[mvapich-discuss] Registration cache problem

[Subramoni, Hari]

Thanks for confirming Alex. Please let us know if it works with the real programs.

I have taken in this patch into the MVAPICH2 code base with an acknowledgement to you. It will be available with the next release of MVAPICH2.

Best,
Hari.

From: Alexander Melnikov <alex.i.melnikov at gmail.com>
Sent: Friday, July 3, 2020 7:08 AM
To: Subramoni, Hari <subramoni.1 at osu.edu>
Cc: mvapich-discuss at cse.ohio-state.edu <mvapich-discuss at mailman.cse.ohio-state.edu>
Subject: Re: Registration cache problem

Many thanks, Hari. For some reason, two lines in the patch were truncated (PRINT_DEBUG in flush_delayed_dregs). However, it looks like your patch is working. We will check further on real programs.
Best regards, Alexander Melnikov.

чт, 2 июл. 2020 г. в 23:20, Subramoni, Hari <subramoni.1 at osu.edu<mailto:subramoni.1 at osu.edu>>:
Hi, Alexander.

We looked at the issue. Your patch does solve the problem. However, we thought the attached patch may be a safer way. We tried it locally and the application seems to run fine. Can you please try it out and see if it solves your issue?

Best,
Hari.

From: alex.i.melnikov at gmail.com<mailto:alex.i.melnikov at gmail.com> <alex.i.melnikov at gmail.com<mailto:alex.i.melnikov at gmail.com>>
Sent: Monday, June 29, 2020 10:36 AM
To: mvapich-discuss at cse.ohio-state.edu<mailto:mvapich-discuss at cse.ohio-state.edu> <mvapich-discuss at mailman.cse.ohio-state.edu<mailto:mvapich-discuss at mailman.cse.ohio-state.edu>>
Cc: Subramoni, Hari <subramoni.1 at osu.edu<mailto:subramoni.1 at osu.edu>>
Subject: Registration cache problem

Perhaps the problem is with the delayed registration. A small patch (in the attachment) solved the problem for me, but I'm not sure of its correctness.
Best regards, Alexander Melnikov.
From: Subramoni, Hari<mailto:subramoni.1 at osu.edu>
Sent: Monday, June 29, 2020 4:46 PM
To: Alexander Melnikov<mailto:alex.i.melnikov at gmail.com> ; mailto:mvapich-discuss at mailman.cse.ohio-state.edu
Cc: Subramoni, Hari<mailto:subramoni.1 at osu.edu>
Subject: RE: [mvapich-discuss] Registration cache problem

Hi, Alexander.

Thanks for reporting the issue. We appreciate it.

We are able to reproduce it. We are looking into it and will get back to you.

Best,
Hari.

From: mvapich-discuss-bounces at cse.ohio-state.edu<mailto:mvapich-discuss-bounces at cse.ohio-state.edu> <mvapich-discuss-bounces at mailman.cse.ohio-state.edu<mailto:mvapich-discuss-bounces at mailman.cse.ohio-state.edu>> On Behalf Of Alexander Melnikov
Sent: Monday, June 29, 2020 1:07 AM
To: mvapich-discuss at cse.ohio-state.edu<mailto:mvapich-discuss at cse.ohio-state.edu> <mvapich-discuss at mailman.cse.ohio-state.edu<mailto:mvapich-discuss at mailman.cse.ohio-state.edu>>
Subject: [mvapich-discuss] Registration cache problem

Hey. Can anyone check out a little test (in the attachment)? On each node, one MPI process starts, a small number of nodes is enough (3-6). We used the MVAPICH2 library versions 2.3.3 and 2.3.4, the compiler - GCC 4.8.5.
Almost all tests fail (bad data in rcv_thread). However, when the registration cache is disabled (MV2_USE_LAZY_MEM_UNREGISTER = 0), the error disappears.


________________________________
[Image removed by sender. Avast logo]<https://urldefense.com/v3/__https:/www.avast.com/antivirus__;!!KGKeukY!haenv9wShyyLNADsgxEeUuE3Fgde1jCoJR_kolJIMLbcwE_H614A43EBreqORyR-WQ$>

Это сообщение проверено на вирусы антивирусом Avast.
www.avast.com<https://urldefense.com/v3/__https:/www.avast.com/antivirus__;!!KGKeukY!haenv9wShyyLNADsgxEeUuE3Fgde1jCoJR_kolJIMLbcwE_H614A43EBreqORyR-WQ$>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20200716/c9354e7a/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ~WRD000.jpg
Type: image/jpeg
Size: 823 bytes
Desc: ~WRD000.jpg
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20200716/c9354e7a/attachment.jpg>