[petsc-users] How to confirm the performance of asynchronous computations

[Jed Brown]

Victor Eijkhout <eijkhout at tacc.utexas.edu> writes:

> On , 2021Jan28, at 10:35, Jed Brown <jed at jedbrown.org<mailto:jed at jedbrown.org>> wrote:
>
> Why plots like this are not _absolutely standard_ on all HPC sites' webpages is a source of continuing mystery to me.
>
> I've been asking for it for years.
>
> Curious. Why? What would it tell you? I mean, other than “42”.

Quite a bit about what sort of performance and variability can be expected for latency-sensitive jobs. Facilities publish HPL numbers despite far less relevance to application needs. People currently write proposals for Supercomputer X, and only find post-award that it delivers latency regressions compared to Humblecomputer Y and thus isn't suitable for the planned workload, so they fabricate a different workload to use the allocated hours.