Differences

This shows you the differences between two versions of the page.

--- doku:vsc5quickstart [2022/04/27 11:13] – [Submit a Job] jz
+++ doku:vsc5quickstart [2023/01/05 21:05] – [SLURM] goldenberg
@@ Line 160: / Line 160: @@
 $ sinfo -o %P
 PARTITION
-gpu_a100_dual* -> Currently the default partition. AMD CPU nodes with 2x AMD Epyc (Milan) and 2x NIVIDA A100 and 512GB RAM
+zen2_0256_a40x2 -> AMD CPU nodes with 2x AMD Epyc (Milan) and 2x NIVIDA A40 and 256GB RAM
+jupyter -> reserved for the jupyterhub
+login5 -> login nodes, not an actual slurm partition
+zen3_2048 -> AMD CPU nodes with 2x AMD Epyc (Milan) and 2TB RAM
+zen3_1024 -> AMD CPU nodes with 2x AMD Epyc (Milan) and 1TB RAM
+zen3_0512* -> The default partition. AMD CPU nodes with 2x AMD Epyc (Milan) and 512GB RAM
 cascadelake_0384 -> Intel CPU nodes with 2x Intel Cascadelake and 384GB RAM
-</code>
+zen3_0512_a100x2 -> AMD CPU nodes with 2x AMD Epyc (Milan) and 2x NIVIDA A100 and 512GB RAM
-<code>
-$ sinfo
-PARTITION         AVAIL  TIMELIMIT  NODES  STATE NODELIST
-gpu_a100_dual        up   infinite     40   idle n571-[001-015],n572-[001-015],n573-[001-010]
-cascadelake_0384*    up   infinite     48   idle n451-[001-024],n452-[001-024]
 </code>
@@ Line 199: / Line 198: @@
 ./my_program
 </file>
+Job Scripts for the AMD CPU nodes:
+<file sh zen3_0512.sh>
+#!/bin/sh
+#SBATCH -J <meaningful name for job>
+#SBATCH -N 1
+#SBATCH --partition=zen3_0512
+#SBATCH --qos goodluck
+./my_program
+</file>
+<file sh zen3_1024.sh>
+#!/bin/sh
+#SBATCH -J <meaningful name for job>
+#SBATCH -N 1
+#SBATCH --partition=zen3_1024
+#SBATCH --qos goodluck
+./my_program
+</file>
+<file sh zen3_2048.sh>
+#!/bin/sh
+#SBATCH -J <meaningful name for job>
+#SBATCH -N 1
+#SBATCH --partition=zen3_2048
+#SBATCH --qos goodluck
+./my_program
+</file>
 Example job script to use both GPUs on a GPU nodes:
@@ Line 229: / Line 258: @@
 Official Slurm documentation: https://slurm.schedmd.com
+===== Intel MPI =====
+When **using Intel-MPI on the AMD nodes and mpirun** please set the following environment variable in your job script to allow for correct process pinning:
+<code>
+export I_MPI_PIN_RESPECT_CPUSET=0
+</code>