[mvapich-discuss] MVAPICH 2-GDR Issue with MPI_Allreduce

[Subramoni, Hari]

Hi All.

We followed up on this issue offline and have resolved the issue. The updated RPM is available for download from the MVAPICH download page.

Best,
Hari.

From: mvapich-discuss-bounces at cse.ohio-state.edu <mvapich-discuss-bounces at mailman.cse.ohio-state.edu> On Behalf Of Herten, Andreas
Sent: Wednesday, January 8, 2020 5:34 AM
To: mvapich-discuss at cse.ohio-state.edu <mvapich-discuss at mailman.cse.ohio-state.edu>
Cc: Alvarez, Damian <d.alvarez at fz-juelich.de>; Breuer, Thomas <t.breuer at fz-juelich.de>; Markus Schmitt <mschmitt at pks.mpg.de>; Hater, Thorsten <t.hater at fz-juelich.de>
Subject: [mvapich-discuss] MVAPICH 2-GDR Issue with MPI_Allreduce

Dear MV2 Support,

We see an issue when calling MPI_Allreduce on GPU memory buffers. The reduction produces wrong results or the program even seg faults.
I wrote up a more detailed description including a minimal reproducing example and some of our experiments here:
                https://gist.github.com/AndiH/b929b50b4c8d25137e0bfee25db63791<https://urldefense.com/v3/__https:/gist.github.com/AndiH/b929b50b4c8d25137e0bfee25db63791__;!!KGKeukY!mzGbEBbFzC6husDgfEvXunnb3UPLwL594G78RXrsJVo4LO7VTQv4egSuOcAN5Ckv8tRa5OTWYyKLHsI$>

Is this a bug?

New year’s greetings,

-Andreas
—
NVIDIA Application Lab
Jülich Supercomputing Centre
Forschungszentrum Jülich, Germany
+49 2461 61 1825

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20200123/49555838/attachment.html>