cory-master/BabelStream
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

CUDA Instructions for Spack

Browse Source
This commit is contained in:
Kaan Olgu 2023-06-17 09:17:32 +00:00
parent 406cc0010e
commit 9b13172235
1 changed files with 26 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
docs/spack_instructions.md
View File

@ -1,5 +1,6 @@
# Spack Instructions # Spack Instructions
## Table of contents ## Table of contents
* [OpenMP](#omp) * [OpenMP](#omp)
* [OpenCL](#ocl) * [OpenCL](#ocl)
@ -104,3 +105,28 @@
# Example 4: ROCM with GPU target and extra flags # Example 4: ROCM with GPU target and extra flags
$ spack install babelstream +rocm amdgpu_target=<gfx701> flags=<xxx> $ spack install babelstream +rocm amdgpu_target=<gfx701> flags=<xxx>
``` ```
## CUDA
* The `cuda_arch` value is mandatory here.
* If a user provides a value for `mem`, device memory mode will be chosen accordingly
* If a user provides a value for `flags`, additional CUDA flags will be passed to NVCC
* In the absence of `mem` and `flags`, the execution will choose **DEFAULT** for device memory mode and no additional flags will be passed
| Flag | Definition |
|-----------| ----------------------------------|
| cuda_arch |- List of supported compute capabilities are provided [here](https://github.com/spack/spack/blob/0f271883831bec6da3fc64c92eb1805c39a9f09a/lib/spack/spack/build_systems/cuda.py#LL19C1-L47C6) <br />- Useful [link](https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/) for matching CUDA gencodes with NVIDIA architectures|
|mem| Device memory mode: <br />- **DEFAULT** allocate host and device memory pointers.<br />- **MANAGED** use CUDA Managed Memory.<br />- **PAGEFAULT** shared memory, only host pointers allocated |
|flags | Extra flags to pass |
```shell
# Example 1: CUDA no mem and flags specified
$ spack install babelstream +cuda cuda_arch=70
# Example 2: for Nvidia GPU for Volta (sm_70)
$ spack install babelstream +cuda cuda_arch=70 mem=managed
# Example 3: CUDA with mem and flags specified
$ spack install babelstream +cuda cuda_arch=70 mem=managed flags=xxx
```