1 |
%define name nvidia-cuda-toolkit |
2 |
%define version 3.2.16 |
3 |
%define release %mkrel 1 |
4 |
|
5 |
%define driver_ver 260.19.21 |
6 |
|
7 |
Summary: NVIDIA CUDA Toolkit libraries |
8 |
Name: %{name} |
9 |
Version: %{version} |
10 |
Release: %{release} |
11 |
Source0: cudatoolkit_%{version}_linux_32_ubuntu10.04.run |
12 |
Source1: cudatoolkit_%{version}_linux_64_ubuntu10.04.run |
13 |
Source2: nvidia |
14 |
License: Freeware |
15 |
Group: System/Libraries |
16 |
Url: http://www.nvidia.com/cuda/ |
17 |
BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-buildroot |
18 |
Suggests: nvidia >= %{driver_ver} |
19 |
# We don't require installation of the NVIDIA graphics drivers so that |
20 |
# folks can do CUDA development on systems without NVIDIA hardware. |
21 |
|
22 |
%description |
23 |
NVIDIA(R) CUDA(TM) is a general purpose parallel computing architecture |
24 |
that leverages the parallel compute engine in NVIDIA graphics |
25 |
processing units (GPUs) to solve many complex computational problems |
26 |
in a fraction of the time required on a CPU. It includes the CUDA |
27 |
Instruction Set Architecture (ISA) and the parallel compute engine in |
28 |
the GPU. To program to the CUDATM architecture, developers can, today, |
29 |
use C, one of the most widely used high-level programming languages, |
30 |
which can then be run at great performance on a CUDATM enabled |
31 |
processor. Other languages will be supported in the future, including |
32 |
FORTRAN and C++. |
33 |
|
34 |
This package contains the libraries and attendant files needed to run |
35 |
programs that make use of CUDA. |
36 |
|
37 |
%package devel |
38 |
Summary: NVIDIA CUDA Toolkit development files |
39 |
Group: Development/C |
40 |
Requires: %{name} = %{version}-%{release} |
41 |
Suggests: nvidia-devel >= %{driver_ver} |
42 |
%define _requires_exceptions libcuda.so.3\\|libcudart.so.3 |
43 |
|
44 |
%description devel |
45 |
NVIDIA(R) CUDA(TM) is a general purpose parallel computing architecture |
46 |
that leverages the parallel compute engine in NVIDIA graphics |
47 |
processing units (GPUs) to solve many complex computational problems |
48 |
in a fraction of the time required on a CPU. It includes the CUDA |
49 |
Instruction Set Architecture (ISA) and the parallel compute engine in |
50 |
the GPU. To program to the CUDATM architecture, developers can, today, |
51 |
use C, one of the most widely used high-level programming languages, |
52 |
which can then be run at great performance on a CUDATM enabled |
53 |
processor. Other languages will be supported in the future, including |
54 |
FORTRAN and C++. |
55 |
|
56 |
This package contains the development files needed to build programs |
57 |
that make use of CUDA. |
58 |
|
59 |
%package -n nvidia-compute-profiler |
60 |
Summary: NVIDIA Compute Visual Profiler |
61 |
Group: Development/Other |
62 |
Requires: %{name} = %{version}-%{release} |
63 |
Obsoletes: nvidia-cuda-profiler, nvidia-opencl-profiler |
64 |
Suggests: nvidia-devel >= %{driver_ver} |
65 |
Suggests: qt4-assistant |
66 |
|
67 |
%description -n nvidia-compute-profiler |
68 |
NVIDIA(R) CUDA(TM) is a general purpose parallel computing architecture |
69 |
that leverages the parallel compute engine in NVIDIA graphics |
70 |
processing units (GPUs) to solve many complex computational problems |
71 |
in a fraction of the time required on a CPU. It includes the CUDA |
72 |
Instruction Set Architecture (ISA) and the parallel compute engine in |
73 |
the GPU. To program to the CUDATM architecture, developers can, today, |
74 |
use C, one of the most widely used high-level programming languages, |
75 |
which can then be run at great performance on a CUDATM enabled |
76 |
processor. Other languages will be supported in the future, including |
77 |
FORTRAN and C++. |
78 |
|
79 |
This package contains the Compute Visual Profiler for CUDA and OpenCL. |
80 |
|
81 |
%prep |
82 |
%setup -q -T -c %{name}-%{version} |
83 |
|
84 |
%install |
85 |
%__rm -rf %{buildroot} |
86 |
|
87 |
%__install -d -m 755 %{buildroot}%{_usr} |
88 |
|
89 |
%ifarch %ix86 |
90 |
bash %SOURCE0 --tar xf -C %{buildroot}%{_usr} |
91 |
%else |
92 |
bash %SOURCE1 --tar xf -C %{buildroot}%{_usr} |
93 |
%__rm -rf %{buildroot}/usr/lib |
94 |
%__sed -i 's/lib/lib64/g' %{buildroot}%{_bindir}/nvcc.profile |
95 |
%endif |
96 |
|
97 |
%__mv %{buildroot}%{_usr}/doc ./ |
98 |
|
99 |
%__rm -rf %{buildroot}%{_usr}/src |
100 |
%__rm -rf %{buildroot}%{_usr}/install-linux.pl |
101 |
|
102 |
%__mv %{buildroot}%{_usr}/computeprof/bin/computeprof %{buildroot}%{_bindir}/ |
103 |
%__mkdir computeprofdoc |
104 |
%__mv %{buildroot}%{_usr}/computeprof/*.txt computeprofdoc/ |
105 |
%__mv %{buildroot}%{_usr}/computeprof/doc/* computeprofdoc/ |
106 |
%__mv %{buildroot}%{_usr}/computeprof/projects computeprofdoc/ |
107 |
%__rm -rf %{buildroot}%{_usr}/computeprof |
108 |
|
109 |
%__install -D -m 755 %SOURCE2 %{buildroot}%{_sysconfdir}/init.d/nvidia |
110 |
|
111 |
%clean |
112 |
%__rm -rf %{buildroot} |
113 |
|
114 |
%post -p /sbin/ldconfig |
115 |
|
116 |
%postun -p /sbin/ldconfig |
117 |
|
118 |
%files |
119 |
%_libdir/*.so.* |
120 |
%_sysconfdir/init.d/* |
121 |
|
122 |
%files devel |
123 |
%defattr(-,root,root) |
124 |
%doc doc/* |
125 |
%_bindir/* |
126 |
%exclude %_bindir/computeprof |
127 |
%_libdir/*.so |
128 |
%_includedir/* |
129 |
%_usr/open64/* |
130 |
|
131 |
%files -n nvidia-compute-profiler |
132 |
%defattr(-,root,root) |
133 |
%doc computeprofdoc/* |
134 |
%_bindir/computeprof |
135 |
|
136 |
|
137 |
|