diff options
| author |  Axel Huebl <axel.huebl@plasma.ninja>
| 2023-06-23 10:08:17 -0700 |
|---|---|---|
| committer |  GitHub <noreply@github.com>
| 2023-06-23 10:08:17 -0700 |
| commit | 0f96d1786e974ae3726902d2e39b24ab900e524c (patch) | |
| tree | ce893e8a61e06fdbdb5307156c75beec54599350 /Source/FieldSolver/FiniteDifferenceSolver/HybridPICModel/HybridPICModel.cpp | |
| parent | 1adbf5e4ebc7df82fd95281aae490e26810fd46c (diff) | |
| download | WarpX-0f96d1786e974ae3726902d2e39b24ab900e524c.tar.gz WarpX-0f96d1786e974ae3726902d2e39b24ab900e524c.tar.zst WarpX-0f96d1786e974ae3726902d2e39b24ab900e524c.zip | |
Doc: HPC3 (UCI) HDF5 & GPU Visibility (#4021)
* Doc: HPC3 (UCI) HDF5 & GPU Visibility
Switch the HDF5 module to one that is picked up well
by our build logic.
Slurm is not well configured to properly address GPU visibility
per MPI rank. Thus, we define the node layouts during compile-time.
* GPUS_PER_NODE: In CMake
Looks like there is no env var support.
* GPU Visibility: Via SLURM_LOCALID
Not necessarily ideally pinned, but a good start.
Diffstat (limited to 'Source/FieldSolver/FiniteDifferenceSolver/HybridPICModel/HybridPICModel.cpp')
0 files changed, 0 insertions, 0 deletions