This shows you the differences between two versions of the page.
Both sides previous revision Previous revision | Next revision Both sides next revision | ||
cluster:164 [2017/10/27 19:30] hmeij07 |
cluster:164 [2017/10/27 19:38] hmeij07 |
||
---|---|---|---|
Line 564: | Line 564: | ||
==== PPMA Bench ==== | ==== PPMA Bench ==== | ||
+ | |||
+ | * Runs fastest when constrined to one gpu with 4 mpi threads | ||
+ | * Room for improvement as gpu and gpu memory are not fully utilized | ||
+ | * Adding mpi threads or more gpus reduces ns/day performance | ||
+ | * No idea if adding omp threads shows a different picture | ||
< | < | ||
+ | nvidia-smi -pm 0; nvidia-smi -c 0 | ||
+ | # gpu_id is done via CUDA_VISIBLE_DEVICES | ||
+ | export CUDA_VISIBLE_DEVCES=[0, | ||
+ | |||
+ | # on n78 | ||
+ | cd / | ||
+ | rm -f / | ||
+ | time / | ||
+ | / | ||
+ | -in nvt.in -var t 310 > /dev/null 2>& | ||
+ | |||
- | PMMA Benchmark Performance Metric (x nr of gpus) | + | PMMA Benchmark Performance Metric |
- | GTX on n78 | + | Lammps 11Aug17 |
-n 1, -gpu_id 3 | -n 1, -gpu_id 3 |