Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] 1.4.5rc3 released
From: Paul H. Hargrove (PHHargrove_at_[hidden])
Date: 2012-01-30 19:25:46

In retesting I see that the following items I reported in 1.4.5rc2 or
rc1 have NOT been resolved in rc3.
This list is meant to be complete, even for issues that I reported so
recently that no fix is expected and those that probably won't be fixed
at all.

0) README for Solaris Studio compilers ("don't set CXX=sunCC") which I
mentioned in my previous post.

1) The 1.4.x version of ticket #2954 has not been fixed yet.
In other words, one must still use GNU Make on FreeBSD and OpenBSD.
I saw George committed the fix to 1.5.x, but Did the CMRs ever get
created for HEAD and 1.4?

2) Must --disable-io-romio on OpenBSD

The README says "OS X (10.4)" has been tested, but doesn't state 10.4 as
a minimum requirement.
I've noted that 1.5.x is smart enough to enforce 10.4 as a minimum at
configure time:
> configure: WARNING: Open MPI does not support OS X prior to version
> 10.4 (Tiger)
> configure: error: Cannot continue

4) Icc-9.1 and icc-10.0 fail to build atomics on IA64 (gcc is fine on
same machine)

5) atomic_spinlock test failure on linux/ppc64/xlc-9.0 (ppc32 PASSes, as
does ppc64 w/ xlc-8.0 and xlc-11.1).
If this is not a OMPI bug, then it may be worth noting in the README.

6) linux/xlc-7.0 can't build atomics (to document in README)


On 1/30/2012 2:25 PM, Jeff Squyres wrote:
> A few little fixes incorporated; posted in the usual location:
> The shared library versions still need to be updated before final release.
> ----
> Combined patch from Fujitsu. Fixes a collections of typos over the
> code and man pages.
> Modify Solaris Studio Complier notes about using xarch
> Fixes #2967: pretty much copy the text from MPI-2.2 in the description
> of the group argument to MPI_COMM_CREATE.
> Fixes #2844: ensure to take the value of --with(out)-memory-manager
> into account when configuring the components of the faramework. If
> --without-memory-manager was given, then we really don't want any
> memory managers to be used.
> NEWS and README updated

Paul H. Hargrove                          PHHargrove_at_[hidden]
Future Technologies Group
HPC Research Department                   Tel: +1-510-495-2352
Lawrence Berkeley National Laboratory     Fax: +1-510-486-6900