We're looking at this as we've got mostly-CASTEP projects on the nationally-allocated portion of Sulis during the first allocation period.
Single node performance is a little ahead of ARCHER2 which @aturner-epcc suggests is likely due to turboboost.
Multinode performance is worse than ARCHER2 as might be expected.
I haven't attempted to muck around with the Jupyter notebook to plots these yet.
We're looking at this as we've got mostly-CASTEP projects on the nationally-allocated portion of Sulis during the first allocation period.
Single node performance is a little ahead of ARCHER2 which @aturner-epcc suggests is likely due to turboboost. Multinode performance is worse than ARCHER2 as might be expected.
I haven't attempted to muck around with the Jupyter notebook to plots these yet.