[Openmp-dev] GPU offloading with clang

[Talita Perciano via Openmp-dev]

Hi,

I'm trying to use openmp offloading to gpu, and I'm getting linking errors
during compilation:

[  7%] Linking CXX executable MPI_PMRF
/usr/bin/ld:
CMakeFiles/MPI_PMRF.dir-------/cfuncs-pmrf.c.o:(.rodata..omp_offloading.device_images[.omp_offloading.descriptor_reg.nvptx64-nvidia-cuda]+0x10):
undefined reference to `.omp_offloading.entries_begin'
/usr/bin/ld:
CMakeFiles/MPI_PMRF.dir-------/cfuncs-pmrf.c.o:(.rodata..omp_offloading.device_images[.omp_offloading.descriptor_reg.nvptx64-nvidia-cuda]+0x18):
undefined reference to `.omp_offloading.entries_end'
/usr/bin/ld:
CMakeFiles/MPI_PMRF.dir/-------/cfuncs-pmrf.c.o:(.rodata..omp_offloading.descriptor[.omp_offloading.descriptor_reg.nvptx64-nvidia-cuda]+0x10):
undefined reference to `.omp_offloading.entries_begin'
/usr/bin/ld:
CMakeFiles/MPI_PMRF.dir/-------/cfuncs-pmrf.c.o:(.rodata..omp_offloading.descriptor[.omp_offloading.descriptor_reg.nvptx64-nvidia-cuda]+0x18):
undefined reference to `.omp_offloading.entries_end'
/usr/bin/ld: link errors found, deleting executable `MPI_PMRF'
clang-9: error: linker command failed with exit code 1 (use -v to see
invocation)
CMakeFiles/MPI_PMRF.dir/build.make:265: recipe for target 'MPI_PMRF' failed
make[2]: *** [MPI_PMRF] Error 1
CMakeFiles/Makefile2:72: recipe for target 'CMakeFiles/MPI_PMRF.dir/all'
failed
make[1]: *** [CMakeFiles/MPI_PMRF.dir/all] Error 2
Makefile:83: recipe for target 'all' failed
make: *** [all] Error 2

I actually can compile the code just fine without the #pragma omp target.

Some additional info:

- I'm compiling the code with *-fopenmp
-fopenmp-targets=nvptx64-nvidia-cuda*

- *clang version 9.0.0* (https://github.com/clang-ykt/clang
e50a43ae7e2fe6e6ff515cedb2a8ae658120e32f) (https://github.com/clang-ykt/llvm
3cd66813b063ab08a389470f1bc53e18fbe595b2)

- *cuda info:*

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 418.67       Driver Version: 418.67       CUDA Version: 10.1
  |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr.
ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute
M. |
|===============================+======================+======================|
|   0  Tesla V100-SXM2...  On   | 00000000:3E:00.0 Off |
 0 |
| N/A   30C    P0    38W / 300W |      0MiB / 16130MiB |      0%
 Default |
+-------------------------------+----------------------+----------------------+

- Linux kernel: *4.4.176-96-default*

Any thoughts?

Thanks!
Talita
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/openmp-dev/attachments/20190702/4d4c1ad7/attachment.html>