[mvapich-discuss] mvapich compilation error with psm

[Cabral, Matias A]

Hi Christof,
> This is also set for the patched mvapich 2.1 version Intel diistributes together with the OmniPath driver tarball
Yes. 
Btw, we are in the process of updating to 2.2. 

_MAC

-----Original Message-----
From: mvapich-discuss [mailto:mvapich-discuss-bounces at cse.ohio-state.edu] On Behalf Of Christof Koehler
Sent: Thursday, December 01, 2016 1:06 AM
To: mvapich-discuss at cse.ohio-state.edu
Subject: Re: [mvapich-discuss] mvapich compilation error with psm

Hello,

> In summary, we have run some tests on the setup you mention (KNL with dual Omni-Path cards) and we have achieved nearly link speed by running different osu_ benchmarks for bandwidth and message rates measurements and using 4 ranks per node on a pair of nodes and not using any special build flags.
> ./configure  --without-mpe --disable-mcast --enable-shared 
> --with-device=ch3:psm --with-psm2 --with-ch3-rank-bits=32 
> --disable-maintainer-mode

I have a question regarding compilation options. In the line above "--disable-mcast" is set. This is also set for the patched mvapich 2.1 version Intel diistributes together with the OmniPath driver tarball.

However, running configure -h shows "--enable-mcast enable hardware multicast" which would indicate to me that "--disable-mcast" is the actual default.

The manual documents the "MV2_USE_MCAST" flag with a default value of "0", so it should be disabled in any case ?

I am confused. Is the "--disable-mcast" flag recommended or required ?
Does it have an effect at all ? 

I built our mvapich2 for psm2 without setting any flag regarding mulitcast. On the other hand we do not have KNL hardware.


Best Regards

Christof



-- 
Dr. rer. nat. Christof Köhler       email: c.koehler at bccms.uni-bremen.de
Universitaet Bremen/ BCCMS          phone:  +49-(0)421-218-62334
Am Fallturm 1/ TAB/ Raum 3.12       fax: +49-(0)421-218-62770
28359 Bremen  

PGP: http://www.bccms.uni-bremen.de/cms/people/c_koehler/