Joost VandeVondele

@vdbergh this is probably an issue now. The performance for workers will differ much more between workers, because hardware differences can't be (easily) corrected. The residual calculation will always differ...

@Vizvezdenec btw, setting prio to 1 for your test made it pass.

yes, that's a workaround till we have something better in place.

yes, I agree changing the default for auto-purging from true to false would be the best right now, but we should consider the chi^2 per arch.

I've pinged him on discord. But there are at least two other workers with a lot of losses.

yes, I agree that this could be done at a higher level like you describe.

BTW, let me plug an advertisement for a framework I wrote a while ago, which allows for methods in the nevergrad suite of optimization methods to be used. Right now...

This worker is probably running in a bug in latest clang. https://github.com/llvm/llvm-project/issues/55377 I see his workers are based on clang 15.

for whatever reason the TLS might be handled differently or the code aligned properly by luck, depending on the OS.

hard to guess, would need a more detailed error message.