Hi all, It looks like most of the nightly tests are passing from the 1.0.x branch. The OSU multi-threaded test is still failing, but I believe the plan is to leave this open for 1.1. Apart from these, here are a few more failures: ========================================================================== 1. threads/pt2pt/alltoall (why is there an all2all in pt2pt?) is failing in the following combination: --with-pm=smpd --with-pmi=smpd --with-device=ch3:sock CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-strict=posix 2. NAS MG is failing in the following combination: --with-pm=mpd --with-device=ch3:nemesis CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-nemesis-dbg-localoddeven ========================================================================== Is there any other bug we should try to fix for this release? You might see two more failures on breadboard with the Intel compilers, but you can ignore them -- we only have 10 licenses for MCS; so depending on who else is running nightly tests, we might not get a license. These failures are an artifact of the limited licenses. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji
Does the OSU multithreaded test fail with ch3:sock? Rajeev
-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Pavan Balaji Sent: Thursday, August 28, 2008 10:37 AM To: [email protected] Subject: [mpich2-core] mpich2-1.0.8 release status
Hi all,
It looks like most of the nightly tests are passing from the 1.0.x branch. The OSU multi-threaded test is still failing, but I believe the plan is to leave this open for 1.1. Apart from these, here are a few more failures:
============================================================== ============
1. threads/pt2pt/alltoall (why is there an all2all in pt2pt?) is failing in the following combination:
--with-pm=smpd --with-pmi=smpd --with-device=ch3:sock CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-strict=posix
2. NAS MG is failing in the following combination:
--with-pm=mpd --with-device=ch3:nemesis CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-nemesis-dbg-localoddeven
============================================================== ============
Is there any other bug we should try to fix for this release?
You might see two more failures on breadboard with the Intel compilers, but you can ignore them -- we only have 10 licenses for MCS; so depending on who else is running nightly tests, we might not get a license. These failures are an artifact of the limited licenses.
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
No, just nemesis. -- Pavan On 08/28/2008 10:38 AM, Rajeev Thakur wrote:
Does the OSU multithreaded test fail with ch3:sock?
Rajeev
-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Pavan Balaji Sent: Thursday, August 28, 2008 10:37 AM To: [email protected] Subject: [mpich2-core] mpich2-1.0.8 release status
Hi all,
It looks like most of the nightly tests are passing from the 1.0.x branch. The OSU multi-threaded test is still failing, but I believe the plan is to leave this open for 1.1. Apart from these, here are a few more failures:
============================================================== ============
1. threads/pt2pt/alltoall (why is there an all2all in pt2pt?) is failing in the following combination:
--with-pm=smpd --with-pmi=smpd --with-device=ch3:sock CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-strict=posix
2. NAS MG is failing in the following combination:
--with-pm=mpd --with-device=ch3:nemesis CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-nemesis-dbg-localoddeven
============================================================== ============
Is there any other bug we should try to fix for this release?
You might see two more failures on breadboard with the Intel compilers, but you can ignore them -- we only have 10 licenses for MCS; so depending on who else is running nightly tests, we might not get a license. These failures are an artifact of the limited licenses.
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
2. NAS MG is failing in the following combination:
--with-pm=mpd --with-device=ch3:nemesis CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-nemesis-dbg-localoddeven
I ran this test about 50 times each on 32-bit and 64-bit, and I can't reproduce this failure. But it seems to have showed up on the nightly tests even with revision 3049. I'll monitor this for one more round of the tests; if we don't see if the next time, I guess we'll need to let this be for the 1.0.8 release. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji
All nightly tests are passing now with the 1.0.x branch. We just need to decide on what bug requests we are going to handle for 1.0.8 and fix them up. Remember that we need to move to 1.1a2 soon as well; we should try to release that before SC|08. -- Pavan On Thu, 28 Aug 2008, Pavan Balaji wrote:
Hi all,
It looks like most of the nightly tests are passing from the 1.0.x branch. The OSU multi-threaded test is still failing, but I believe the plan is to leave this open for 1.1. Apart from these, here are a few more failures:
==========================================================================
1. threads/pt2pt/alltoall (why is there an all2all in pt2pt?) is failing in the following combination:
--with-pm=smpd --with-pmi=smpd --with-device=ch3:sock CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-strict=posix
2. NAS MG is failing in the following combination:
--with-pm=mpd --with-device=ch3:nemesis CC=gcc CXX=g++ F77=g77 F90=gfortran --enable-nemesis-dbg-localoddeven
==========================================================================
Is there any other bug we should try to fix for this release?
You might see two more failures on breadboard with the Intel compilers, but you can ignore them -- we only have 10 licenses for MCS; so depending on who else is running nightly tests, we might not get a license. These failures are an artifact of the limited licenses.
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
participants (2)
-
Pavan Balaji -
Rajeev Thakur