Recent Updates¶
Phase 2 hardware is available¶
Phase 2 hardware has been installed and is available to use for everyone! Here is how you use it:
- New (Phase 2) compute nodes (64 "Sapphire Rapid" CPUs per node) can be requested with
#SBATCH --partition=compute-p2
. The old (Phase 1) compute nodes are now calledcompute-p1
, and they remain available with the aliascompute
for backward-compatibility. - New (Phase 2) GPU nodes (4x A100 cards per node) can be requested with
#SBATCH --partition=gpu-a100
. The old (Phase 1) GPU nodes are now calledgpu-v100
, and they remain available with the aliasgpu
for backward-compatibility.
Intel OneAPI module has been updated following the update to Infiniband drivers¶
Following the update to Infiniband drivers, the intel/oneapi-all
module was updated to the version 2022.3. If your MPI application was compiled with Intel compilers, you might want to recompile it to make sure it works correctly.
For the time being, there are two modules, the old one and the new one:
Please note: after the maintenance weekend 17-18 December 2022, the "old" intel/oneapi-all
will be removed, and the "new" intel/oneapi-all_2022.3
will be renamed into intel/oneapi-all
.