/VirtualGL/bin</code> on Linux/Unix/Mac/Cygwin VirtualGL installations or in <code>c:\program files\VirtualGL-</code><em><code>{version}</code></em><code>-</code><em><code>{build}</code></em> if using the VirtualGL Client for Exceed.</p> <p>To use NetTest, first start up the NetTest server on one end of the connection:</p> <pre class="verbatim"> nettest -server [-ssl] </pre> <p>(Use <code>-ssl</code> if you want to test the performance of SSL encryption over this particular connection. VirtualGL must have been compiled with OpenSSL support for this option to be available.)</p> <p>Next, start the client on the other end of the connection:</p> <pre class="verbatim"> nettest -client {server} [-ssl] </pre> <p>Replace <em><code>{server}</code></em> with the hostname or IP address of the machine on which the NetTest server is running. (Use <code>-ssl</code> if the NetTest server is running in SSL mode. VirtualGL must have been compiled with OpenSSL support for this option to be available.)</p> <p>The NetTest client will produce output similar to the following:</p> <pre class="verbatim"> TCP transfer performance between localhost and {server}: Transfer size 1/2 Round-Trip Throughput Throughput (bytes) (msec) (MBytes/sec) (Mbits/sec) 1 0.093402 0.010210 0.085651 2 0.087308 0.021846 0.183259 4 0.087504 0.043594 0.365697 8 0.088105 0.086595 0.726409 16 0.090090 0.169373 1.420804 32 0.093893 0.325026 2.726514 64 0.102289 0.596693 5.005424 128 0.118493 1.030190 8.641863 256 0.146603 1.665318 13.969704 512 0.205092 2.380790 19.971514 1024 0.325896 2.996542 25.136815 2048 0.476611 4.097946 34.376065 4096 0.639502 6.108265 51.239840 8192 1.033596 7.558565 63.405839 16384 1.706110 9.158259 76.825049 32768 3.089896 10.113608 84.839091 65536 5.909509 10.576174 88.719379 131072 11.453894 10.913319 91.547558 262144 22.616389 11.053931 92.727094 524288 44.882406 11.140223 93.450962 1048576 89.440702 11.180592 93.789603 2097152 178.536997 11.202160 93.970529 4194304 356.754396 11.212195 94.054712 </pre> <p>We can see that the throughput peaks at about 94 megabits/sec, which is pretty good for a 100 Megabit connection. We can also see that, for small transfer sizes, the round-trip time is dominated by latency. The “latency” is the same thing as the one-way (1/2 round-trip) transit time for a zero-byte packet, which is about 93 microseconds in this case.</p> <h3 id="hd0018003002">CPUstat</h3> <p>CPUstat is available only for Linux and is installed in the same place as NetTest (<code>/opt/VirtualGL/bin</code>). It measures the average, minimum, and peak CPU usage for all processors combined and for each processor individually. On Windows, this same functionality is provided in the Windows Performance Monitor, which is part of the operating system. On Solaris, the same data can be obtained through <code>vmstat</code>.</p> <p>CPUstat measures the CPU usage over a given sample period (a few seconds) and continuously reports how much the CPU was utilized since the last sample period. Output for a particular sample looks something like this:</p> <pre class="verbatim"> ALL : 51.0 (Usr= 47.5 Nice= 0.0 Sys= 3.5) / Min= 47.4 Max= 52.8 Avg= 50.8 cpu0: 20.5 (Usr= 19.5 Nice= 0.0 Sys= 1.0) / Min= 19.4 Max= 88.6 Avg= 45.7 cpu1: 81.5 (Usr= 75.5 Nice= 0.0 Sys= 6.0) / Min= 16.6 Max= 83.5 Avg= 56.3 </pre> <p>The first column indicates what percentage of time the CPU was active since the last sample period (this is then broken down into what percentage of time the CPU spent running user, nice, and system/kernel code.) “ALL” indicates the average utilization across all CPUs since the last sample period. “Min”, “Max”, and “Avg” indicate a running minimum, maximum, and average of all samples since CPUstat was started.</p> <p>Generally, if an application’s CPU usage is fairly steady, you can run CPUstat for a bit and wait for the Max. and Avg. for the “ALL” category to stabilize, then that will tell you what the application’s peak and average % CPU utilization is.</p> <h3 id="hd0018003003">TCBench</h3> <p>TCBench was born out of the need to compare VirtualGL’s performance to that of other thin client software, some of which had frame spoiling features that could not be disabled. TCBench measures the frame rate of a thin client system as seen from the client’s point of view. It does this by attaching to one of the client windows and continuously reading back a small area at the center of the window. While this may seem to be a somewhat non-rigorous test, experiments have shown that, if care is taken to ensure that the application is updating the center of the window on every frame (such as in a spin animation), TCBench can produce quite accurate results. It has been sanity checked with VirtualGL’s internal profiling mechanism and with a variety of system-specific techniques, such as monitoring redraw events on the client’s windowing system.</p> <p>TCBench can be found in <code>/opt/VirtualGL/bin</code> on Linux/Unix/Mac/Cygwin VirtualGL installations or in <code>c:\program files\VirtualGL-</code><em><code>{version}</code></em><code>-</code><em><code>{build}</code></em> if using the VirtualGL Client for Exceed. Run <code>tcbench</code> from the command line, and it will prompt you to click in the window you want to benchmark. That window should already have an automated animation of some sort running before you launch TCBench. Note that GLXSpheres (see below) is an ideal benchmark to use with TCBench, since GLXSpheres draws a new sphere to the center of its window on every frame.</p> <pre class="verbatim"> tcbench -? </pre> <p>lists the relevant command-line arguments, which can be used to adjust the benchmark time, the sampling rate, and the x and y offset of the sampling area within the window.</p> <h3 id="hd0018003004">GLXSpheres</h3> <p>GLXSpheres is a benchmark that produces very similar images to nVidia’s (long-discontinued) SphereMark benchmark. Back in the early days of VirtualGL’s existence, it was discovered (quite by accident) that SphereMark was a pretty good test of VirtualGL’s end-to-end performance, because that benchmark generated images with about the same proportion of solid color and similar frequency components to the images generated by volume visualization applications.</p> <p>Thus, the goal of GLXSpheres was to create an open source Unix version of SphereMark (the original SphereMark was for Windows only) completely from scratch. GLXSpheres does not use any code from the original benchmark, but it does attempt to mimic the visual output of the original as closely as possible. GLXSpheres lacks some of the advanced rendering features of the original, such as the ability to use vertex arrays, but since GLXspheres was primarily designed as a benchmark for VirtualGL, display lists are more than fast enough for that purpose.</p> <p>GLXSpheres has some additional modes that its predecessor lacked, modes that are designed specifically to test the performance of various VirtualGL features:</p> <dl class="Description"> <dt class="Description-1 Description">Stereographic rendering (<code>glxspheres -s</code>)</dt> <dd class="Description-1 Description"> </dd> <dt class="Description-1 Description">Color index rendering (<code>glxspheres -c</code>)</dt> <dd class="Description-1 Description"> In color index mode, GLXSpheres will draw the spheres using an 8-bit color map and will change the color map periodically. </dd> <dt class="Description-1 Description">Overlay rendering (<code>glxspheres -o</code>)</dt> <dd class="Description-1 Description"> This renders text, a moving crosshair cursor, and a block of pixels to an 8-bit transparent overlay while animating the spheres on the underlay. The color map of the overlay is changed periodically. </dd> <dt class="Description-1 Description">Immediate mode rendering (<code>glxspheres -m</code>)</dt> <dd class="Description-1 Description"> Want to really see the benefit of VirtualGL? Run <code>glxspheres -m</code> over a remote X connection, then run <code>vglrun -sp glxspheres -m</code> over the same connection and compare. Immediate mode does not use display lists, so when immediate mode OpenGL is rendered indirectly (over a remote X connection), this causes every OpenGL command to be sent as a separate network request to the X server … with every frame. Many applications do not use display lists (because the geometry they are rendering is dynamic, or for other reasons), so this test models how such applications might perform when displayed remotely without VirtualGL. </dd> <dt class="Description-1 Description">Interactive mode (<code>glxspheres -i</code>)</dt> <dd class="Description-1 Description"> In interactive mode, GLXSpheres will wait to draw a frame until it receives a mouse event. Continuously dragging the mouse in the window should produce a steady frame rate, and this frame rate is a reasonable model of the frame rate that you can achieve when running interactive applications in VirtualGL. Comparing this interactive frame rate (<code>vglrun glxspheres -i</code>) with the non-interactive frame rate (<code>vglrun -sp glxspheres</code>) allows you to quantify the effect of X latency on the performance of interactive applications in a VirtualGL environment. </dd> </dl> <p>GLXSpheres is installed in <code>/opt/VirtualGL/bin</code> on Linux and Unix VirtualGL servers. 64-bit VirtualGL packages name this program <code>glxspheres64</code> so as to allow both a 64-bit and a 32-bit version of GLXSpheres to be installed on the same system.</p> <p><br /></p> <hr class="break" /> <h1 id="hd0019"><a name="file019"></a>19 The VirtualGL Configuration Dialog</h1> <p><a name="Config_Dialog"></a></p> <p>Several of VirtualGL’s configuration parameters can be changed on the fly once a 3D application has been started. This is accomplished by using the VirtualGL Configuration dialog, which can be popped up by holding down the <code>CTRL</code> and <code>SHIFT</code> keys and pressing the <code>F9</code> key while any one of the 3D application’s windows is active. This displays the following dialog box:</p> <p><img src="configdialog.gif" alt="configdialog" class="inline" id="imgid_5" name="imgid_5"/></p> <p>You can use this dialog to adjust various image compression and display parameters in VirtualGL. Changes are communicated immediately to VirtualGL.</p> <dl class="Description"> <dt class="Description-1 Description">Image Compression (Transport)</dt> <dd class="Description-1 Description"> This is a drop-down menu with the following options: <br /><br /> <em>None (X11 Transport)</em> : equivalent to setting <code>VGL_COMPRESS=proxy</code>. This option can be activated at any time, regardless of which transport was active when VirtualGL started. <br /><br /> <em>JPEG (VGL Transport)</em> : equivalent to setting <code>VGL_COMPRESS=jpeg</code>. This option is only available if the VGL Transport was active when VirtualGL started. <br /><br /> <em>RGB (VGL Transport)</em> : equivalent to setting <code>VGL_COMPRESS=rgb</code>. This option is only available if the VGL Transport was active when VirtualGL started. <br /><br /> <em>YUV (XV Transport)</em> : equivalent to setting <code>VGL_COMPRESS=xv</code>. This option is only available if the 2D X server has the X Video extension and the X Video implementation supports the YUV420P (AKA “I420”) pixel format. <br /><br /> <em>YUV (VGL Transport)</em> : equivalent to setting <code>VGL_COMPRESS=yuv</code>. This option is only available if the 2D X server has the X Video extension, the X Video implementation supports the YUV420P (AKA “I420”) pixel format, and the VGL Transport was active when VirtualGL started. <br /><br /> See Section <a href="#VGL_COMPRESS" class="ref">20.1</a> for more information about the <code>VGL_COMPRESS</code> configuration option. <div class="important"><p class="important"> If an image transport plugin is loaded, then this menu’s name changes to “Image Compression”, and it has options “0” through “10”. </p></div> </dd> <dt class="Description-1 Description">Chrominance Subsampling</dt> <dd class="Description-1 Description"> This drop-down menu is active only when using JPEG compression or an image transport plugin. It has the following options: <br /><br /> <em>Grayscale</em> : equivalent to setting <code>VGL_SUBSAMP=gray</code> <br /><br /> <em>1X</em> : equivalent to setting <code>VGL_SUBSAMP=1x</code> <br /><br /> <em>2X</em> : equivalent to setting <code>VGL_SUBSAMP=2x</code> <br /><br /> <em>4X</em> : equivalent to setting <code>VGL_SUBSAMP=4x</code> <br /><br /> See Section <a href="#VGL_SUBSAMP" class="ref">20.1</a> for more information about the <code>VGL_SUBSAMP</code> configuration option. <div class="important"><p class="important"> If an image transport plugin is loaded, then this menu has two additional options, “8X” and “16X”. </p></div> </dd> <dt class="Description-1 Description">JPEG Image Quality</dt> <dd class="Description-1 Description"> This slider gadget is active only when using JPEG compression or an image transport plugin. It is the equivalent of setting <code>VGL_QUAL</code>. See Section <a href="#VGL_QUAL" class="ref">20.1</a> for more information about the <code>VGL_QUAL</code> configuration option. <div class="important"><p class="important"> If an image transport plugin is loaded, then this gadget’s name changes to “Image Quality”. </p></div> </dd> <dt class="Description-1 Description">Connection Profile</dt> <dd class="Description-1 Description"> This drop-down menu is active only if the VGL Transport was active when VirtualGL started. It has the following options: <br /><br /> <em>Low Qual (Low-Bandwidth Network)</em> : Sets the image compression type to JPEG (VGL Transport), sets the Chrominance Subsampling to 4X, and sets the JPEG Image Quality to 30. <br /><br /> <em>Medium Qual</em> : Sets the image compression type to JPEG (VGL Transport), sets the Chrominance Subsampling to 2X, and sets the JPEG Image Quality to 80. <br /><br /> <em>High Qual (High-Bandwidth Network)</em> : Sets the image compression type to JPEG (VGL Transport), sets the Chrominance Subsampling to 1X, and sets the JPEG Image Quality to 95. </dd> <dt class="Description-1 Description">Gamma Correction Factor</dt> <dd class="Description-1 Description"> This floating point input gadget is the equivalent of setting <code>VGL_GAMMA</code>. This enables VirtualGL’s internal gamma correction system with the specified gamma correction factor. See Section <a href="#VGL_GAMMA" class="ref">20.1</a> for more information about the <code>VGL_GAMMA</code> configuration option. </dd> <dt class="Description-1 Description">Frame Spoiling</dt> <dd class="Description-1 Description"> This toggle button is the equivalent of setting <code>VGL_SPOIL</code>. See Section <a href="#Frame_Spoiling" class="ref">18.2</a> and Section <a href="#VGL_SPOIL" class="ref">20.1</a> for more information about the <code>VGL_SPOIL</code> configuration option. </dd> <dt class="Description-1 Description">Interframe Comparison</dt> <dd class="Description-1 Description"> This toggle button is the equivalent of setting <code>VGL_INTERFRAME</code>. See Section <a href="#VGL_INTERFRAME" class="ref">20.1</a> for more information about the <code>VGL_INTERFRAME</code> configuration option. </dd> <dt class="Description-1 Description">Stereographic Rendering Method</dt> <dd class="Description-1 Description"> This drop-down menu has the following options: <br /><br /> <em>Send Left Eye Only</em> : equivalent to setting <code>VGL_STEREO=left</code>. <br /><br /> <em>Send Right Eye Only</em> : equivalent to setting <code>VGL_STEREO=right</code> <br /><br /> <em>Quad-Buffered (if available)</em> : equivalent to setting <code>VGL_STEREO=quad</code> <br /><br /> <em>Anaglyphic (Red/Cyan)</em> : equivalent to setting <code>VGL_STEREO=rc</code> <br /><br /> <em>Anaglyphic (Green/Magenta)</em> : equivalent to setting <code>VGL_STEREO=gm</code> <br /><br /> <em>Anaglyphic (Blue/Yellow)</em> : equivalent to setting <code>VGL_STEREO=by</code> <br /><br /> <em>Passive (Interleaved)</em> : equivalent to setting <code>VGL_STEREO=i</code> <br /><br /> <em>Passive (Top/Bottom)</em> : equivalent to setting <code>VGL_STEREO=tb</code> <br /><br /> <em>Passive (Side-by-Side)</em> : equivalent to setting <code>VGL_STEREO=ss</code> <br /><br /> See Section <a href="#VGL_STEREO" class="ref">20.1</a> for more information about the <code>VGL_STEREO</code> configuration option. </dd> <dt class="Description-1 Description">Limit Frames/second</dt> <dd class="Description-1 Description"> This floating point input gadget is the equivalent of setting <code>VGL_FPS</code>. See Section <a href="#VGL_FPS" class="ref">20.1</a> for more information about the <code>VGL_FPS</code> configuration option. </dd> </dl> <p>You can set the <code>VGL_GUI</code> environment variable to change the key sequence used to pop up the VirtualGL Configuration dialog. If the default of <code>CTRL-SHIFT-F9</code> is not suitable, then set <code>VGL_GUI</code> to any combination of <code>ctrl</code>, <code>shift</code>, <code>alt</code>, and one of <code>f1, f2,..., f12</code> (these are not case sensitive.) For example:</p> <pre class="verbatim"> export VGL_GUI=CTRL-F9 </pre> <p>will cause the dialog box to pop up whenever <code>CTRL-F9</code> is pressed.</p> <p>To disable the VirtualGL dialog altogether, set <code>VGL_GUI</code> to <code>none</code>.</p> <div class="important"><p class="important"> VirtualGL monitors the application’s X event loop to determine whenever a particular key sequence has been pressed. If an application is not monitoring key press events in its X event loop, then the VirtualGL Configuration dialog might not pop up at all. There is unfortunately no workaround for this, but it should be a rare occurrence. </p></div> <p><br /></p> <hr class="break" /> <h1 id="hd0020"><a name="file020"></a>20 Advanced Configuration</h1> <p><a name="Advanced_Configuration"></a></p> <h2 id="hd0020001">20.1 Server Settings</h2> <p>You can control the operation of the VirtualGL faker in four different ways. Each method of configuration takes precedence over the previous method:</p> <ol class="Ordered numeric"> <li class="Ordered-1 Ordered"> Setting a configuration environment variable globally (for instance, in <code>/etc/profile</code>) </li> <li class="Ordered-1 Ordered"> Setting a configuration environment variable on a per-user basis (for instance, in <code>~/.bashrc</code>) </li> <li class="Ordered-1 Ordered"> Setting a configuration environment variable only for the current shell session (for instance, <code>export VGL_XXX={whatever}</code>) </li> <li class="Ordered-1 Ordered"> Passing a configuration option as an argument to <code>vglrun</code>. This effectively overrides any previous environment variable setting corresponding to that configuration option. </li> </ol> <p></p> <div class="important"><p class="important"> If “Custom (if supported)” is listed as one of the available Image Transports, then this means that image transport plugins are free to handle or ignore the configuration option as they see fit. </p></div> <p><a name="VGL_ALLOWINDIRECT"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_ALLOWINDIRECT = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Allow applications to request an indirect OpenGL context</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">0 (all OpenGL contexts use direct rendering, unless rendering to a <a href="#overlays">transparent overlay</a><a name="idx0049"></a>)</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, when VirtualGL maps a Pbuffer to a window and establishes an OpenGL rendering context with the Pbuffer, it forces direct rendering to be used with this context. Some 3D applications erroneously try to create indirect OpenGL contexts because they detect that the X display is remote and assume that the 3D rendering commands will be sent over the network. Thus, VirtualGL normally forces all contexts to be direct in order to prevent severe readback performance degradation with such apps (even on modern 3D adapters, and even when the connection to the 3D X server is local, <code>glReadPixels()</code> can perform very slowly if an indirect OpenGL context is used.) <br /><br /> However, some applications intentionally try to create indirect contexts so that these contexts can be shared, and those apps may not work properly when the contexts are forced to be direct. For such apps, setting <code>VGL_ALLOWINDIRECT</code> to <code>1</code> will cause VirtualGL to honor the application’s request for an indirect OpenGL context. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_CLIENT = </code><em><code>{c}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-cl </code><em><code>{c}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{c}</code></em> = the hostname or IP address of the VirtualGL client</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Automatically set by <code>vglconnect</code> or <code>vglrun</code></td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> When using the VGL Transport, <code>VGL_CLIENT</code> should be set to the hostname or IP address of the machine on which <code>vglclient</code> is running. Normally, <code>VGL_CLIENT</code> is set automatically when executing <code>vglconnect</code> or <code>vglrun</code>, so don’t override it unless you know what you’re doing. </dd> </dl> <p><a name="VGL_COMPRESS"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_COMPRESS = </code><em><code>proxy | jpeg | rgb | xv | yuv</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-c </code><em><code>proxy | jpeg | rgb | xv | yuv</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Set image transport and image compression type</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">(See description)</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> <em>proxy</em> = Send images uncompressed using the X11 Transport. This is useful when displaying to a local 2D X server or X proxy (see Section <a href="#X11_Proxy_Usage_Local" class="ref">9.1</a>.) <br /><br /> <em>jpeg</em> = Compress images using JPEG and send using the VGL Transport. This is useful when displaying to a remote 2D X server (see Chapter <a href="#VGL_Transport_Usage" class="ref">8</a>.) <br /><br /> <em>rgb</em> = Encode images as uncompressed RGB and send using the VGL Transport. This is useful when displaying to a remote 2D X server or X proxy across a very fast network (see Section <a href="#X11_Proxy_Usage_Remote" class="ref">9.2</a>.) <br /><br /> <em>xv</em> = Encode images as YUV420P (planar YUV with 4X chrominance subsampling) and display them to the 2D X server using the XV Transport. This transport is designed for use with X proxies that support the X Video extension (see Chapter <a href="#X_Video_Support" class="ref">10</a>.) <br /><br /> <em>yuv</em> = Encode images as YUV420P, send using the VGL Transport, and display on the client machine using the X Video extension. This greatly reduces the CPU usage on both server and client and uses only about half the network bandwidth as RGB, but the use of 4X chrominance subsampling does produce some visible artifacts (see Chapter <a href="#X_Video_Support" class="ref">10</a>.) <br /><br /> If <code>VGL_COMPRESS</code> is not specified, then the default is set as follows: <br /><br /> If the <code>DISPLAY</code> environment variable begins with <code>:</code> or <code>unix:</code>, then VirtualGL assumes that the X display connection is local and will default to using <em>proxy</em> compression. <br /><br /> If VirtualGL detects that the 2D X server is remote, then it will default to using <em>jpeg</em> compression. <div class="important"><p class="important"> If an image transport plugin is being used, then you can set <code>VGL_COMPRESS</code> to any numeric value >= 0 (Default value = 0.) The plugin can choose to respond to this value as it sees fit. </p></div> </dd> </dl> <p><a name="VGL_DEFAULTFBCONFIG"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_DEFAULTFBCONFIG = </code><em><code>{attrib_list}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{attrib_list}</code></em> = Attributes of the default GLX framebuffer config, which VirtualGL uses if a 3D application does not call <code>glXChooseVisual()</code> to specify the visual attributes it desires</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">None</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, a Unix OpenGL application would call the <code>glXChooseVisual()</code> function to obtain an X11 visual with a desired set of OpenGL attributes (such as a specific Z buffer depth, etc.) The application would then use that X visual when creating an X window for OpenGL rendering. VirtualGL’s fundamental purpose is to redirect OpenGL rendering from a window on one X server (the 2D X server) to a Pbuffer on another X server (the 3D X server.) Thus, for every OpenGL-enabled X visual that the application tries to obtain, VirtualGL needs to obtain an equivalent “GLX FB config”, which is like an X visual for Pbuffers. VirtualGL does this by intercepting <code>glXChooseVisual()</code> and using the attributes passed to that function to build an attribute list for <code>glXChooseFBConfig()</code>, which is called on the 3D X server. The FB config returned from <code>glXChooseFBConfig()</code> is mapped internally to an X visual on the 2D X server, and that visual is returned from <code>glXChooseVisual()</code>. The FB config is later used when creating the Pbuffer that backs a 3D application window. <br /><br /> In rare cases, an application may choose to implement its own visual selection mechanism rather than call <code>glXChooseVisual()</code>. Such applications will iterate through the list of X visuals and examine the OpenGL attributes of each using <code>glXGetConfig()</code>. The problem is this: whereas in a “normal” GLX environment, there would be a 1:1 correspondence between X visuals and GLX FB configs, in VirtualGL’s split rendering environment, X visuals are on the 2D X server and GLX FB configs are on the 3D X server. Thus, if an application calls <code>glXGetConfig()</code> before calling <code>glXChooseVisual()</code>, VirtualGL has not yet mapped the X visual in question to a GLX FB config, and furthermore, VirtualGL has no idea what type of visual the application is looking for. In such cases, VGL has to map the visual to a default FB config. Since this default FB config is very basic, if the application is hunting for a visual with a particular OpenGL attribute (such as an alpha channel or a stencil buffer), then it may fail to find one. <br /><br /> <code>VGL_DEFAULTFBCONFIG</code> allows the user to specify the attributes of VirtualGL’s default FB config. This may be necessary to make certain applications work, if those applications do not use <code>glXChooseVisual()</code> to obtain a visual. The attribute list is specified in the same way that you would specify an attribute list for <code>glXChooseFBConfig()</code>. Example: <code>VGL_DEFAULTFBCONFIG = GLX_ALPHA_SIZE,8,GLX_STENCIL_SIZE,8</code>. See <a href="#Application_Recipes">Application Recipes</a><a name="idx0050"></a> for a list of applications that are known to require the use of this configuration option. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_DISPLAY = </code><em><code>{d}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-d </code><em><code>{d}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{d}</code></em> = the X display to use for 3D rendering</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">:0</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> If the VirtualGL server has multiple GPUs, each attached to a separate X screen or a separate X server, then you can use this option to specify which GPU should be used for 3D rendering. For instance, setting <code>VGL_DISPLAY</code> to (or invoking <code>vglrun -d</code> with) <code>:0.1</code> would cause VirtualGL to redirect all of the 3D rendering from the application to a GPU attached to Screen 1 on X display :0. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_DRAWABLE = </code><em><code>pbuffer | pixmap</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Specify the drawable type to use for 3D rendering</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">pbuffer</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, VirtualGL will redirect the OpenGL rendering from each of the application’s windows on the 2D X server into a corresponding Pbuffer on the 3D X server, and it will likewise redirect the OpenGL rendering from each of the application’s pixmaps on the 2D X server into a corresponding pixmap on the 3D X server. Setting <code>VGL_DRAWABLE</code> to <code>pixmap</code> causes VirtualGL to use pixmaps on the 3D X server to represent all of the application’s OpenGL drawables. This is meant as an advanced option for those who are experimenting with VirtualGL in unsupported environments that may lack complete or stable Pbuffer support. </dd> </dl> <p><a name="VGL_FORCEALPHA"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_FORCEALPHA = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Force the Pbuffers used for 3D rendering to have an 8-bit alpha channel</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard"><code>VGL_FORCEALPHA=1</code> if PBO readback mode is used, <code>VGL_FORCEALPHA=0</code> otherwise</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, VirtualGL will create Pbuffers whose attributes match those of the visuals requested by the 3D application. Setting <code>VGL_FORCEALPHA</code> to <code>1</code> causes VirtualGL to always create Pbuffers with alpha channels. This means that a 32-bit-per-pixel (BGRA) Pbuffer will be created if the application requests a 24-bit-per-pixel visual. <br /><br /> The primary purpose of this option is to work around a limitation of certain consumer-grade GPUs, whereby the pixel format requested by the pixel readback operation must match the pixel format of the Pbuffer in order for pixel buffer objects (PBOs) to behave correctly. Since displaying to an X proxy typically requires VirtualGL to read back pixels in the BGRA format, enabling <code>VGL_FORCEALPHA</code> might be necessary in order to use PBO readback mode with the afore-mentioned GPUs (as of this writing, nVidia GeForce adapters are known to require this.) See the <a href="#VGL_READBACK"><code>VGL_READBACK</code></a><a name="idx0051"></a> option for further information. <div class="important"><p class="important"> <code>VGL_FORCEALPHA</code> overrides the application’s choice of visuals. It has no effect if the application is not explicitly choosing a visual. In that case, use <a href="#VGL_DEFAULTFBCONFIG"><code>VGL_DEFAULTFBCONFIG</code></a><a name="idx0052"></a> instead. </p></div> </dd> </dl> <p><a name="VGL_FPS"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_FPS = </code><em><code>{f}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-fps </code><em><code>{f}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Limit the end-to-end frame rate to <em><code>{f}</code></em> frames/second, where <em><code>{f}</code></em> is a floating point number > 0.0</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, X11, XV, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">0.0 (No limit)</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> This option prevents VirtualGL from sending frames at a rate faster than the specified limit. It can be used, for instance, as a crude way to control network bandwidth or CPU usage in multi-user environments in which those resources are constrained. <br /><br /> If frame spoiling is disabled, then setting <code>VGL_FPS</code> effectively limits the server’s 3D rendering frame rate as well. </dd> </dl> <p><a name="VGL_GAMMA"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_GAMMA = </code><em><code>{g}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-gamma </code><em><code>{g}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Specify gamma correction factor</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">1.00 (no gamma correction)</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> “Gamma” refers to the relationship between the intensity of light that your computer’s monitor is instructed to display and the intensity that it actually displays. The curve is an exponential curve of the form <em>Y = X<sup>G</sup></em>, where X is between 0 and 1. G is called the “gamma” of the monitor. PC monitors and TVs usually have a gamma of around 2.2. <br /><br /> Some of the math involved in 3D rendering assumes a linear gamma (G = 1.0), so technically speaking, 3D applications will not display with mathematical correctness unless the pixels are “gamma corrected” to counterbalance the non-linear response curve of the monitor. However, some systems do not have any form of built-in gamma correction, so the applications developed for such systems have usually been designed to display properly without gamma correction. Gamma correction involves passing pixels through a function of the form <em>X = W<sup>1/G</sup></em>, where G is the “gamma correction factor” and should be equal to the gamma of the monitor. So, the final output is <em>Y = X<sup>G</sup> = (W<sup>1/G</sup>)<sup>G</sup> = W</em>, which describes a linear relationship between the intensity of the pixels drawn by the application and the intensity of the pixels displayed by the monitor. <br /><br /> If <code>VGL_GAMMA</code> is set to an arbitrary floating point value, then VirtualGL will perform gamma correction on all of the rendered 3D images from the application, using the specified value as the gamma correction factor. You can also specify a negative value to apply a “de-gamma” function. Specifying a gamma correction factor of G (where G < 0) is equivalent to specifying a gamma correction factor of -1/G. </dd> </dl> <p><a name="VGL_GLFLUSHTRIGGER"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_GLFLUSHTRIGGER = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable reading back and sending the front buffer contents whenever the 3D application calls <code>glFlush()</code> while rendering to the front buffer</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Enabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> <code>glFlush()</code> is a sort of “asynchronous synchronization” command. It flushes the OpenGL command buffers, which generally has the effect of ensuring that the commands have been delivered to the GPU. However, unlike <code>glFinish()</code>, <code>glFlush()</code> does not wait until the commands have been rendered before it returns. <br /><br /> The use of <code>glFlush()</code> can vary widely from application to application. When doing front buffer rendering, some applications call <code>glFlush()</code> after each object is rendered. Some call it only at the end of the frame. Others call <code>glFlush()</code> much more often, even as frequently as every time a few primitives are rendered. This creates problems for VirtualGL, since it has to guess what the application is intending to do. Not all applications that use front buffer rendering call <code>glFinish()</code> to signal the end of a frame, so VirtualGL cannot usually get away with ignoring <code>glFlush()</code>. However, some applications call <code>glFlush()</code> so often that VirtualGL cannot get away with reading back/sending a frame every time <code>glFlush()</code> is called, either (see <a href="#VGL_SPOILLAST">VGL_SPOILLAST</a><a name="idx0053"></a> for more information on how VirtualGL tries to handle this, under normal circumstances.) <br /><br /> Some 3D applications use <code>glFlush()</code> very liberally and intend for it to be an intermediate rather than a final synchronization command. Such applications will call <code>glFinish()</code> after a sequence of <code>glFlush()</code> calls, so for those applications, reading back and sending the rendered 3D image in response to <code>glFlush()</code> calls is a waste of resources and can sometimes create visual artifacts (for instance, if the application clears the front buffer with a particular color, calls <code>glFlush()</code>, then clears it again with another color. We wouldn’t mention it if it hadn’t happened before.) For such applications, setting <code>VGL_GLFLUSHTRIGGER</code> to <code>0</code> should make them display properly in VirtualGL. See <a href="#Application_Recipes">Application Recipes</a><a name="idx0054"></a> for a list of applications that are known to require this. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_GLLIB = </code><em><code>{l}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{l}</code></em> = the location of an alternate OpenGL library</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, VirtualGL will try to use the OpenGL library against which it was linked (usually <code>libGL.so.1</code>, in the system library path) to make any GLX or OpenGL calls it needs to make. Failing this, VirtualGL will try to use the first compatible library named <code>libGL.so.1</code> that is found in the dynamic loader path. You can use the <code>VGL_GLLIB</code> environment variable to override this behavior and specify a dynamic library from which VirtualGL should call “real” GLX and OpenGL functions. For instance, when VirtualGL intercepts a <code>glXSwapBuffers()</code> call from the application, it modifies the arguments to redirect the function to the 3D X server, then it calls the “real” <code>glXSwapBuffers()</code> function. <code>VGL_GLLIB</code> allows one to specify the library from which this “real” function (and others) should be called. <br /><br /> You shouldn’t need to change this unless something doesn’t work. However, setting this environment variable is necessary when using VirtualGL with <a href="#Chromium">Chromium</a><a name="idx0055"></a>. It is also potentially useful if one wishes to insert another OpenGL interposer between VirtualGL and the system’s OpenGL library. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_GUI = </code><em><code>{k}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{k}</code></em> = the key sequence used to pop up the VirtualGL Configuration dialog, or <code>none</code> to disable the dialog</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">shift-ctrl-f9</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> VirtualGL will normally monitor an application’s X event queue and pop up the VirtualGL Configuration dialog whenever <code>CTRL-SHIFT-F9</code> is pressed. In the event that this interferes with a key sequence that the application is already using, then you can redefine the key sequence used to pop up the VirtualGL Configuration dialog by setting <code>VGL_GUI</code> to some combination of <code>shift</code>, <code>ctrl</code>, <code>alt</code>, and one of <code>f1, f2, ..., f12</code>. You can also set <code>VGL_GUI</code> to <code>none</code> to disable the configuration dialog altogether. See Chapter <a href="#Config_Dialog" class="ref">19</a> for more details. </dd> </dl> <p><a name="VGL_INTERFRAME"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_INTERFRAME = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Enable or disable interframe image comparison</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL (JPEG, RGB), Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Enabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> The VGL Transport will normally compare each frame with the previous frame and send only the portions of the image that have changed. Setting <code>VGL_INTERFRAME</code> to <code>0</code> disables this behavior. <br /><br /> This setting was introduced in order to work around a specific application interaction issue, but since a proper fix for that issue was introduced in VirtualGL 2.1.1, this option isn’t really useful anymore. <div class="important"><p class="important"> When using the VGL Transport, interframe comparison is affected by the <a href="#VGL_TILESIZE"><code>VGL_TILESIZE</code></a><a name="idx0056"></a> option </p></div> </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_LOG = </code><em><code>{l}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Redirect all messages from VirtualGL to a log file specified by <em><code>{l}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Print all messages to stderr</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Setting this environment variable to the pathname of a log file on the VirtualGL server will cause VirtualGL to redirect all of its messages (including profiling and trace output) to the specified log file rather than to stderr. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_LOGO = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Enable or disable the display of a VGL logo in the 3D window(s)</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Setting <code>VGL_LOGO</code> to <code>1</code> will cause VirtualGL to display a small logo in the bottom right-hand corner of all of the application’s 3D windows. This is meant as a debugging tool to allow users to determine whether or not VirtualGL is active. </dd> </dl> <p><a name="VGL_NPROCS"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_NPROCS = </code><em><code>{n}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-np </code><em><code>{n}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{n}</code></em> = the number of CPUs to use for multi-threaded compression</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL (JPEG, RGB), Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">1</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> The VGL Transport can divide the task of compressing each frame among multiple server CPUs. This might speed up the overall throughput in rare circumstances in which the server CPUs are significantly slower than the client CPUs. <br /><br /> VirtualGL will not allow more than 4 CPUs total to be used for compression, nor will it allow you to set this parameter to a value greater than the number of CPUs in the system. <div class="important"><p class="important"> When using the VGL Transport, multi-threaded compression is affected by the <a href="#VGL_TILESIZE"><code>VGL_TILESIZE</code></a><a name="idx0057"></a> option </p></div> </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_PORT = </code><em><code>{p}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-p </code><em><code>{p}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{p}</code></em> = the TCP port to use when connecting to the VirtualGL Client</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Read from X property stored by VirtualGL Client</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> The connection port for the VGL Transport is normally determined by reading an X property that <code>vglclient</code> stores on the 2D X server, so don’t override this unless you know what you’re doing. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_PROFILE = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-pr</code> / <code>+pr</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable profiling output</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, X11, XV, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> If profiling output is enabled, then VirtualGL will continuously benchmark itself and periodically print out the throughput of various stages in its image pipeline. <br /><br /> See Chapter <a href="#Perf_Measurement" class="ref">18</a> for more details. </dd> </dl> <p><a name="VGL_QUAL"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_QUAL = </code><em><code>{q}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-q </code><em><code>{q}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{q}</code></em> = the JPEG compression quality, 1 <= <em><code>{q}</code></em> <= 100</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL (JPEG), Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">95</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> In digital images, “frequency” refers to how quickly the color changes between light and dark as you move either horizontally or vertically in the image. Images with very sharp, bright features on a dark background, for instance, consist of both low-frequency and high-frequency components, whereas images with smooth transitions between neighboring pixels contain only low-frequency components. JPEG compression works by breaking down the image into its constituent frequencies and then throwing out the highest of these frequencies. The JPEG image “quality” determines which frequencies are thrown out. A JPEG quality of 1 throws out all but the lowest frequencies and thus produces a very impressionistic, but generally not very useful, compressed image. A JPEG quality of 100 retains all frequencies in the original image (but, due to roundoff errors, the compressed image is still not completely lossless.) <br /><br /> Because the human eye usually cannot detect the highest frequencies in the image, and often because the image lacks those high-frequency elements to begin with, a sufficiently high JPEG quality setting can produce a “perceptually lossless” image. A “perceptually lossless” image contains a small amount of mathematical error when compared to the original image, but this error is so small that, under normal circumstances, human vision cannot detect it. The threshold quality level at which JPEG compression becomes perceptually lossless is different for each image, but experiments with various visual difference benchmarks (such as <span class="remote"><a href="http://www.mpi-inf.mpg.de/resources/hdr/vdp/" class="remote">HDR-VDP</a></span><a name="idx0058"></a>) suggest that a JPEG quality of 95 is sufficient to guarantee perceptual losslessness for the types of applications (volume visualization apps, in particular) in which image quality is critical. As with any benchmarks, Your Mileage May Vary. Those who are particularly paranoid about image quality can set the JPEG quality to 100 or use RGB encoding, but a fast network is required for both. <br /> <div class="important"><p class="important"> If using an image transport plugin, then this setting need not necessarily correspond to JPEG image quality. The plugin can choose to respond to the <code>VGL_QUAL</code> option as it sees fit. </p></div> </dd> </dl> <p><a name="VGL_READBACK"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_READBACK = </code><em><code>none | pbo | sync</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Specify the method used by VirtualGL to read back the 3D pixels from the 3D graphics adapter</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">pbo</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> </dd> </dl> <ul class="Itemize"> <li class="Itemize-1 Itemize asterisk"> <em>none</em> = Do not read back the 3D pixels at all. On rare occasions, it might be desirable to have VirtualGL redirect OpenGL rendering from an application’s window into a Pbuffer but not automatically read back and send the rendered pixels. Some applications have their own mechanisms for reading back the rendered pixels, so setting <code>VGL_READBACK=none</code> disables VirtualGL’s readback mechanism and prevents duplication of effort. <br /><br /> This option was developed initially to support running <span class="remote"><a href="http://www.paraview.org/" class="remote">ParaView</a></span><a name="idx0059"></a> in parallel using MPI. ParaView/MPI normally uses MPI Processes 1 through N as rendering servers, each rendering a portion of the geometry into a separate window on a separate X display. ParaView reads back these server windows and composites the pixels into the main application window, which is controlled by MPI Process 0. By creating a script that passes a different value of <code>VGL_DISPLAY</code> and <code>VGL_READBACK</code> to each MPI process, it is possible to make all of the ParaView server processes render to off-screen buffers on different GPUs while preventing VirtualGL from displaying any pixels except those generated by Process 0. <br /><br /> This setting can also be used to force Chromium server processes to render into Pbuffers instead of windows. See Section <a href="#Force_Pbuffer" class="ref">13.2</a>. <br /><br /> </li> <li class="Itemize-1 Itemize asterisk"> <em>pbo</em> = PBO readback mode. Attempt to use pixel buffer objects (PBOs) to read back the 3D pixels from the GPU. A PBO is an opaque memory buffer managed by OpenGL, so it can be locked down for direct DMA transfers. This improves readback performance as well as makes the readback operation non-blocking. Because PBOs are managed buffers, VirtualGL has to perform an additional memory copy to transfer the pixels out of the PBO and into the image transport’s buffer. However, on high-end GPUs, PBO readback mode will still generally perform better than synchronous readback mode, even with this additional memory copy. Further, since the non-blocking nature of PBO readback reduces the load on the GPU, PBOs can improve performance dramatically when multiple simultaneous users are sharing a professional-grade GPU. <br /><br /> As of this writing, some nVidia GeForce adapters will fall back to using blocking readbacks if the pixel format requested in the pixel read operation does not match the pixel format of the Pbuffer. If VirtualGL detects that this is occurring– that is, if PBOs are no longer behaving asynchronously– then VGL will fall back to synchronous readback mode until the next time the compression type is changed. If you are using an X proxy, then this situation could occur because the X proxy’s virtual framebuffer is BGRA but, unless the application has requested an alpha channel, its Pbuffer is likely BGR. In this specific case, setting the <a href="#VGL_FORCEALPHA"><code>VGL_FORCEALPHA</code></a><a name="idx0060"></a> option to <code>1</code> could alleviate the issue. <br /><br /> </li> <li class="Itemize-1 Itemize asterisk"> <em>sync</em> = Synchronous readback mode. This disables the use of PBOs altogether, which causes VirtualGL to always use blocking pixel readback operations. <br /><br /> Setting <code>VGL_VERBOSE=1</code> will cause VirtualGL to print the current readback mode being used, as well as the pixel format requested by the readback operation and the pixel format of the Pbuffer. Additionally, a notification will be printed if VirtualGL falls back from PBO readback mode to synchronous readback mode. </li> </ul> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_REFRESHRATE = </code><em><code>{r}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{r}</code></em> = the “virtual” refresh rate, in Hz, for the GLX_EXT_swap_control and GLX_SGI_swap_control extensions</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">60.0</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> The GLX_EXT_swap_control and GLX_SGI_swap_control extensions allow applications to specify that buffer swaps should be synchronized with the refresh rate of the monitor. When one of these extensions is used, <code>glXSwapBuffers()</code> will not return until a specified number of refreshes (the “swap interval”) has occurred. Although refresh rate has no meaning when rendering into an off-screen buffer, VirtualGL still emulates the swap control extensions so that applications can control their own frame rate (this is often used by games, for instance, in which maintaining a constant frame rate is important.) VirtualGL uses an internal timer to emulate the refresh rate, and setting <code>VGL_REFRESHRATE</code> changes the interval of that timer. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_SAMPLES = </code><em><code>{s}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-ms </code><em><code>{s}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Force OpenGL multisampling to be enabled with <em><code>{s}</code></em> samples. <em><code>{s}</code></em> = 0 to force OpenGL multisampling to be disabled.</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Allow the 3D application to determine the level of multisampling</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> This option was added primarily because certain vendor-specific methods of enabling full-scene antialiasing at a global level (such as nVidia’s <code>__GL_FSAA_MODE</code> environment variable) do not work with Pbuffers and, subsequently, do not work with VirtualGL. If <code>VGL_SAMPLES</code> is > 0, then VirtualGL will attempt to create Pbuffers with the specified number (or a greater number) of samples. This effectively forces the 3D application to render with the specified multisampling level, as if the application had explicitly passed attributes of <code>GLX_SAMPLES</code>, <em><code>{s}</code></em> to <code>glXChooseVisual()</code>. If <code>VGL_SAMPLES</code> is <code>0</code>, then VirtualGL forces multisampling to be disabled, even if the 3D application explicitly tries to enable it. <div class="important"><p class="important"> <code>VGL_SAMPLES</code> overrides the application’s choice of visuals. It has no effect if the application is not explicitly choosing a visual. In that case, use <a href="#VGL_DEFAULTFBCONFIG"><code>VGL_DEFAULTFBCONFIG</code></a><a name="idx0061"></a> instead. </p></div> </dd> </dl> <p><a name="VGL_SPOIL"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_SPOIL = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-sp</code> / <code>+sp</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable frame spoiling</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, X11, XV, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Enabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> In remote display environments, the mouse movement is generally sampled at least 40 and sometimes 60 times per second. Therefore, unless VirtualGL is able to deliver at least that number of frames per second, the movement of a 3D scene will appear to lag behind the mouse motion. This is because the server is trying to render a new frame for every mouse motion event, but if the image transport and (if applicable) the client cannot process the frames fast enough, the server’s TCP buffers quickly fill up, causing increased delay in the delivery of each frame. VirtualGL’s default behavior is to compensate for this by dropping (spoiling) each frame that the transport isn’t ready to receive. This ensures that the movement of the 3D scene will appear to “keep up” with the mouse, even though not all rendered frames are actually being delivered. <br /><br /> Frame spoiling is usually necessary with interactive applications, but it should be turned off when running benchmarks or other non-interactive applications. Turning off frame spoiling will force every frame rendered on the server to be delivered through the image transport, and thus the frame rate reported by OpenGL benchmarks will accurately reflect the end-to-end performance of VirtualGL (though, in X proxy environments, this may still not accurately reflect the frame rate seen by the user. See Section <a href="#Frame_Spoiling" class="ref">18.2</a>.) Disabling frame spoiling also prevents non-interactive applications from wasting graphics resources by rendering frames that will never be seen. </dd> </dl> <p><a name="VGL_SPOILLAST"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_SPOILLAST = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable “spoil last” frame spoiling algorithm for frames triggered by <code>glFlush()</code></td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, X11, XV, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Enabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> VirtualGL will normally read back a rendered 3D image if the 3D application calls <code>glXSwapBuffers()</code> while rendering to the back buffer or if the 3D application calls <code>glFinish()</code>, <code>glFlush()</code>, or <code>glXWaitGL()</code> while rendering to the front buffer. When frame spoiling is enabled and the image transport is busy compressing/sending a frame, the newly-rendered frame is normally promoted to the head of the queue, and the rest of the frames in the queue are “spoiled” (discarded.) This algorithm, called “spoil first”, ensures that when a frame is actually delivered through the image transport (rather than spoiled), the delivered frame will be the most recently rendered frame. However, this algorithm requires that VirtualGL read back every frame that the application renders, even if the frame is ultimately discarded. <br /><br /> Some applications call <code>glFlush()</code> many thousands of times per frame while rendering to the front buffer. Thus, VirtualGL’s default behavior is to use a different spoiling algorithm, “spoil last”, to process frames triggered by <code>glFlush()</code> calls. “Spoil last” discards the most recently rendered frame if the image transport is busy. Thus, the only frames that are read back from the Pbuffer are the frames that are actually delivered through the image transport. However, there is no guarantee in this case that the delivered frame will be the most recently rendered frame, so applications that perform front buffer rendering and call <code>glFlush()</code> in response to an interactive operation may not display properly. For such applications, setting the <code>VGL_SPOILLAST</code> environment variable to <code>0</code> prior to launching the application with <code>vglrun</code> will cause the “spoil first” algorithm to be used for all frame triggers, including <code>glFlush()</code>. This should fix the display problem, at the expense of increased load on the GPU (because VirtualGL is now reading back the rendered 3D image every time <code>glFlush()</code> is called.) See <a href="#Application_Recipes">Application Recipes</a><a name="idx0062"></a> for a list of applications that are known to require this. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_SSL = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-s</code> / <code>+s</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable SSL encryption of the image transport</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Enabling this option causes the VGL Transport to be tunneled through a secure socket layer (SSL.) <div class="important"><p class="important"> This option has no effect unless both the VirtualGL server and client were built with OpenSSL support. </p></div> </dd> </dl> <p><a name="VGL_STEREO"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_STEREO = </code><em><code>left | right | quad | rc | gm | by | i | tb | ss</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-st </code><em><code>left | right | quad | rc | gm | by | i | tb | ss</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Specify the delivery method for stereo images</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">quad</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> <em>left</em> = When an application renders a stereo frame, send only the left eye buffer <br /><br /> <em>right</em> = When an application renders a stereo frame, send only the right eye buffer <br /><br /> <em>quad</em> = Attempt to use quad-buffered stereo, which will result in a pair of images being sent to the VirtualGL Client for every frame. Using quad-buffered stereo requires the VGL Transport (or a transport plugin that can handle stereo image pairs.) Using quad-buffered stereo with the VGL Transport also requires that the 2D X server support OpenGL and be connected to a GPU that supports stereo rendering. The 2D X server should additionally be configured to export stereo visuals. Quad-buffered stereo is not supported when using the VGL Transport with YUV encoding. If quad-buffered stereo is requested but the transport or the client does not support it, then VirtualGL will fall back to using anaglyphic stereo. <br /><br /> <em>rc</em> = Use Red/Cyan (anaglyphic) stereo, even if quad-buffered is available <br /><br /> <em>gm</em> = Use Green/Magenta (anaglyphic) stereo, even if quad-buffered is available <br /><br /> <em>by</em> = Use Blue/Yellow (anaglyphic) stereo, even if quad-buffered is available <br /><br /> <em>i</em> = Use Interleaved (passive) stereo, even if quad-buffered is available <br /><br /> <em>tb</em> = Use Top/Bottom (passive) stereo, even if quad-buffered is available <br /><br /> <em>ss</em> = Use Side-by-Side (passive) stereo, even if quad-buffered is available <br /><br /> See Chapter <a href="#Advanced_OpenGL" class="ref">17</a> for more details. </dd> </dl> <p><a name="VGL_SUBSAMP"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_SUBSAMP = </code><em><code>gray | 1x | 2x | 4x | 8x | 16x</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-samp </code><em><code>gray | 1x | 2x | 4x | 8x | 16x</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Specify the level of chrominance subsampling in the JPEG image compressor</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL (JPEG), Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">1x</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> When an image is compressed using JPEG, each pixel in the image is first converted from RGB (Red/Green/Blue) to YCbCr. An RGB pixel has three values that specify the amounts of red, green, and blue that make up the pixel’s color. A YCbCr pixel has three values that specify the overall brightness of the pixel (Y, or “luminance”) and the overall color of the pixel (Cb and Cr, or “chrominance”.) <div class="important"><p class="important"> NOTE: in the digital world, the terms “YCbCr” and “YUV” are often used interchangeably. Per the convention of the image processing and digital video communities, we use “YCbCr” when discussing JPEG compression and “YUV” when discussing video formats, but they are really the same thing. </p></div> Since the human eye is less sensitive to changes in color than it is to changes in brightness, the chrominance components for some of the pixels can be discarded without much noticeable loss in image quality. This technique, called “chrominance subsampling”, significantly reduces the size of the compressed image. <br /><br /> <em>1x</em> = no chrominance subsampling <br /><br /> <em>2x</em> = discard the chrominance components for every other pixel along the image’s X direction (this is also known as “4:2:2” or “2:1” subsampling.) All else being equal, 2x subsampling generally reduces the image size by about 20-25% when compared to no subsampling.<br /> <br /> <em>4x</em> = discard the chrominance components for every other pixel along both the X and Y directions of the image (this is also known as “4:2:0” or “2:2” subsampling.) All else being equal, 4x subsampling generally reduces the image size by about 35-40% when compared to no subsampling. <br /><br /> <em>8x</em> = discard the chrominance components for 3 out of every 4 pixels along the image’s X direction and half the pixels along the image’s Y direction (this is also known as “4:1:0” or “4:2” subsampling.) <em>This option is available only when using an image transport plugin that supports it.</em> <br /><br /> <em>16x</em> = discard the chrominance components for 3 out of every 4 pixels along both the X and Y directions of the image (this is also known as “4:4” subsampling.) <em>This option is available only when using an image transport plugin that supports it.</em> <br /><br /> <em>gray</em> = discard all chrominance components. This is useful when running applications (such as medical visualization applications) that are already generating grayscale images. <br /><br /> Subsampling artifacts are less noticeable with volume data, since it usually only contains 256 colors to begin with, but narrow, aliased lines and other sharp features on a black background will tend to produce very noticeable artifacts when subsampling is enabled. <br /><br /> The axis indicator from a popular visualization app displayed with 1x, 2x, and 4x chrominance subsampling (respectively):<br /> <img src="444.gif" alt="444" class="inline" id="imgid_6" name="imgid_6"/><img src="422.gif" alt="422" class="inline" id="imgid_7" name="imgid_7"/><img src="411.gif" alt="411" class="inline" id="imgid_8" name="imgid_8"/> <br /> <div class="important"><p class="important"> If using an image transport plugin, then this setting need not necessarily correspond to JPEG chrominance subsampling. How the plugin responds to the <code>VGL_SUBSAMP</code> option is implementation-specific. </p></div> </dd> </dl> <p><a name="VGL_SYNC"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_SYNC = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-sync</code> / <code>+sync</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable strict 2D/3D synchronization</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL, X11, XV, Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, VirtualGL’s operation is asynchronous from the point of view of the application. The application swaps the buffers or calls <code>glFinish()</code> or <code>glFlush()</code> or <code>glXWaitGL()</code>, and VirtualGL will read back the Pbuffer and deliver the pixels to the 2D X server … eventually. This is fine for the vast majority of applications, but it does not strictly conform to the GLX spec. Technically speaking, when an application calls <code>glXWaitGL()</code> or <code>glFinish()</code>, it is well within its rights to expect the 3D image to be immediately available in the X window. Fortunately, very few applications actually do expect this, but on rare occasions, an application may try to use <code>XGetImage()</code> or other X11 functions to obtain a bitmap of the pixels that were rendered by OpenGL. Enabling <code>VGL_SYNC</code> is a somewhat extreme measure that may be needed to make such applications display properly with VirtualGL. It was developed initially as a way to pass the GLX conformance suite (<code>conformx</code>, specifically), but at least one commercial application is known to require it as well (see <a href="#Application_Recipes">Application Recipes</a><a name="idx0063"></a>.)<br /><br />When <code>VGL_SYNC</code> is enabled, every call to <code>glFinish()</code>, <code>glXWaitGL()</code>, and <code>glXSwapBuffers()</code> will cause the contents of the Pbuffer to be read back and <em>synchronously</em> drawn into the application’s window <em>using the X11 Transport and no frame spoiling</em>. The call to <code>glFinish()</code>, <code>glXWaitGL()</code>, or <code>glXSwapBuffers()</code> will not return until VirtualGL has verified that the pixels have been delivered into the application’s window. As such, this mode can have potentially dire effects on performance when used with a remote 2D X server. It is strongly recommended that <code>VGL_SYNC</code> be used only in conjunction with an X proxy running on the same server as VirtualGL. <div class="important"><p class="important"> If an image transport plugin is being used, then VirtualGL does not automatically enable the X11 Transport or disable frame spoiling when <code>VGL_SYNC</code> is set. This allows the plugin to handle synchronous image delivery as it sees fit (or to simply ignore this option.) </p></div> </dd> </dl> <p><a name="VGL_TILESIZE"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_TILESIZE = </code><em><code>{t}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{t}</code></em> = the image tile size (<em><code>{t}</code></em> x <em><code>{t}</code></em> pixels) to use for multi-threaded compression and interframe comparison (8 <= <em><code>{t}</code></em> <= 1024)</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">VGL (JPEG, RGB), Custom (if supported)</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">256</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, the VGL Transport will divide an OpenGL window into equal-sized square tiles, compare each tile vs. the same tile in the previous frame, then compress and send only the tiles that have changed (assuming <a href="#VGL_INTERFRAME">interframe comparison</a><a name="idx0064"></a> is enabled.) The VGL Transport will also divide up the task of compressing these tiles among the available CPUs in a round robin fashion, if multi-threaded compression is enabled (see <a href="#VGL_NPROCS">VGL_NPROCS</a><a name="idx0065"></a>.) <br /><br /> There are several tradeoffs that must be considered when choosing a tile size: <br /><br /> <em>Parallel scalability:</em> <ul class="Itemize"><li class="Itemize-0"> Smaller tiles can more easily be divided up among multiple CPUs. If the tile size if too large, then there may not be enough tiles to go around. </li></ul> <br /> <em>Compression efficiency:</em> <ul class="Itemize"><li class="Itemize-0"> There is a certain amount of fixed CPU overhead required when compressing images, so compressing larger tiles makes more efficient use of the CPUs and increases the aggregate throughput of the compressor. </li></ul> <br /> <em>Inter-frame optimization:</em> <ul class="Itemize"><li class="Itemize-0"> When using smaller tiles, there is more of a chance that a given tile will remain unchanged from frame to frame and thus not need to be re-transmitted to the client. </li></ul> <br /> <em>Network efficiency:</em> <ul class="Itemize"><li class="Itemize-0"> Each tile has a certain amount of fixed overhead required to represent it on the network. Thus, using smaller tiles will increase the total number of tiles per frame, which will increase the total network usage. </li></ul> <br /> 256x256 was chosen as the default because, in experiments, it provided the best balance between scalability and efficiency on the platforms that VirtualGL supports. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_TRACE = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-tr</code> / <code>+tr</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable tracing</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> When tracing is enabled, VirtualGL will log all calls to the GLX and X11 functions it is intercepting, as well as the arguments, return values, and execution times for those functions. This is useful when diagnosing interaction problems between VirtualGL and a particular OpenGL application. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_TRANSPORT = </code><em><code>{t}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-trans </code><em><code>{t}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Use an image transport plugin</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">None</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> If this option is specified, then VirtualGL will attempt to load an image transport plugin contained in a dynamic library named <code>libtransvgl_</code><em><code>{t}</code></em><code>.so</code> located in the dynamic linker path. See Chapter <a href="#Transport_Plugins" class="ref">11</a> for more information. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_TRAPX11 = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable VirtualGL’s X11 error handler</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> If an application does not install its own X11 error handler, then the default X11 error handler is used, thus causing the application to exit if an X11 error occurs. Enabling the <code>VGL_TRAPX11</code> option will cause VirtualGL to install its own X11 error handler, which prints a warning message but allows the application to continue running. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_VERBOSE = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-v</code> / <code>+v</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable verbose VirtualGL messages</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> When verbose mode is enabled, VirtualGL will reveal some of the decisions it is making behind the scenes, such as which type of X11 drawing it is using, etc. This can be helpful when diagnosing performance problems. </dd> </dl> <p><a name="VGL_WM"></a></p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_WM = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglrun</code> argument</td> <td class="standard"><code>-wm</code> / <code>+wm</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable window manager mode</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> When window manager mode is enabled, VirtualGL will disable some of its internal features that interfere with the correct operation of 3D window managers such as compiz. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_X11LIB = </code><em><code>{l}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{l}</code></em> = the location of an alternate X11 library</td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Normally, VirtualGL will try to use the X11 library against which it was linked (usually <code>libX11.so.6</code>, in the system library path) to make any X11 calls it needs to make. Failing this, VirtualGL will then try to use the first compatible library named <code>libX11.so.6</code> that is found in the dynamic loader path. You can use the <code>VGL_X11LIB</code> environment variable to override this behavior and specify a dynamic library from which VirtualGL should call “real” X11 functions. <br /><br /> You shouldn’t need to muck with this unless something doesn’t work. However, it is potentially useful if one wishes to insert another X11 interposer between VirtualGL and the system’s X11 library. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_XVENDOR = </code><em><code>{v}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{v}</code></em> = a fake X11 vendor string to return when the application calls <code>XServerVendor()</code> or <code>ServerVendor()</code></td> </tr> <tr class="standard"> <td class="high standard">Image Transports</td> <td class="standard">All</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> Some applications expect the X11 vendor string to contain a particular value, which the application (sometimes erroneously) uses to figure out whether it is being displayed to a local or a remote X server. This setting allows you to fool such applications into thinking that they are being displayed to a “local” X server rather than a remote one. </dd> </dl> <h2 id="hd0020002">20.2 Client Settings</h2> <p>These settings control the VirtualGL Client, which is used only with the VGL Transport. <code>vglclient</code> is normally launched automatically from <code>vglconnect</code> and should not require any further configuration except in exotic circumstances. These settings are meant only for advanced users or those wishing to build additional infrastructure around VirtualGL.</p> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGLCLIENT_DRAWMODE = </code><em><code>ogl | x11</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglclient</code> argument</td> <td class="standard"><code>-gl</code> / <code>-x</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Specify the method used to draw pixels into the application window</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard"><code>x11</code></td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> If the client machine has a GPU, then it may be faster in some rare instances to draw pixels using OpenGL rather than using 2D (X11) commands. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGLCLIENT_LISTEN = </code><em><code>sslonly | nossl</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglclient</code> argument</td> <td class="standard"><code>-sslonly</code> / <code>-nossl</code></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Accept only unencrypted or only SSL connections from the VirtualGL server</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Accept both SSL and unencrypted connections</td> </tr> </table> </div> <div class="important"><p class="important"> This option is available only if the VirtualGL client was built with OpenSSL support. </p></div> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGLCLIENT_PORT = </code><em><code>{p}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglclient</code> argument</td> <td class="standard"><code>-port </code><em><code>{p}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{p}</code></em> = TCP port on which to listen for unencrypted connections from the VirtualGL server</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Automatically select a free port</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> The default behavior for <code>vglclient</code> is to first try listening for unencrypted connections on port 4242, to maintain backward compatibility with VirtualGL v2.0.x. If port 4242 is not available, then <code>vglclient</code> will try to find a free port in the range of 4200-4299. If none of those ports is available, then <code>vglclient</code> will request a free port from the operating system. <br /><br /> Setting this option circumvents the automatic behavior described above and causes <code>vglclient</code> to listen only on the specified TCP port. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_PROFILE = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable profiling output</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> If profiling output is enabled, then VirtualGL will continuously benchmark itself and periodically print out the throughput of various stages in its image pipelines. <br /><br /> See Chapter <a href="#Perf_Measurement" class="ref">18</a> for more details. </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGLCLIENT_SSLPORT = </code><em><code>{p}</code></em></td> </tr> <tr class="standard"> <td class="high standard"><code>vglclient</code> argument</td> <td class="standard"><code>-sslport </code><em><code>{p}</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard"><em><code>{p}</code></em> = TCP port on which to listen for SSL connections from the VirtualGL server</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Automatically select a free port</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> The default behavior for <code>vglclient</code> is to first try listening for SSL connections on port 4243, to maintain backward compatibility with VirtualGL v2.0.x. If port 4243 is not available, then <code>vglclient</code> will try to find a free port in the range of 4200-4299. If none of those ports is available, then <code>vglclient</code> will request a free port from the operating system. <br /><br /> Setting this option circumvents the automatic behavior described above and causes <code>vglclient</code> to listen only on the specified TCP port. <div class="important"><p class="important"> This option is available only if the VirtualGL client was built with OpenSSL support. </p></div> </dd> </dl> <div class="table"> <table class="standard"> <tr class="standard"> <td class="high standard">Environment Variable</td> <td class="standard"><code>VGL_VERBOSE = </code><em><code>0 | 1</code></em></td> </tr> <tr class="standard"> <td class="high standard">Summary</td> <td class="standard">Disable/enable verbose VirtualGL messages</td> </tr> <tr class="standard"> <td class="high standard">Default Value</td> <td class="standard">Disabled</td> </tr> </table> </div> <dl class="Description"> <dt class="Description-1 Description">Description</dt> <dd class="Description-1 Description"> When verbose mode is enabled, VirtualGL will reveal some of the decisions it is making behind the scenes, such as which type of X11 drawing it is using, etc. This can be helpful when diagnosing performance problems. </dd> </dl> <p><br /></p> </body> </html>