Choose timezone
Your profile timezone:
US HPC Facilities Meeting
→
US/Central
Danila, David, Vakho, Wei, Xin, Doug, Taylor, Kaushik
Danila Titan IO: Still struggling, begun identifying the problem. Will send around slides to describe the problem.
Vakho commented on the large number of paths in LD_LIBRARY_PATH causes many failed file system accesses at start up. Merge step also peaks into each worker's input files causing a spike at the end of the run as well.
New US ATLAS HPC egroup: atlas-usa-hpc@cern.ch
Discussing software distribution solutions: Aim to present conclusions at September meeting at CERN. Wei is working on getting the BNL container system into production. Current containers running at NERSC have software inside, but the condb.cern.ch is on the shared filesystem still.
NERSC Edison is now running containers, Cori-KNL will likely be running containers by end of weekend.
Sergey will grab images from Wei and run those on Titan in singularity.
NERSC ERCAP allocation was increased by 40M. ALCC is at 2.4M and NERSC is asking if we want the remaining 12M hours. I was told that if we use up the ALCC early we can likely get more.
There are minutes attached to this event.
Show them.