On Wed, 2021-12-22 at 16:02:00 +0000, Stuart MacLachlan wrote: > Hi Steffan, > > Not sure if the output from 'numactl --hardware' is more consistent and > easier to parse with a script or similar?
Hi, I'm getting confusing results. For an older dual 7351, there are 8 NUMA nodes, 4 physical cores each. (This already works with Slurm "Sockes=8 CorePerSocket=4".) For a dual 7713 running Ubuntu, kernel 5.11, I get 2 NUMA nodes, one per processor (64 physical cores, times 2). I've seen "lscpu" output for a 7313 which also shows 8 nodes, 4 cores, 2 threads each, kernel 4.19, Debian Buster. Does Ubuntu (or the 5.11 kernel) handle NUMA nodes difefrently? Thanks, Steffen > Hello, > > I'm wondering whether there is some rule-of-thumb to translate the core > config listed in > https://urldefense.proofpoint.com/v2/url?u=https-3A__en.wikipedia.org_wiki_Epyc&d=DwIDAw&c=euGZstcaTDllvimEN8b7jXrwqOf-v5A_CdpgnVfiiMM&r=WRHOCjWNhD-hk2AQTjbVUqX9gELcVS7wxFXQqbJ02hk&m=tinTJJMaVo42Y9_sd_6GFqebHHPSHkAU7HWSxq9pJJk&s=4qtrnAVSz-nrQqws1E9H4JixtZQpj0dg3dtPszJpW1g&e= > to the node information Slurm expects in "Sockets=x CoresPerSocket=y"? > ("ThreadsPerCore=2" is clear.) > > We'll be getting Epyc 7313 and 7513 machines, and perhaps add a single 7713 > one - "lscpu" outputs are wildly different, while the total number of cores > is correct. > > Will I have to wait until the machines have arrived, and do some experiments, > or did someone already retrieve the right numbers, and is willing to share? > > Thanks, > Steffen -- Steffen Grunewald, Cluster Administrator Max Planck Institute for Gravitational Physics (Albert Einstein Institute) Am Mühlenberg 1 * D-14476 Potsdam-Golm * Germany ~~~ Fon: +49-331-567 7274 Mail: steffen.grunewald(at)aei.mpg.de ~~~
