
Hi all, I'm trying to do MergerTree for 10 snapshots on Nautilus. Each of the snapshots has around 500,000,000 particles. I tried #PBS -l ncpus=256 #PBS -l mem=1024GB #PBS -l walltime=12:00:00 but it ended 11 minutes later with the error: MPI WARNING: Could not allocate an internal buffer in the last 30 seconds on rank 184. Try increasing MPI_BUFS_PER_PROC and/or MPI_BUFS_PER_HOST. Could someone please give me an estimation about how much memory I need to use? Thanks! Pengfei

Hi Pengfei, I see those warnings often when running big yt jobs on Nautilus. You can get rid of them by setting those environment variables to something like: export MPI_BUFS_PER_PROC=1024 export MPI_BUFS_PER_HOST=1024 That should get rid of those warnings, but may not totally solve the problem of the job ending prematurely, so let us know if you're still having problems. Britton On Wed, Dec 5, 2012 at 12:40 AM, Pengfei Chen <madcpf@gmail.com> wrote:
Hi all,
I'm trying to do MergerTree for 10 snapshots on Nautilus. Each of the snapshots has around 500,000,000 particles. I tried
#PBS -l ncpus=256 #PBS -l mem=1024GB #PBS -l walltime=12:00:00
but it ended 11 minutes later with the error:
MPI WARNING: Could not allocate an internal buffer in the last 30 seconds on rank 184. Try increasing MPI_BUFS_PER_PROC and/or MPI_BUFS_PER_HOST.
Could someone please give me an estimation about how much memory I need to use?
Thanks!
Pengfei
_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
participants (2)
-
Britton Smith
-
Pengfei Chen