Created attachment 507796 [details] build error after changing the system profile to 17.0/systemd, nvidia-cuda-sdk-9.0.176 couldn't be build successfully, though I have successfully installed it before the profile change. The error is attached. #$ cicc --c++14 --gnu_version=60400 --allow_managed -arch compute_70 -m64 -ftz=0 -prec_div=1 -prec_sqrt=1 -fmad=1 --include_file_name "tmpxft_0001f50c_00000000-2_simpleIPC.fatbin.c" -tused -nvvmir-library "/opt/cuda/bin/../nvvm/libdevice/libdevice.10.bc" --gen_module_id_file --module_id_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f50c_00000000-3_simpleIPC.module_id" --orig_src_file_name "simpleIPC.cu" --gen_c_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f50c_00000000-5_simpleIPC.compute_70.cudafe1.c" --stub_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f50c_00000000-5_simpleIPC.compute_70.cudafe1.stub.c" --gen_device_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f50c_00000000-5_simpleIPC.compute_70.cudafe1.gpu" "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f50c_00000000-12_simpleIPC.compute_70.cpp1.ii" -o "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f50c_00000000-5_simpleIPC.compute_70.ptx" #$ cicc --c++14 --gnu_version=60400 --allow_managed -arch compute_70 -m64 -ftz=0 -prec_div=1 -prec_sqrt=1 -fmad=1 --include_file_name "tmpxft_0001f593_00000000-2_template.fatbin.c" -tused -nvvmir-library "/opt/cuda/bin/../nvvm/libdevice/libdevice.10.bc" --gen_module_id_file --module_id_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f593_00000000-3_template.module_id" --orig_src_file_name "template.cu" --gen_c_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f593_00000000-5_template.compute_70.cudafe1.c" --stub_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f593_00000000-5_template.compute_70.cudafe1.stub.c" --gen_device_file_name "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f593_00000000-5_template.compute_70.cudafe1.gpu" "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f593_00000000-12_template.compute_70.cpp1.ii" -o "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f593_00000000-5_template.compute_70.ptx" /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined 2 errors detected in the compilation of "/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/temp/tmpxft_0001f535_00000000-6_vectorAdd_kernel.cpp1.ii". /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined /usr/include/bits/floatn.h(61): error: invalid argument to attribute "__mode__" /usr/include/bits/floatn.h(73): error: identifier "__float128" is undefined make[1]: Entering directory '/home/portage/portage/dev-util/nvidia-cuda-sdk-9.0.176/work/samples/0_Simple/cudaOpenMP' # --error 0x1 --
Created attachment 507798 [details] emerge info
I confirm this bug. This bug is probably caused by either sys-libs/glibc-2.26-r3 or Gentoo profile version 17 (default/linux/amd64/17.0/systemd).
And I faced the same issue when emerging dev-lang/gnat-gpl on Gentoo profile 17. dev-lang/gnat-gpl-2017 dev-util/nvidia-cuda-sdk-9.0.176
Probably similar bug: https://bugs.gentoo.org/639808
Same here. Failed to compile dev-util/nvidia-cuda-toolkit-9.0.176. On the internet it is said that it is due to incompatibility with glibc-2.26. My profile is default/linux/amd64/13.0/desktop/plasma
This hack works for me with glibc-2.26 and gcc-6.3.0 https://devtalk.nvidia.com/default/topic/1023776/cuda-programming-and-performance/-request-add-nvcc-compatibility-with-glibc-2-26/post/5210367/#5210367 But it seems to me not the best way to make thing up ...
I confirm this bug. Seems to be blocker for me.
The fix seems to be not available yet: https://devtalk.nvidia.com/default/topic/1023776/cuda-programming-and-performance/-request-add-nvcc-compatibility-with-glibc-2-26/2 See the latest post.
this is a duplicate of 638230.
*** This bug has been marked as a duplicate of bug 638230 ***
