Skip to content
Snippets Groups Projects

How to check if OpenMPI has support for CUDA

2 unresolved threads
1 file
+ 9
1
Compare changes
  • Side-by-side
  • Inline
@@ -200,7 +200,15 @@ detail in [nvcc documentation](https://docs.nvidia.com/cuda/cuda-compiler-driver
@@ -200,7 +200,15 @@ detail in [nvcc documentation](https://docs.nvidia.com/cuda/cuda-compiler-driver
This compiler is available via several `CUDA` packages, a default version can be loaded via
This compiler is available via several `CUDA` packages, a default version can be loaded via
`module load CUDA`. Additionally, the `NVHPC` modules provide CUDA tools as well.
`module load CUDA`. Additionally, the `NVHPC` modules provide CUDA tools as well.
#### Usage of the CUDA compiler
For using CUDA with OpenMPI at multiple nodes, the OpenMPI module loaded shall have be compiled with
 
CUDA support. If you aren't sure if the module you are using has support for it you can check it as
 
following:
    • Comment on lines +204 to +205
      Suggested change
      204 CUDA support. If you aren't sure if the module you are using has support for it you can check it as
      205 following:
      204 CUDA support. If you aren't sure if the module you are using has support for it you can check it
      205 using the following commands:
Please register or sign in to reply
 
 
```console
 
ompi_info --parsable --all | grep mpi_built_with_cuda_support:value | awk -F":" '{print "OpenMPI supports CUDA:",$7}'
    • Suggested change
      208 ompi_info --parsable --all | grep mpi_built_with_cuda_support:value | awk -F":" '{print "OpenMPI supports CUDA:",$7}'
      208 marie@login$ ompi_info --parsable --all | grep mpi_built_with_cuda_support:value | awk -F":" '{print "OpenMPI supports CUDA:",$7}'
Please register or sign in to reply
 
```
 
 
#### Usage of the CUDA Compiler
The simple invocation `nvcc <code.cu>` will compile a valid CUDA program. `nvcc` differentiates
The simple invocation `nvcc <code.cu>` will compile a valid CUDA program. `nvcc` differentiates
between the device and the host code, which will be compiled in separate phases. Therefore, compiler
between the device and the host code, which will be compiled in separate phases. Therefore, compiler
Loading