Greetings, First, please accept my apologies for the three-month delay in getting back to you. I will try to be much more responsive in the future.
I downloaded the SLURM patch, and it looks pretty invasive. In particular, the README says: "This patch makes use of SLURM's srun command to launch all tasks." Will this continue to work well with existing mpich-based applications? Also: "You must explicitly enable shared memory when building MPICH with the following configure line: ./configure --with-device=ch_p4 --with-comm=shared" The Debian mpich package includes the ch_p4, ch_p4mpd and ch_shmem transports, for clusters via rsh, clusters via mpd, and shared memory. Why does SLURM want to re-configure ch_p4 for shared memory? This sounds like it could break a lot of cluster applications. If you can address these concerns and help me understand how SLURM support will not cause regressions, I will apply the patch to mpich. Thank you, Adam Powell On Tue, 2007-10-16 at 13:20 +0200, [EMAIL PROTECTED] wrote: > Package: mpich > Version: 1.2.7-4 > > It would be desirable to add SLURM support to mpich. SLURM stands for > Simple Linux Utility for Resource Management, it is an open-source > cluster resource management and job scheduling system. SLURM recently > entered the Debian system as slurm-llnl package. > To add support to MPICH you will need the patch available here. > http://sourceforge.net/projects/slurm/ > Thanks, -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Engineering consulting with open source tools http://www.opennovation.com/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
