Sophie

Sophie

distrib > Mageia > 5 > x86_64 > media > nonfree-release-src > by-pkgid > b13c3f32910e50a883ce135d893dcf5c > files > 5

nvidia-cuda-toolkit-6.5.14-6.mga5.nonfree.src.rpm

%define		__jar_repack %nil
%define		driver_ver 340.29

%define		_enable_debug_packages %{nil}
%define		debug_package %{nil}

Summary:	NVIDIA CUDA runtime libraries
Name:		nvidia-cuda-toolkit
Version:	6.5.14
Release:	%mkrel 6
Source0:	http://developer.download.nvidia.com/compute/cuda/5_5/rel/installers/cuda_%{version}_linux_32.run
Source1:	http://developer.download.nvidia.com/compute/cuda/5_5/rel/installers/cuda_%{version}_linux_64.run
Source2:	nvidia
Source10:	nvvp.desktop
Source11:	nsight.desktop
License:	Freeware
Group:		Development/Other
Url:		http://www.nvidia.com/cuda/
Recommends:	nvidia >= %{driver_ver}
BuildRequires:	imagemagick
BuildRequires:	jpackage-utils
%ifnarch x86_64
Obsoletes:	nvidia-cuda-toolkit-samples < 6.5.14
Obsoletes:	nvidia-visual-profiler < 6.5.14
Obsoletes:	nvidia-nsight < 6.5.14
%endif

# We don't require installation of the NVIDIA graphics drivers so that 
# folks can do CUDA development on systems without NVIDIA hardware.

# A library, libcudainj.so, was introduced in CUDA 4.1, which depends
# on libcuda.so. It is not needed to compile CUDA programs, though.
# python(abi) auto-require is triggered by runant.py script somewhere
# inside Eclipse plugins used by NVVP.
%global __requires_exclude libcuda.so.1|devel\\(libcuda\\)|devel\\(libcuda\\(64bit\\)\\)|python\\(abi\\)
%global __provides_exclude libcairo.so.2

%description
NVIDIA® CUDA™ is a general purpose parallel computing architecture
that leverages the parallel compute engine in NVIDIA graphics
processing units (GPUs) to solve many complex computational problems
in a fraction of the time required on a CPU. It includes the CUDA
Instruction Set Architecture (ISA) and the parallel compute engine in
the GPU. To program to the CUDA™ architecture, developers can, today,
use C++, one of the most widely used high-level programming languages,
which can then be run at great performance on a CUDA™ enabled
processor. Support for other languages, like FORTRAN, Python or Java,
is available from third parties.

This package contains the libraries and attendant files needed to run
programs that make use of CUDA.

%package devel
Summary:	NVIDIA CUDA Toolkit development files
Group:		Development/Other
Requires:	%{name} = %{version}-%{release}
Recommends:	nvidia-devel >= %{driver_ver}
Recommends:	gcc-c++

%description devel
NVIDIA® CUDA™ is a general purpose parallel computing architecture
that leverages the parallel compute engine in NVIDIA graphics
processing units (GPUs) to solve many complex computational problems
in a fraction of the time required on a CPU. It includes the CUDA
Instruction Set Architecture (ISA) and the parallel compute engine in
the GPU. To program to the CUDA™ architecture, developers can, today,
use C++, one of the most widely used high-level programming languages,
which can then be run at great performance on a CUDA™ enabled
processor. Support for other languages, like FORTRAN, Python or Java,
is available from third parties.

This package contains the development files needed to build programs
that make use of CUDA.

%ifarch x86_64
%package samples
Summary:	NVIDIA CUDA Toolkit samples
Recommends:	%{name}-devel = %{version}-%{release}

%description samples
NVIDIA® CUDA™ is a general purpose parallel computing architecture
that leverages the parallel compute engine in NVIDIA graphics
processing units (GPUs) to solve many complex computational problems
in a fraction of the time required on a CPU. It includes the CUDA
Instruction Set Architecture (ISA) and the parallel compute engine in
the GPU. To program to the CUDA™ architecture, developers can, today,
use C++, one of the most widely used high-level programming languages,
which can then be run at great performance on a CUDA™ enabled
processor. Support for other languages, like FORTRAN, Python or Java,
is available from third parties.

This package contains numerous CUDA code samples (formerly CUDA SDK).

%package -n nvidia-visual-profiler
Summary:	NVIDIA Visual Profiler
Group:		Development/Other
Requires:	java
Obsoletes:	nvidia-cuda-profiler < 4.0, nvidia-opencl-profiler < 4.0, nvidia-compute-profiler < 5.0
Recommends:	nvidia-devel >= %{driver_ver}
Recommends:	%{name} = %{version}-%{release}

# We don't strictly require NVIDIA CUDA Toolkit, because the profiler
# could be used to analyze CSV profile logs obtained elsewhere.

%description -n nvidia-visual-profiler
NVIDIA® CUDA™ is a general purpose parallel computing architecture
that leverages the parallel compute engine in NVIDIA graphics
processing units (GPUs) to solve many complex computational problems
in a fraction of the time required on a CPU. It includes the CUDA
Instruction Set Architecture (ISA) and the parallel compute engine in
the GPU. To program to the CUDA™ architecture, developers can, today,
use C++, one of the most widely used high-level programming languages,
which can then be run at great performance on a CUDA™ enabled
processor. Support for other languages, like FORTRAN, Python or Java,
is available from third parties.

This package contains NVIDIA Visual Profiler for CUDA and OpenCL.

%package -n nvidia-nsight
Summary:	NVIDIA Nsight IDE
Group:		Development/Other
Requires:	java
Recommends:	nvidia-devel >= %{driver_ver}

# We don't strictly require NVIDIA CUDA Toolkit, because Nsight IDE
# could be used to develop CUDA programs on a remote node.

%description -n nvidia-nsight
NVIDIA® CUDA™ is a general purpose parallel computing architecture
that leverages the parallel compute engine in NVIDIA graphics
processing units (GPUs) to solve many complex computational problems
in a fraction of the time required on a CPU. It includes the CUDA
Instruction Set Architecture (ISA) and the parallel compute engine in
the GPU. To program to the CUDA™ architecture, developers can, today,
use C++, one of the most widely used high-level programming languages,
which can then be run at great performance on a CUDA™ enabled
processor. Support for other languages, like FORTRAN, Python or Java,
is available from third parties.

This package contains Nsight Eclipse Edition, a full-featured CUDA IDE.
%endif

%prep
%setup -q -T -c %{name}-%{version}

%build
# Nothing to do

%install
%__install -d -m 755 %{buildroot}%{_usr}
%__install -d -m 755 %{buildroot}%{_datadir}/%{name}
%__install -d -m755 %{buildroot}%{_datadir}/applications
%__install -d -m755 %{buildroot}%{_docdir}/%{name}-devel

%ifarch %ix86
bash %SOURCE0 --tar xf -C .
./run_files/cuda-linux-rel-%{version}-18745345.run --tar xf -C %{buildroot}%{_usr}
%else
bash %SOURCE1 --tar xf -C .
./run_files/cuda-linux64-rel-%{version}-18749181.run --tar xf -C %{buildroot}%{_usr}
%__rm -rf %{buildroot}%{_usr}/lib
%__rm -rf %{buildroot}%{_usr}/extras/CUPTI/lib
sed -i 's/lib/lib64/g' %{buildroot}%{_bindir}/nvcc.profile
# (tmb) restore libdevice
sed -i 's/lib64device/libdevice/g' %{buildroot}%{_bindir}/nvcc.profile
%endif

# Move compiler components from /usr/{nvvm,open64} to libdir
%__mv %{buildroot}%{_usr}/nvvm %{buildroot}%{_libdir}/
%__mv %{buildroot}%{_usr}/open64 %{buildroot}%{_libdir}/

# Fix nvcc.profile to reflect the move
sed -i 's/nvvm/%{_lib}\/nvvm/g' %{buildroot}%{_bindir}/nvcc.profile
sed -i 's/open64/%{_lib}\/open64/g' %{buildroot}%{_bindir}/nvcc.profile

# Unpack samples (SDK) 64bit only
%ifarch x86_64
./run_files/cuda-samples-linux-%{version}-18745345.run --tar xf -C %{buildroot}%{_datadir}/%{name}
%endif

# Remove duplicates (these will be shipped with -devel)
%__rm -rf %{buildroot}%{_datadir}/%{name}/sdk/{doc,tools}
%__rm -rf %{buildroot}%{_datadir}/%{name}/sdk/Documentation.html
%__rm -rf %{buildroot}%{_usr}/InstallUtils.pm

%__rm -f  %{buildroot}%{_datadir}/%{name}/install-sdk-linux.pl
%__rm -rf %{buildroot}%{_usr}/install-linux.pl
%__rm -rf %{buildroot}%{_usr}/uninstall_cuda.pl

%ifarch x86_64
%__mv %{buildroot}%{_usr}/doc/* %{buildroot}%{_docdir}/%{name}-devel/
%__rm -r %{buildroot}%{_usr}/doc
%__mv %{buildroot}%{_usr}/src %{buildroot}%{_datadir}/%{name}
%__mv %{buildroot}%{_usr}/libnvvp %{buildroot}%{_libdir}/nvvp
%__mv %{buildroot}%{_usr}/libnsight %{buildroot}%{_libdir}/nsight
%endif
%__mv %{buildroot}%{_usr}/EULA.txt %{buildroot}%{_docdir}/%{name}-devel/
%__mv %{buildroot}%{_usr}/CUDA_Toolkit_Release_Notes.txt %{buildroot}%{_docdir}/%{name}-devel/
%__mv %{buildroot}%{_usr}/{extras,tools} %{buildroot}%{_datadir}/%{name}

# dont ship gdb files
%__rm -rf %{buildroot}%_datadir/gdb

# Remove bundled JRE and fix paths for Java
%__rm -rf %{buildroot}%{_usr}/jre
%ifarch x86_64
sed -i 's|\.\./jre/bin/java|%{_jvmdir}/jre/bin/java|g' %{buildroot}%{_libdir}/nvvp/nvvp.ini
sed -i 's|\.\./jre/bin/java|%{_jvmdir}/jre/bin/java|g' %{buildroot}%{_libdir}/nsight/nsight.ini


for S in 16 24 32 48 64 128 192 256; do
 %__install -d -m755 %{buildroot}%{_iconsdir}/hicolor/$S\x$S/apps
 convert -scale $S\x$S %{buildroot}/%{_libdir}/nvvp/icon.xpm %{buildroot}%{_iconsdir}/hicolor/$S\x$S/apps/nvvp.png
 convert -scale $S\x$S %{buildroot}/%{_libdir}/nsight/icon.xpm %{buildroot}%{_iconsdir}/hicolor/$S\x$S/apps/nsight.png
done

%__install -m644 %{SOURCE10} %{buildroot}%{_datadir}/applications/
%__install -m644 %{SOURCE11} %{buildroot}%{_datadir}/applications/
%endif
%__install -D -m 755 %SOURCE2 %{buildroot}%{_sysconfdir}/init.d/nvidia

# Don't prevent the use of gcc 4.9
sed -i 's|__GNUC_MINOR__ > 8|__GNUC_MINOR__ > 9|g' %{buildroot}%{_includedir}/host_config.h

%files
%_libdir/*.so.*
%_sysconfdir/init.d/*

%files devel
%doc %{_docdir}/%{name}-devel/*
%_bindir/*
%ifarch x86_64
%exclude %_bindir/nvvp
%exclude %_bindir/nsight
%endif
%_libdir/*.so
%_libdir/*.a
%_includedir/*
%_libdir/nvvm/*
%_libdir/open64/*
%_libdir/stubs/*
%_datadir/%{name}/*
%ifarch x86_64
%exclude %_datadir/%{name}/cuda-samples

%files samples
%_datadir/%{name}/cuda-samples

%files -n nvidia-visual-profiler
%_bindir/nvvp
%_libdir/nvvp/.eclipseproduct
%_libdir/nvvp/*
%_datadir/applications/nvvp.desktop
%_iconsdir/hicolor/*/apps/nvvp.png

%files -n nvidia-nsight
%_bindir/nsight
%_libdir/nsight/.eclipseproduct
%_libdir/nsight/*
%_datadir/applications/nsight.desktop
%_iconsdir/hicolor/*/apps/nsight.png
%endif


%changelog
* Tue Mar 10 2015 akien <akien> 6.5.14-6.mga5
+ Revision: 818296
- Do not prevent the use of GCC 4.9

* Fri Feb 06 2015 tmb <tmb> 6.5.14-5.mga5.nonfree
+ Revision: 813665
- fix gdb files removal
- dont ship gdb files (mga#15187)

* Wed Oct 15 2014 umeabot <umeabot> 6.5.14-3.mga5.nonfree
+ Revision: 748872
- Second Mageia 5 Mass Rebuild

* Tue Sep 16 2014 umeabot <umeabot> 6.5.14-2.mga5.nonfree
+ Revision: 682957
- Mageia 5 Mass Rebuild

  + tv <tv>
    - use %%global for req/prov exclude
    - autoconvert to new prov/req excludes
    - s/uggests:/Recommends:/

* Tue Sep 09 2014 tmb <tmb> 6.5.14-1.mga5.nonfree
+ Revision: 674217
- SDK, samples, nvidia-visual-profiler and nvidia-nsight is now 64bit only
- update filelists
- update to 6.5.14

* Sun Dec 15 2013 tmb <tmb> 5.5.22-2.mga4.nonfree
+ Revision: 557042
- fix libdevice path (mga#11941)

* Sun Dec 08 2013 tmb <tmb> 5.5.22-1.mga4.nonfree
+ Revision: 555985
- update filelists
- update to 5.5.22

* Thu Oct 31 2013 tmb <tmb> 5.0.35-4.mga4.nonfree
+ Revision: 548445
- Mageia 4 rebuild

* Sun Apr 07 2013 mitya <mitya> 5.0.35-3.mga3.nonfree
+ Revision: 408642
- Fix BR jpackage-utils
- Fix #8755

* Sat Jan 12 2013 umeabot <umeabot> 5.0.35-2.mga3.nonfree
+ Revision: 361085
- Mass Rebuild - https://wiki.mageia.org/en/Feature:Mageia3MassRebuild

* Fri Dec 14 2012 mitya <mitya> 5.0.35-1.mga3.nonfree
+ Revision: 331007
- New version 5.0.35
- Optimize directory structure
- Use SLES packages
- Include samples (ex-SDK)

* Sat May 12 2012 mitya <mitya> 4.2.9-2.mga3.nonfree
+ Revision: 235438
- Restore nvidia RC script; add LSB headers

* Sun Apr 22 2012 mitya <mitya> 4.2.9-1.mga2.nonfree
+ Revision: 232730
- New version 4.2.9

* Thu Apr 05 2012 mitya <mitya> 4.1.28-7.mga2.nonfree
+ Revision: 228788
+ rebuild (emptylog)

* Fri Mar 23 2012 mitya <mitya> 4.1.28-6.mga2.nonfree
+ Revision: 225837
+ rebuild (emptylog)

* Fri Mar 23 2012 mitya <mitya> 4.1.28-5.mga2.nonfree
+ Revision: 225788
+ rebuild (emptylog)

* Mon Mar 19 2012 mitya <mitya> 4.1.28-4.mga2.nonfree
+ Revision: 224451
+ rebuild (emptylog)

* Mon Mar 12 2012 mitya <mitya> 4.1.28-3.mga2.nonfree
+ Revision: 223087
+ rebuild (emptylog)

* Mon Mar 12 2012 mitya <mitya> 4.1.28-2.mga2.nonfree
+ Revision: 223086
+ rebuild (emptylog)

* Wed Mar 07 2012 mitya <mitya> 4.1.28-1.mga2.nonfree
+ Revision: 221359
- New version 4.1.28

* Sat Oct 01 2011 supp <supp> 4.0.17-2.mga2.nonfree
+ Revision: 150695
- apply official NV patches for cuda profiler

* Fri Sep 23 2011 lebedov <lebedov> 4.0.17-1.mga2
+ Revision: 147061
- Update to 4.0.17.

  + tmb <tmb>
    - rebuild to get repo back in sync

* Thu Feb 24 2011 ahmad <ahmad> 3.2.16-1.mga1
+ Revision: 58188
- imported package nvidia-cuda-toolkit


* Thu Nov 25 2010 Lev Givon <lev@mandriva.org> 3.2.16-1mdv2011.0
+ Revision: 601414
- Update to 3.2.
  Use Ubuntu instead of RHEL tarballs because their linkage
  is more conformant with Mandriva lib versions.

* Fri Sep 03 2010 Lev Givon <lev@mandriva.org> 3.1-1mdv2011.0
+ Revision: 575702
- Update to 3.1.

* Fri Mar 26 2010 Lev Givon <lev@mandriva.org> 3.0-1mdv2010.1
+ Revision: 527582
- Update to 3.0.

* Wed Mar 17 2010 Lev Givon <lev@mandriva.org> 2.3-4mdv2010.1
+ Revision: 524731
- Set /dev/nvidia* perms in startup script even when files
  already exist.

* Tue Nov 10 2009 Lev Givon <lev@mandriva.org> 2.3-3mdv2010.1
+ Revision: 464182
- Suppress problematic automatically generated dependency
  in nvidia-cuda-toolkit-devel.

* Mon Nov 09 2009 Lev Givon <lev@mandriva.org> 2.3-2mdv2010.1
+ Revision: 463330
- Fix issues reported in #55425.

* Sun Nov 08 2009 Lev Givon <lev@mandriva.org> 2.3-1mdv2010.1
+ Revision: 463066
- Update to 2.3.

* Mon Jul 13 2009 Lev Givon <lev@mandriva.org> 2.2-2mdv2010.0
+ Revision: 395463
- Don't build profiler package for x86_64 because the profiler binary
  is only 32 bit.

* Fri Jul 10 2009 Lev Givon <lev@mandriva.org> 2.2-1mdv2010.0
+ Revision: 394322
- Update to 2.2.

* Thu Mar 05 2009 Lev Givon <lev@mandriva.org> 2.1-1mdv2009.1
+ Revision: 348730
- Update to 2.1.

* Tue Feb 24 2009 Lev Givon <lev@mandriva.org> 2.0-4mdv2009.1
+ Revision: 344449
- Don't require nvidia or nvidia-devel so that the cuda
  tools can be run on non-nvidia systems.

* Mon Feb 23 2009 Lev Givon <lev@mandriva.org> 2.0-3mdv2009.1
+ Revision: 344315
- Rebuilt against new nvidia drivers.

* Mon Feb 02 2009 Lev Givon <lev@mandriva.org> 2.0-2mdv2009.1
+ Revision: 336532
- Rebuild.
- import nvidia-cuda-toolkit