[OMPI devel] 1.4.5rc2 now released
Please test: http://www.open-mpi.org/software/ompi/v1.4/ -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
Re: [OMPI devel] 1.4.5rc2 now released
Is there a summary of fixes since rc1? Such a list would guide my retesting of the issues I reported with rc1. -Paul On 1/19/2012 9:55 AM, Jeff Squyres wrote: Please test: http://www.open-mpi.org/software/ompi/v1.4/ -- Paul H. Hargrove phhargr...@lbl.gov Future Technologies Group HPC Research Department Tel: +1-510-495-2352 Lawrence Berkeley National Laboratory Fax: +1-510-486-6900
Re: [OMPI devel] [OMPI svn-full] svn:open-mpi r25742
This is a critical change, with a significant impact on the code base. Basically by moving the binding later in the code after the modex was completed, all memory allocated before (which is all memory allocated during the registration of all OMPI modules) will endue being on the wrong NUMA node (at least by default on Linux). This can have tremendous impact on performance! george. On Jan 19, 2012, at 13:38 , r...@osl.iu.edu wrote: > Author: rhc > Date: 2012-01-19 13:38:58 EST (Thu, 19 Jan 2012) > New Revision: 25742 > URL: https://svn.open-mpi.org/trac/ompi/changeset/25742 > > Log: > Correct ordering in MPI_Init so that we do the modex prior to attempting to > bind ourselves in the direct launch case as the modex contains info required > for self-binding. > > Text files modified: > trunk/ompi/runtime/ompi_mpi_init.c | 238 > > 1 files changed, 119 insertions(+), 119 deletions(-)
Re: [OMPI devel] [OMPI svn-full] svn:open-mpi r25742
On Jan 19, 2012, at 11:58 AM, George Bosilca wrote: > This is a critical change, with a significant impact on the code base. > Basically by moving the binding later in the code after the modex was > completed, all memory allocated before (which is all memory allocated during > the registration of all OMPI modules) will endue being on the wrong NUMA node > (at least by default on Linux). This can have tremendous impact on > performance! Yes and no. First, any process launched via mpirun is already bound (both cpu and memory) prior to being exec'd, so this change has no impact on them. Second, any process direct-launched where the RM is setting the binding is also bound prior to being exec'd, so this change has no impact there. Thus, the change -only- impacts those processes that are direct-launched but want OMPI to bind them anyway. In that one case, this change is required because the proc has no way to self-determine its binding prior to doing the modex. Jeff/Nathan are adding language to btl.h to highlight the need to avoid allocating memory during registration - we informally have always stated that any allocation should only occur during add_procs, but people may have forgotten over time. Memory allocated by other frameworks should follow similar rules. > > george. > > On Jan 19, 2012, at 13:38 , r...@osl.iu.edu wrote: > >> Author: rhc >> Date: 2012-01-19 13:38:58 EST (Thu, 19 Jan 2012) >> New Revision: 25742 >> URL: https://svn.open-mpi.org/trac/ompi/changeset/25742 >> >> Log: >> Correct ordering in MPI_Init so that we do the modex prior to attempting to >> bind ourselves in the direct launch case as the modex contains info required >> for self-binding. >> >> Text files modified: >> trunk/ompi/runtime/ompi_mpi_init.c | 238 >> >> 1 files changed, 119 insertions(+), 119 deletions(-) > > > ___ > devel mailing list > de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/devel
[OMPI devel] 1.4.5rc2 test results
Here is a report on my findings with "configure && make all && make install && make check" with ompi-1.4.5rc2. All platforms built the "self", "sm" and "tcp" btls. I've noted where addition btl/mtl components were built. PASS with no special configure arguments (just --prefix and things like "CFLAGS=-m64" when appropriate). This means GNU compilers were used in all cases. linux/x86 [including gm1, gm2 and elan] linux/x86-64 [including gm2, elan, ofud, openib and psm] linux/ia64 (SGI Altix) linux/ppc32 linux/ppc64 macos-10.4/x86 macos-10.4/ppc macos-10.5/x86 macos-10.5/ppc macos-10.6/x86 solaris-11(snv_151a)/amd64 [including ofud, openib and dapl] freebsd-6.3-RELEASE/amd64 freebsd-7.2-RELEASE/amd64 freebsd-8.2-RELEASE/amd64 freebsd-8.2-RELEASE/i386 netbsd-5.0.2/i386 netbsd-5.1/amd64 netbsd-5.1/i386 PASS with addition of --disable-io-romio openbsd-5.0/amd64 openbsd-5.0/i386 This need do disable ROMIO is "known", having been reported against 1.4.5rc1 (see http://www.open-mpi.org/community/lists/devel/2011/12/10086.php ) as well as against 1.4.3rc1 in the past (see http://www.open-mpi.org/community/lists/devel/2010/08/8338.php ). FAIL macos-10.3/ppc Same problem as reported against 1.4.5rc1 in http://www.open-mpi.org/community/lists/devel/2011/12/10086.php I have additional Solaris platforms which I will report on later. -Paul -- Paul H. Hargrove phhargr...@lbl.gov Future Technologies Group HPC Research Department Tel: +1-510-495-2352 Lawrence Berkeley National Laboratory Fax: +1-510-486-6900
Re: [OMPI devel] 1.4.5rc2 test results
I neglected to mention that all FreeBSD and OpenBSD tests were conducted using GNU Make. If I use /usr/bin/make (BSD make), then I encounter the same problem I reported against ompi-1.5.5rc1: https://svn.open-mpi.org/trac/ompi/ticket/2954 Note that there is a patch (against 1.5.5rc1) attached to that ticket that may be sufficient to fix this. Oddly, NetBSD doesn't display the problem. I don't know if there is something different happening with respect to the atomics that simply avoids the problem, or if NetBSD's version of Make may share the GNU Make behavior with respect to VPATH. -Paul On 1/19/2012 1:33 PM, Paul H. Hargrove wrote: Here is a report on my findings with "configure && make all && make install && make check" with ompi-1.4.5rc2. All platforms built the "self", "sm" and "tcp" btls. I've noted where addition btl/mtl components were built. PASS with no special configure arguments (just --prefix and things like "CFLAGS=-m64" when appropriate). This means GNU compilers were used in all cases. linux/x86 [including gm1, gm2 and elan] linux/x86-64 [including gm2, elan, ofud, openib and psm] linux/ia64 (SGI Altix) linux/ppc32 linux/ppc64 macos-10.4/x86 macos-10.4/ppc macos-10.5/x86 macos-10.5/ppc macos-10.6/x86 solaris-11(snv_151a)/amd64 [including ofud, openib and dapl] freebsd-6.3-RELEASE/amd64 freebsd-7.2-RELEASE/amd64 freebsd-8.2-RELEASE/amd64 freebsd-8.2-RELEASE/i386 netbsd-5.0.2/i386 netbsd-5.1/amd64 netbsd-5.1/i386 PASS with addition of --disable-io-romio openbsd-5.0/amd64 openbsd-5.0/i386 This need do disable ROMIO is "known", having been reported against 1.4.5rc1 (see http://www.open-mpi.org/community/lists/devel/2011/12/10086.php ) as well as against 1.4.3rc1 in the past (see http://www.open-mpi.org/community/lists/devel/2010/08/8338.php ). FAIL macos-10.3/ppc Same problem as reported against 1.4.5rc1 in http://www.open-mpi.org/community/lists/devel/2011/12/10086.php I have additional Solaris platforms which I will report on later. -Paul -- Paul H. Hargrove phhargr...@lbl.gov Future Technologies Group HPC Research Department Tel: +1-510-495-2352 Lawrence Berkeley National Laboratory Fax: +1-510-486-6900
Re: [OMPI devel] 1.4.5rc2 now released
Minor documentation nit, which might apply to the 1.5 branch as well (didn't check). README says: - Open MPI does not support the Sparc v8 CPU target, which is the default on Sun Solaris. The v8plus (32 bit) or v9 (64 bit) targets must be used to build Open MPI on Solaris. This can be done by including a flag in CFLAGS, CXXFLAGS, FFLAGS, and FCFLAGS, -xarch=v8plus for the Sun compilers, -mcpu=v9 for GCC. However, following that instruction w/ Sun Studio 12 Update 2 yields: cc: Warning: -xarch=v8plus is deprecated, use -m32 -xarch=sparc instead for every single compilation. I vaguely recall noting this once before, perhaps 2 years or so. Additionally, it appears that the "Sun" example is for the 32-bit ABI and the "GCC" example for the 64-bit ABI. This is potentially a bit confusing. -Paul On 1/19/2012 9:55 AM, Jeff Squyres wrote: Please test: http://www.open-mpi.org/software/ompi/v1.4/ -- Paul H. Hargrove phhargr...@lbl.gov Future Technologies Group HPC Research Department Tel: +1-510-495-2352 Lawrence Berkeley National Laboratory Fax: +1-510-486-6900
Re: [OMPI devel] 1.4.5rc2 now released
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 20/01/12 04:55, Jeff Squyres wrote: > Please test: Great - we can now silence that warning for NFS, thanks! - -- Christopher Samuel - Senior Systems Administrator VLSCI - Victorian Life Sciences Computation Initiative Email: sam...@unimelb.edu.au Phone: +61 (0)3 903 55545 http://www.vlsci.unimelb.edu.au/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk8YzF8ACgkQO2KABBYQAh/nZACgipxqx8FesM3hN4HO6Qoo+Oag pQEAnAuUMkyk1twDwxbaSJQFTX5Sb5Dy =Y6j9 -END PGP SIGNATURE-
