<!DOCTYPE html
PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en-us" xml:lang="en-us">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"></meta>
<meta http-equiv="X-UA-Compatible" content="IE=edge"></meta>
<meta name="copyright" content="(C) Copyright 2005"></meta>
<meta name="DC.rights.owner" content="(C) Copyright 2005"></meta>
<meta name="DC.Type" content="cuda_reference"></meta>
<meta name="DC.Title" content="API synchronization behavior"></meta>
<meta name="DC.Format" content="XHTML"></meta>
<meta name="DC.Identifier" content="api-sync-behavior"></meta>
<link rel="stylesheet" type="text/css" href="../common/formatting/commonltr.css"></link>
<link rel="stylesheet" type="text/css" href="../common/formatting/site.css"></link>
<title>CUDA Driver API :: CUDA Toolkit Documentation</title>
<!--[if lt IE 9]>
<script src="../common/formatting/html5shiv-printshiv.min.js"></script>
<![endif]-->
<script type="text/javascript" charset="utf-8" src="//assets.adobedtm.com/b92787824f2e0e9b68dc2e993f9bd995339fe417/satelliteLib-7ba51e58dc61bcb0e9311aadd02a0108ab24cc6c.js"></script>
<script type="text/javascript" charset="utf-8" src="../common/formatting/jquery.min.js"></script>
<script type="text/javascript" charset="utf-8" src="../common/formatting/jquery.ba-hashchange.min.js"></script>
<script type="text/javascript" charset="utf-8" src="../common/formatting/jquery.scrollintoview.min.js"></script>
<script type="text/javascript" src="../search/htmlFileList.js"></script>
<script type="text/javascript" src="../search/htmlFileInfoList.js"></script>
<script type="text/javascript" src="../search/nwSearchFnt.min.js"></script>
<script type="text/javascript" src="../search/stemmers/en_stemmer.min.js"></script>
<script type="text/javascript" src="../search/index-1.js"></script>
<script type="text/javascript" src="../search/index-2.js"></script>
<script type="text/javascript" src="../search/index-3.js"></script>
<link rel="canonical" href="http://docs.nvidia.com/cuda/cuda-driver-api/index.html"></link>
<link rel="stylesheet" type="text/css" href="../common/formatting/qwcode.highlight.css"></link>
</head>
<body>
<header id="header"><span id="company">NVIDIA</span><span id="site-title">CUDA Toolkit Documentation</span><form id="search" method="get" action="search">
<input type="text" name="search-text"></input><fieldset id="search-location">
<legend>Search In:</legend>
<label><input type="radio" name="search-type" value="site"></input>Entire Site</label>
<label><input type="radio" name="search-type" value="document"></input>Just This Document</label></fieldset>
<button type="reset">clear search</button>
<button id="submit" type="submit">search</button></form>
</header>
<div id="site-content">
<nav id="site-nav">
<div class="category closed"><a href="../index.html" title="The root of the site.">CUDA Toolkit
v10.1.168</a></div>
<div class="category"><a href="index.html" title="CUDA Driver API">CUDA Driver API</a></div>
<ul>
<li>
<div class="section-link"><a href="driver-vs-runtime-api.html#driver-vs-runtime-api">1. Difference between the driver and runtime APIs </a></div>
</li>
<li>
<div class="section-link"><a href="api-sync-behavior.html#api-sync-behavior">2. API synchronization behavior </a></div>
</li>
<li>
<div class="section-link"><a href="stream-sync-behavior.html#stream-sync-behavior">3. Stream synchronization behavior</a></div>
</li>
<li>
<div class="section-link"><a href="graphs-thread-safety.html#graphs-thread-safety">4. Graph object thread safety</a></div>
</li>
<li>
<div class="section-link"><a href="modules.html#modules">5. Modules</a></div>
<ul>
<li>
<div class="section-link"><a href="group__CUDA__TYPES.html#group__CUDA__TYPES">5.1. Data types used by CUDA driver</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__ERROR.html#group__CUDA__ERROR">5.2. Error Handling</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__INITIALIZE.html#group__CUDA__INITIALIZE">5.3. Initialization</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__VERSION.html#group__CUDA__VERSION">5.4. Version Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__DEVICE.html#group__CUDA__DEVICE">5.5. Device Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__DEVICE__DEPRECATED.html#group__CUDA__DEVICE__DEPRECATED">5.6. Device Management [DEPRECATED]</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__PRIMARY__CTX.html#group__CUDA__PRIMARY__CTX">5.7. Primary Context Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__CTX.html#group__CUDA__CTX">5.8. Context Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__CTX__DEPRECATED.html#group__CUDA__CTX__DEPRECATED">5.9. Context Management [DEPRECATED]</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__MODULE.html#group__CUDA__MODULE">5.10. Module Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__MEM.html#group__CUDA__MEM">5.11. Memory Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__UNIFIED.html#group__CUDA__UNIFIED">5.12. Unified Addressing</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__STREAM.html#group__CUDA__STREAM">5.13. Stream Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__EVENT.html#group__CUDA__EVENT">5.14. Event Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__EXTRES__INTEROP.html#group__CUDA__EXTRES__INTEROP">5.15. External Resource Interoperability</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__MEMOP.html#group__CUDA__MEMOP">5.16. Stream memory operations</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__EXEC.html#group__CUDA__EXEC">5.17. Execution Control</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__EXEC__DEPRECATED.html#group__CUDA__EXEC__DEPRECATED">5.18. Execution Control [DEPRECATED]</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__GRAPH.html#group__CUDA__GRAPH">5.19. Graph Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__OCCUPANCY.html#group__CUDA__OCCUPANCY">5.20. Occupancy</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__TEXREF__DEPRECATED.html#group__CUDA__TEXREF__DEPRECATED">5.21. Texture Reference Management [DEPRECATED]</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__SURFREF__DEPRECATED.html#group__CUDA__SURFREF__DEPRECATED">5.22. Surface Reference Management [DEPRECATED]</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__TEXOBJECT.html#group__CUDA__TEXOBJECT">5.23. Texture Object Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__SURFOBJECT.html#group__CUDA__SURFOBJECT">5.24. Surface Object Management</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__PEER__ACCESS.html#group__CUDA__PEER__ACCESS">5.25. Peer Context Memory Access</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__GRAPHICS.html#group__CUDA__GRAPHICS">5.26. Graphics Interoperability</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__PROFILER.html#group__CUDA__PROFILER">5.27. Profiler Control</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__GL.html#group__CUDA__GL">5.28. OpenGL Interoperability</a></div>
<ul>
<li>
<div class="section-link"><a href="group__CUDA__GL__DEPRECATED.html#group__CUDA__GL__DEPRECATED">5.28.1. OpenGL Interoperability [DEPRECATED]</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="group__CUDA__D3D9.html#group__CUDA__D3D9">5.29. Direct3D 9 Interoperability</a></div>
<ul>
<li>
<div class="section-link"><a href="group__CUDA__D3D9__DEPRECATED.html#group__CUDA__D3D9__DEPRECATED">5.29.1. Direct3D 9 Interoperability [DEPRECATED]</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="group__CUDA__D3D10.html#group__CUDA__D3D10">5.30. Direct3D 10 Interoperability</a></div>
<ul>
<li>
<div class="section-link"><a href="group__CUDA__D3D10__DEPRECATED.html#group__CUDA__D3D10__DEPRECATED">5.30.1. Direct3D 10 Interoperability [DEPRECATED]</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="group__CUDA__D3D11.html#group__CUDA__D3D11">5.31. Direct3D 11 Interoperability</a></div>
<ul>
<li>
<div class="section-link"><a href="group__CUDA__D3D11__DEPRECATED.html#group__CUDA__D3D11__DEPRECATED">5.31.1. Direct3D 11 Interoperability [DEPRECATED]</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="group__CUDA__VDPAU.html#group__CUDA__VDPAU">5.32. VDPAU Interoperability</a></div>
</li>
<li>
<div class="section-link"><a href="group__CUDA__EGL.html#group__CUDA__EGL">5.33. EGL Interoperability</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="annotated.html#annotated">6. Data Structures</a></div>
<ul>
<li>
<div class="section-link"><a href="structCUDA__ARRAY3D__DESCRIPTOR.html#structCUDA__ARRAY3D__DESCRIPTOR">6.1. CUDA_ARRAY3D_DESCRIPTOR</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__ARRAY__DESCRIPTOR.html#structCUDA__ARRAY__DESCRIPTOR">6.2. CUDA_ARRAY_DESCRIPTOR</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__EXTERNAL__MEMORY__BUFFER__DESC.html#structCUDA__EXTERNAL__MEMORY__BUFFER__DESC">6.3. CUDA_EXTERNAL_MEMORY_BUFFER_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__EXTERNAL__MEMORY__HANDLE__DESC.html#structCUDA__EXTERNAL__MEMORY__HANDLE__DESC">6.4. CUDA_EXTERNAL_MEMORY_HANDLE_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__EXTERNAL__MEMORY__MIPMAPPED__ARRAY__DESC.html#structCUDA__EXTERNAL__MEMORY__MIPMAPPED__ARRAY__DESC">6.5. CUDA_EXTERNAL_MEMORY_MIPMAPPED_ARRAY_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__EXTERNAL__SEMAPHORE__HANDLE__DESC.html#structCUDA__EXTERNAL__SEMAPHORE__HANDLE__DESC">6.6. CUDA_EXTERNAL_SEMAPHORE_HANDLE_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__EXTERNAL__SEMAPHORE__SIGNAL__PARAMS.html#structCUDA__EXTERNAL__SEMAPHORE__SIGNAL__PARAMS">6.7. CUDA_EXTERNAL_SEMAPHORE_SIGNAL_PARAMS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__EXTERNAL__SEMAPHORE__WAIT__PARAMS.html#structCUDA__EXTERNAL__SEMAPHORE__WAIT__PARAMS">6.8. CUDA_EXTERNAL_SEMAPHORE_WAIT_PARAMS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__HOST__NODE__PARAMS.html#structCUDA__HOST__NODE__PARAMS">6.9. CUDA_HOST_NODE_PARAMS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__KERNEL__NODE__PARAMS.html#structCUDA__KERNEL__NODE__PARAMS">6.10. CUDA_KERNEL_NODE_PARAMS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__LAUNCH__PARAMS.html#structCUDA__LAUNCH__PARAMS">6.11. CUDA_LAUNCH_PARAMS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__MEMCPY2D.html#structCUDA__MEMCPY2D">6.12. CUDA_MEMCPY2D</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__MEMCPY3D.html#structCUDA__MEMCPY3D">6.13. CUDA_MEMCPY3D</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__MEMCPY3D__PEER.html#structCUDA__MEMCPY3D__PEER">6.14. CUDA_MEMCPY3D_PEER</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__MEMSET__NODE__PARAMS.html#structCUDA__MEMSET__NODE__PARAMS">6.15. CUDA_MEMSET_NODE_PARAMS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__POINTER__ATTRIBUTE__P2P__TOKENS.html#structCUDA__POINTER__ATTRIBUTE__P2P__TOKENS">6.16. CUDA_POINTER_ATTRIBUTE_P2P_TOKENS</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__RESOURCE__DESC.html#structCUDA__RESOURCE__DESC">6.17. CUDA_RESOURCE_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__RESOURCE__VIEW__DESC.html#structCUDA__RESOURCE__VIEW__DESC">6.18. CUDA_RESOURCE_VIEW_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUDA__TEXTURE__DESC.html#structCUDA__TEXTURE__DESC">6.19. CUDA_TEXTURE_DESC</a></div>
</li>
<li>
<div class="section-link"><a href="structCUdevprop.html#structCUdevprop">6.20. CUdevprop</a></div>
</li>
<li>
<div class="section-link"><a href="structCUeglFrame.html#structCUeglFrame">6.21. CUeglFrame</a></div>
</li>
<li>
<div class="section-link"><a href="structCUipcEventHandle.html#structCUipcEventHandle">6.22. CUipcEventHandle</a></div>
</li>
<li>
<div class="section-link"><a href="structCUipcMemHandle.html#structCUipcMemHandle">6.23. CUipcMemHandle</a></div>
</li>
<li>
<div class="section-link"><a href="unionCUstreamBatchMemOpParams.html#unionCUstreamBatchMemOpParams">6.24. CUstreamBatchMemOpParams</a></div>
</li>
</ul>
</li>
<li>
<div class="section-link"><a href="functions.html#functions">7. Data Fields</a></div>
</li>
<li>
<div class="section-link"><a href="deprecated.html#deprecated">8. Deprecated List</a></div>
</li>
<li>
<div class="section-link"><a href="notices-header.html#notices-header">Notices</a></div>
<ul></ul>
</li>
</ul>
</nav>
<div id="resize-nav"></div>
<nav id="search-results">
<h2>Search Results</h2>
<ol></ol>
</nav>
<div id="contents-container">
<div id="breadcrumbs-container">
<div id="breadcrumbs"><a href="driver-vs-runtime-api.html" shape="rect">< Previous</a> | <a href="stream-sync-behavior.html" shape="rect">Next ></a></div>
<div id="release-info">CUDA Driver API
(<a href="../../pdf/CUDA_Driver_API.pdf">PDF</a>)
-
v10.1.168
(<a href="https://developer.nvidia.com/cuda-toolkit-archive">older</a>)
-
Last updated April 24, 2019
-
<a href="mailto:CUDAIssues@nvidia.com?subject=CUDA Toolkit Documentation Feedback: CUDA Driver API">Send Feedback</a></div>
</div>
<article id="contents">
<div class="topic nested1" id="api-sync-behavior"><a name="api-sync-behavior" shape="rect">
<!-- --></a><h2 class="topictitle2">2. API synchronization behavior </h2>
<div class="body refbody">
<div class="section">
<p class="p">The API provides memcpy/memset functions in both synchronous and asynchronous forms, the latter having an "Async"
suffix. This is a misnomer as each function may exhibit synchronous or asynchronous behavior depending on the
arguments passed to the function.
</p>
</div>
<div class="section" id="api-sync-behavior__memcpy"><a name="api-sync-behavior__memcpy" shape="rect">
<!-- --></a><h2 class="title sectiontitle">Memcpy</h2>
<p class="p">In the reference documentation, each memcpy function is categorized as synchronous or asynchronous, corresponding to the definitions
below.
</p>
<p class="p apiDesc_subtitle" id="api-sync-behavior__memcpy-sync"><a name="api-sync-behavior__memcpy-sync" shape="rect">
<!-- --></a><strong class="ph b">Synchronous</strong></p>
<div class="p">
<ol class="ol">
<li class="li">
<p class="p">All transfers involving Unified Memory regions are fully synchronous with respect to the host.</p>
</li>
<li class="li">
<p class="p">For transfers from pageable host memory to device memory, a stream sync is performed before the copy is initiated. The function
will return once the pageable buffer has been copied to the staging memory for DMA transfer to device memory, but the DMA
to final destination may not have completed.
</p>
</li>
<li class="li">
<p class="p">For transfers from pinned host memory to device memory, the function is synchronous with respect to the host.</p>
</li>
<li class="li">
<p class="p">For transfers from device to either pageable or pinned host memory, the function returns only once the copy has completed.</p>
</li>
<li class="li">
<p class="p">For transfers from device memory to device memory, no host-side synchronization is performed.</p>
</li>
<li class="li">
<p class="p">For transfers from any host memory to any host memory, the function is fully synchronous with respect to the host. </p>
</li>
</ol>
</div>
<p class="p apiDesc_subtitle" id="api-sync-behavior__memcpy-async"><a name="api-sync-behavior__memcpy-async" shape="rect">
<!-- --></a><strong class="ph b">Asynchronous</strong></p>
<div class="p">
<ol class="ol">
<li class="li">
<p class="p">For transfers from device memory to pageable host memory, the function will return only once the copy has completed.</p>
</li>
<li class="li">
<p class="p">For transfers from any host memory to any host memory, the function is fully synchronous with respect to the host.</p>
</li>
<li class="li">
<p class="p">For all other transfers, the function is fully asynchronous. If pageable memory must first be staged to pinned memory, this
will be handled asynchronously with a worker thread.
</p>
</li>
</ol>
</div>
</div>
<div class="section" id="api-sync-behavior__memset"><a name="api-sync-behavior__memset" shape="rect">
<!-- --></a><h2 class="title sectiontitle">Memset</h2>
<p class="p">The synchronous memset functions are asynchronous with respect to the host except when the target is pinned host memory or
a Unified Memory region, in which case they are fully synchronous. The Async versions are always asynchronous with respect
to the host.
</p>
</div>
<div class="section" id="api-sync-behavior__kernel-launch-details"><a name="api-sync-behavior__kernel-launch-details" shape="rect">
<!-- --></a><h2 class="title sectiontitle">Kernel Launches</h2>
<p class="p">Kernel launches are asynchronous with respect to the host. Details of concurrent kernel execution and data transfers can be
found in the CUDA Programmers Guide.
</p>
</div>
</div>
</div>
<hr id="contents-end"></hr>
</article>
</div>
</div>
<script language="JavaScript" type="text/javascript" charset="utf-8" src="../common/formatting/common.min.js"></script>
<script language="JavaScript" type="text/javascript" charset="utf-8" src="../common/scripts/google-analytics/google-analytics-write.js"></script>
<script language="JavaScript" type="text/javascript" charset="utf-8" src="../common/scripts/google-analytics/google-analytics-tracker.js"></script>
<script type="text/javascript">var switchTo5x=true;</script><script type="text/javascript" src="http://w.sharethis.com/button/buttons.js"></script><script type="text/javascript">stLight.options({publisher: "998dc202-a267-4d8e-bce9-14debadb8d92", doNotHash: false, doNotCopy: false, hashAddressBar: false});</script><script type="text/javascript">_satellite.pageBottom();</script></body>
</html>
distrib
>
Mageia
>
7
>
x86_64
>
media
>
nonfree-updates
>
by-pkgid
>
b86a85131cc739c1c53d0b55840a4328
>
files
>
1366
