Merge pull request EESSI#165 from casparvl/snellius_h100

Update Snellius config with H100 nodes
reframe-hpc · Jul 25, 2024 · 5cabf95 · 5cabf95
2 parents 926c3e4 + 93186d4
commit 5cabf95
Showing 1 changed file with 39 additions and 3 deletions.
diff --git a/config/surf_snellius.py b/config/surf_snellius.py
@@ -84,13 +84,12 @@
  },
  'descr': 'AMD Genoa CPU partition with native EESSI stack'
  },
-
  {
- 'name': 'gpu',
+ 'name': 'gpu_A100',
  'scheduler': 'slurm',
  'prepare_cmds': ['source %s' % common_eessi_init()],
  'launcher': 'mpirun',
- 'access': ['-p gpu', '--export=None'],
+ 'access': ['-p gpu_a100', '--export=None'],
  'environs': ['default'],
  'max_jobs': 60,
  'devices': [
@@ -121,6 +120,43 @@
  },
  'descr': 'Nvidia A100 GPU partition with native EESSI stack'
  },
+ {
+ 'name': 'gpu_H100',
+ 'scheduler': 'slurm',
+ 'prepare_cmds': ['source %s' % common_eessi_init()],
+ 'launcher': 'mpirun',
+ 'access': ['-p gpu_h100', '--export=None'],
+ 'environs': ['default'],
+ 'max_jobs': 60,
+ 'devices': [
+ {
+ 'type': DEVICE_TYPES[GPU],
+ 'num_devices': 4,
+ }
+ ],
+ 'resources': [
+ {
+ 'name': '_rfm_gpu',
+ 'options': ['--gpus-per-node={num_gpus_per_node}'],
+ },
+ {
+ 'name': 'memory',
+ 'options': ['--mem={size}'],
+ }
+ ],
+ 'features': [
+ FEATURES[GPU],
+ FEATURES[ALWAYS_REQUEST_GPUS],
+ ] + valid_scales_snellius_gpu,
+ 'extras': {
+ GPU_VENDOR: GPU_VENDORS[NVIDIA],
+ # Make sure to round down, otherwise a job might ask for more mem than is available
+ # per node
+ 'mem_per_node': 737280 # in MiB
+ },
+ 'descr': 'Nvidia H100 GPU partition with native EESSI stack'
+ },
+
  ]
  },
  ],