Skip to content
This repository has been archived by the owner on Mar 21, 2024. It is now read-only.

Workaround nvcxx compiler error. #1591

Merged
merged 1 commit into from
Jan 19, 2022
Merged

Conversation

alliepiper
Copy link
Collaborator

This resolves the following error on nvcxx 22.1:

nvdd-Fatal-/opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/bin/tools/nvvmd TERMINATED by signal 11
Arguments to /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/bin/tools/nvvmd
/opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/bin/tools/nvvmd -opt=3 -arch=compute_80 -ftz=0 -prec-div=1 -prec-sqrt=1 -fma=1 /tmp/pgaccTzSchA7vvoIJ.gpu /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/cuda/11.5/nvvm/libdevice/libdevice.10.bc /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_cuda_runtime_cc80.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_curand_runtime.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_nvshmem_runtime.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_utils_runtime.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_cuda_wmma_runtime_cc80.ll -ptx /tmp/pgacc9zSc3cmTLiu1.ptx
NVC++-F-0155-Compiler failed to translate accelerator region (see -Minfo messages): Device compiler exited with error status code (/workspace/testing/minmax_element.cu: 1)
NVC++/x86-64 Linux 22.1-0: compilation aborted
make[2]: *** [testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/build.make:75: testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/minmax_element.cu.o] Error 2
make[2]: Target 'testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/build' not remade because of errors.
make[1]: *** [CMakeFiles/Makefile2:6124: testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/all] Error 2

@dkolsen-pgi FYI

This resolves the following error on nvcxx 22.1:

```
nvdd-Fatal-/opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/bin/tools/nvvmd TERMINATED by signal 11
Arguments to /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/bin/tools/nvvmd
/opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/bin/tools/nvvmd -opt=3 -arch=compute_80 -ftz=0 -prec-div=1 -prec-sqrt=1 -fma=1 /tmp/pgaccTzSchA7vvoIJ.gpu /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/cuda/11.5/nvvm/libdevice/libdevice.10.bc /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_cuda_runtime_cc80.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_curand_runtime.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_nvshmem_runtime.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_utils_runtime.ll /opt/nvidia/hpc_sdk/Linux_x86_64/22.1/compilers/lib/nvvm70/nvhpc_cuda_wmma_runtime_cc80.ll -ptx /tmp/pgacc9zSc3cmTLiu1.ptx
NVC++-F-0155-Compiler failed to translate accelerator region (see -Minfo messages): Device compiler exited with error status code (/workspace/testing/minmax_element.cu: 1)
NVC++/x86-64 Linux 22.1-0: compilation aborted
make[2]: *** [testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/build.make:75: testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/minmax_element.cu.o] Error 2
make[2]: Target 'testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/build' not remade because of errors.
make[1]: *** [CMakeFiles/Makefile2:6124: testing/CMakeFiles/thrust.cpp.cuda.cpp17.test.minmax_element.dir/all] Error 2
```
@alliepiper alliepiper added P0: must have Absolutely necessary. Critical issue, major blocker, etc. compiler: nvc++ Specific to the NVC++ compiler. type: bug: compiler Bug in a compiler, not this library. labels Jan 18, 2022
@alliepiper alliepiper added this to the 1.16.0 milestone Jan 18, 2022
@alliepiper
Copy link
Collaborator Author

run tests

@alliepiper
Copy link
Collaborator Author

DVS CL: 30887415

@alliepiper alliepiper added testing: gpuCI in progress Started gpuCI testing. testing: internal ci in progress Currently testing on internal NVIDIA CI (DVS). labels Jan 18, 2022
@alliepiper alliepiper merged commit fe4a124 into NVIDIA:main Jan 19, 2022
@alliepiper alliepiper added the nvc++ flyspray Has an associated internal NVIDIA NVC++ flyspray. label Jan 19, 2022
@alliepiper alliepiper deleted the nvcxx-war branch January 19, 2022 19:29
@alliepiper
Copy link
Collaborator Author

Flyspray ID: 31222

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
compiler: nvc++ Specific to the NVC++ compiler. nvc++ flyspray Has an associated internal NVIDIA NVC++ flyspray. P0: must have Absolutely necessary. Critical issue, major blocker, etc. testing: gpuCI in progress Started gpuCI testing. testing: internal ci in progress Currently testing on internal NVIDIA CI (DVS). type: bug: compiler Bug in a compiler, not this library.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant