Differences
This shows you the differences between two versions of the page.
Both sides previous revision Previous revision | |||
doku:vsc3_gpu [2018/04/10 09:47] – markus | doku:vsc3_gpu [2018/04/10 09:52] (current) – [Slurm integration] markus | ||
---|---|---|---|
Line 37: | Line 37: | ||
==== Slurm integration ==== | ==== Slurm integration ==== | ||
- | There is one partition called '' | + | There are several partions |
- | < | + | < |
- | PARTITION AVAIL TIMELIMIT | + | gpu_gtx1080single |
- | gpu | + | gpu_gtx1080multi |
- | gpu | + | gpu_k20m |
+ | gpu_m60 | ||
</ | </ | ||
- | and which needs to be specified via: | ||
- | < | ||
- | GPU nodes are selected via the **generic resource (--gres=)** and **constraints (-C, | + | For each partition a identically named QOS is defined. Slurm usage is eg.: |
- | * k20m (kepler) GPU nodes: | + | |
- | #SBATCH --gres=gpu:2 | + | < |
+ | #SBATCH -p gpu_gtx1080single | ||
+ | #SBATCH --qos gpu_gtx1080single | ||
</ | </ | ||
- | * m60 (maxwell) GPU nodes: < | ||
- | #SBATCH --gres=gpu: | ||
- | </ | ||
- | * gtx1080 (pascal) GPU nodes: < | ||
- | #SBATCH --gres=gpu: | ||
- | </ | ||
- | * at idle times of private-shared gtx1080 (pascal) GPU nodes: < | ||
- | #SBATCH -C gtx1080 | ||
- | #SBATCH --gres=gpu: | ||
- | </ | ||
- | |||
- | To use a gpu node for computing purposes the quality of service (QoS) '' | ||
- | < | ||
- | For visualization the QoS '' | ||
- | < | ||
- | When a job is submitted within the '' | ||
-------------------------------------------------- | -------------------------------------------------- |