Jerome Soumagne

I think it should be feasible to simply default to whatever OFI returns and just have a warning printed in that case with some information so that we have a...

Hi Scott, I need to investigate some more but yes it seems to happen after a specific size. I need to do some more comparisons to see if this is...

Sorry for not responding earlier, that's interesting I like the idea. Although that's still 2 passes :) Yes the proc component needs to be improved to get rid of the...

We've always kept windows support as a possibility, without really committing to supporting it as we've never really had any user who asked for it. It's not regularly tested but...

Yes I had looked at that in the past but it was not supported. See https://github.com/ofiwg/libfabric/issues/3062 We should have a look at it again and maybe have a query call...

@carns All providers set mr_iov_limit and rma_iov_limit to 1, I am still doing some more testing and we'll support it internally now but I would keep hopes very low at...

@carns I was looking in more details at the mpich code that you had sent. I think it's still limited though since they also always set iov_count and rma_iov_count to...

Supporting code was added as part of https://github.com/mercury-hpc/mercury/commit/a6bbce8514011e21e5ef1b37283a17606a0d39cb but there are issues remaining with libfabric that will need to be further investigated.

I would need to look at it again but I had added support for it, you can turn it on by uncommenting that line: https://github.com/mercury-hpc/mercury/blob/master/src/na/na_ofi.c#L196 I think if I recall...

Ah ok from what I see it looks like it's because of the rxm provider. When you're running fi_info, are you specifying verbs;ofi_rxm ? or only verbs?