Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] 1.5.5rc1 tested: REGRESSION on FreeBSD
From: Brice Goglin (Brice.Goglin_at_[hidden])
Date: 2011-12-14 01:53:44

Le 14/12/2011 07:17, Paul H. Hargrove a écrit :
> My OpenBSD and NetBSD testers have the same behavior, but now I see
> that I was at warned...
> On all the affected systems I found the following (modulo the system
> tuple) in the configure output:
>> checking which OS support to include... Unsupported!
>> (x86_64-unknown-openbsd5.0)
>> configure: WARNING:
>> ***********************************************************
>> configure: WARNING: *** hwloc does not support this system.
>> configure: WARNING: *** hwloc will *attempt* to build (but it may not
>> work).
>> configure: WARNING: *** hwloc run-time results may be reduced to
>> showing just one processor.
>> configure: WARNING: *** You have been warned.
>> configure: WARNING: *** Pausing to give you time to read this message...
>> configure: WARNING:
>> ***********************************************************
> Clearly my failures are "known" to somebody.
> However, I have multiple "issues" with the current behavior.
> 1) At an minimum the WARNING mention --without-hwloc
> 2) Is this "build and pray" approach to unknown platforms really wise?
> 3) Shouldn't something appear in the README about this? The
> --without-hwloc option doesn't even appear in README.

Maybe the wording is a bit too strong, but you can ignore this. What
happens in the vast majority of cases like this is that hwloc will only
know how many processors the system has, and hwloc won't be able to bind
tasks to processors. But that's still not worse than disabling hwloc.