This shows you the differences between two versions of the page.
Both sides previous revision Previous revision | |||
singularity-apptainer [2023/06/14 20:49] root |
singularity-apptainer [2023/06/14 20:54] (current) root |
||
---|---|---|---|
Line 134: | Line 134: | ||
$ docker pull nvcr.io/ | $ docker pull nvcr.io/ | ||
$ docker save nvcr.io/ | $ docker save nvcr.io/ | ||
- | $ scp / | + | $ scp / |
</ | </ | ||
- | As myhpcuserid | + | As " |
< | < | ||
- | $ cd /storage/myhpcuserid | + | $ cd /storage/rfeynman |
$ singularity build hpc-benchmarks.sif docker-archive:// | $ singularity build hpc-benchmarks.sif docker-archive:// | ||
</ | </ | ||
- | Then for nvidia get a gpu node with srun (and use --nv), or continue with cloud node (without --nv) for serial computing: | + | Then for nvidia, get a gpu node with srun (and use shell --nv), or continue with cloud node (without --nv) for serial |
< | < | ||
- | $ singularity shell --nv --bind / | + | $ singularity shell --nv --bind / |
</ | </ | ||
- | For comparison, we'll run HPL with the cpu on the same node in bare metal. At this memory size it' | + | For comparison, we'll run HPL with the cpu on the same node in bare metal. At this memory size it runs about 2.4 TF for this dual 7543 AMD. |
< | < | ||
Line 193: | Line 193: | ||
-------------------------------------------------------------------------------- | -------------------------------------------------------------------------------- | ||
WC32L2C4 | WC32L2C4 | ||
- | HPL_pdgesv() start time Wed Jun 14 15:42:54 2023 | ||
- | HPL_pdgesv() end time Wed Jun 14 15:44:03 2023 | ||
================================================================================ | ================================================================================ | ||
Line 209: | Line 207: | ||
</ | </ | ||
- | We'll copy this to apptainer and change the MPI grid from 4x4 to 1x1 in HPL.dat. In this measure, a single NVidia A100 GPU is a little over 5 times as fast as two AMD 7543 CPUs. The 40 GB memory of this GPU is close to full. | + | We'll copy this HPL.dat |
< | < |