Jump to content

Photo

Why enabling "cuda" use flags portage requires gcc-4.8.5-r1


Best Answer cardinal , 14 April 2017 - 04:07 AM

dev-util/nvidia-cuda-toolkit-6.5.19-r1.ebuild
RDEPEND="${DEPEND}
<sys-devel/gcc-4.9[cxx] Go to the full post »


2 replies to this topic

#1
Sandro

Sandro

    Advanced Member

  • Members
  • PipPipPip
  • 350 posts
  • LocationSaronno, Via IV Novembre, 23 21047 Italy

Hi to all ....

Adding "cuda" use flag as global, portage wants to install GCC-4.8.5-r1.

And i don't know why.

sandro@ci74771ht ~ $ USE="cuda" emerge -DNupv world 

These are the packages that would be merged, in order: 

Calculating dependencies... done!
[ebuild  NS    ] sys-devel/gcc-4.8.5-r1:4.8.5::gentoo [5.3.0-r1:5.3.0::gentoo] USE="cxx fo
rtran go (multilib) nls objc objc++ openmp (-altivec) -doc (-hardened) (-libssp) -vanilla"
0 KiB
[ebuild  N     ] dev-util/nvidia-cuda-toolkit-6.5.19-r1:0/6.5.19::gentoo  USE="-debugger -
doc -eclipse -profiler" 733066 KiB 
[ebuild   R    ] sys-apps/hwloc-1.11.2:0/5::gentoo  USE="X cairo cuda* numa pci svg xml -d
ebug -gl (-opencl) -plugins -static-libs" ABI_X86="32 (64) (-x32)" 3925 KiB 
[ebuild   R    ] sys-cluster/openmpi-2.0.2::gentoo  USE="cuda* cxx fortran ipv6 java mpi-t
hreads* romio threads -cma -heterogeneous -numa" ABI_X86="32 (64) (-x32)" OPENMPI_FABRICS=
"-knem -ofed -psm" OPENMPI_OFED_FEATURES="-connectx-xrc -control-hdr-padding -dynamic-sl -
failover -rdmacm -udcm" OPENMPI_RM="-pbs -slurm" 8117 KiB 
[ebuild   R   ~] sci-misc/boinc-7.6.33-r3::gentoo  USE="X cuda* -static-libs" CURL_SSL="op
enssl -libressl" 42830 KiB 
[ebuild   R    ] media-libs/opencv-2.4.13:0/2.4::gentoo  USE="cuda* eigen ffmpeg gstreamer
gtk ieee1394 java jpeg jpeg2k opencl openexr opengl openmp png python qt5 threads tiff v4
l xine -examples (-ipp) (-libav) -pch -qt4 -testprograms (-vtk)" PYTHON_TARGETS="python2_7
" 92110 KiB 

Total: 6 packages (1 new, 1 in new slot, 4 reinstalls), Size of downloads: 880047 KiB

I'm with Funtoo Stable 64.

 

Thanks for any suggestion :)



#2
cardinal

cardinal

    Advanced Member

  • Members
  • PipPipPip
  • 239 posts
✓  Best Answer
dev-util/nvidia-cuda-toolkit-6.5.19-r1.ebuild
RDEPEND="${DEPEND}
<sys-devel/gcc-4.9[cxx]
  • Sandro likes this

#3
Sandro

Sandro

    Advanced Member

  • Members
  • PipPipPip
  • 350 posts
  • LocationSaronno, Via IV Novembre, 23 21047 Italy
>>> Installing (1 of 1) dev-util/nvidia-cuda-toolkit-6.5.19-r1::gentoo 

* gcc >= 4.8 will not work with CUDA 
* Make sure you set an earlier version of gcc with gcc-config 
* or append --compiler-bindir= pointing to a gcc bindir like 
* --compiler-bindir=/usr/*pc-linux-gnu/gcc-bin/gcc4.8 
* to the nvcc compiler flags 


* Messages for package dev-util/nvidia-cuda-toolkit-6.5.19-r1: 

* gcc >= 4.8 will not work with CUDA 
* Make sure you set an earlier version of gcc with gcc-config 
* or append --compiler-bindir= pointing to a gcc bindir like 
* --compiler-bindir=/usr/*pc-linux-gnu/gcc-bin/gcc4.8 
* to the nvcc compiler flags 
>>> Auto-cleaning packages...

>>> No outdated packages were found on your system.

* GNU info directory index is up-to-date.
ci74771ht ~ # gcc-config -f 1 
* Switching native-compiler to x86_64-pc-linux-gnu-4.8.5 ... 
>>> Regenerating /etc/ld.so.cache...                                                [ ok ] 

* If you intend to use the gcc from the new profile in an already 
* running shell, please remember to do: 

*   . /etc/profile 

ci74771ht ~ # . /etc/profile 
ci74771ht ~ # emerge -1 nvidia-cuda-toolkit 
Calculating dependencies... done!

>>> Verifying ebuild manifests
>>> Running pre-merge checks for dev-util/nvidia-cuda-toolkit-6.5.19-r1 
* Checking for at least 1500 MiB disk space at "/var/tmp/portage/dev-util/nvidia-cuda-too
lkit-6.5.19-r1/temp" ...                                                           [ ok ] 

>>> Emerging (1 of 1) dev-util/nvidia-cuda-toolkit-6.5.19-r1::gentoo 
* cuda_6.5.19_linux_64.run SHA256 SHA512 WHIRLPOOL size ;-) ...                   [ ok ] 
* Checking for at least 1500 MiB disk space at "/var/tmp/portage/dev-util/nvidia-cuda-too
lkit-6.5.19-r1/temp" ...                                                           [ ok ] 
>>> Unpacking source...
>>> Unpacking cuda_6.5.19_linux_64.run to /var/tmp/portage/dev-util/nvidia-cuda-toolkit-6.
5.19-r1/work
86512+1 records in
1918818+1 records out
982434882 bytes (982 MB) copied, 4,46839 s, 220 MB/s
>>> Unpacking cuda-linux64-rel-6.5.19-18849900.run to /var/tmp/portage/dev-util/nvidia-cud
a-toolkit-6.5.19-r1/work
82325+1 records in
1592477+1 records out
815348401 bytes (815 MB) copied, 7,98509 s, 102 MB/s
>>> Source unpacked in /var/tmp/portage/dev-util/nvidia-cuda-toolkit-6.5.19-r1/work
>>> Preparing source in /var/tmp/portage/dev-util/nvidia-cuda-toolkit-6.5.19-r1/work ...
>>> Source prepared.
>>> Configuring source in /var/tmp/portage/dev-util/nvidia-cuda-toolkit-6.5.19-r1/work ...
>>> Source configured.
>>> Compiling source in /var/tmp/portage/dev-util/nvidia-cuda-toolkit-6.5.19-r1/work ...
>>> Source compiled.
>>> Test phase [not enabled]: dev-util/nvidia-cuda-toolkit-6.5.19-r1

>>> Install nvidia-cuda-toolkit-6.5.19-r1 into /var/tmp/portage/dev-util/nvidia-cuda-toolk
it-6.5.19-r1/image/ category dev-util
* Cleaning doc... ...                                                             [ ok ] 
* Cleaning jre... ...                                                             [ ok ] 
* Cleaning run_files... ...                                                       [ ok ] 
* Cleaning install-linux.pl... ...                                                [ ok ] 
* Cleaning bin/cuda-gdb... ...                                                    [ ok ] 
* Cleaning extras/Debugger... ...                                                 [ ok ] 
* Cleaning libnsight... ...                                                       [ ok ] 
* Cleaning libnvvp... ...                                                         [ ok ] 
* Cleaning extras/CUPTI... ...                                                    [ ok ] 
>>> Completed installing nvidia-cuda-toolkit-6.5.19-r1 into /var/tmp/portage/dev-util/nvid
ia-cuda-toolkit-6.5.19-r1/image/

* Final size of build directory: 0 KiB 
* Final size of installed tree: 696160 KiB 

making executable: opt/cuda/lib64/libOpenCL.so
strip: x86_64-pc-linux-gnu-strip --strip-unneeded -R .comment -R .GCC.command.line -R .not
e.gnu.gold-version
  opt/cuda/lib64/libcublas_device.a
  opt/cuda/lib64/libcudart_static.a
  opt/cuda/lib64/libcufft_static.a
  opt/cuda/open64/lib/inline
  opt/cuda/lib64/libnpps_static.a
  opt/cuda/lib64/libcusparse_static.a
  opt/cuda/lib64/libcublas_static.a
  opt/cuda/lib64/libnppi_static.a
  opt/cuda/lib64/libcudadevrt.a
  opt/cuda/lib64/libnppc_static.a
  opt/cuda/lib64/libcufftw_static.a
  opt/cuda/lib64/libculibos.a
  opt/cuda/lib64/libcurand_static.a
  opt/cuda/lib/libcudart_static.a
  opt/cuda/lib/libcudadevrt.a
  opt/cuda/open64/bin/nvopencc
  opt/cuda/bin/cuda-gdbserver
  opt/cuda/open64/lib/be
  opt/cuda/open64/lib/gfec
ecompressdir: bzip2 -9 /usr/share/man
* QA Notice: Missing soname symlink(s): 
*  
*      opt/cuda/lib64/libOpenCL.so.1 -> libOpenCL.so 
*  

>>> Installing (1 of 1) dev-util/nvidia-cuda-toolkit-6.5.19-r1::gentoo 
>>> Auto-cleaning packages...

>>> No outdated packages were found on your system.

* GNU info directory index is up-to-date. 
ci74771ht ~ # 

W the "multislot" :D

 

Thanx Cardinal ... You're ever a "magic man" :)





Reply to this topic



  


0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users