...
With/out Hyperthreading (SMT)
Model | User | Pro SMT | Contra SMT |
---|---|---|---|
-- | admin | ~2500 kJ (∑ Esocket[0-7] according to lm_sensors), ~30% less | Nodes need more power (3500 kJ) and get warmer in idle state |
FESOM | NEC | Using 128 Threads per node: 3% faster (probably because the (buggy) GXFS daemon can use a virtual core) | Using 256 Thread per node: 10% slower |
Python AI | vhelm | no impact/difference | no impact/difference |
GPU nodes
- no difference in runtime between A40 and A100 ( tensorflow-gpu AI application)
- ???
...