[[MPC Group]]/[[並列計算機]] *Appro GPU Cluster [#r3403299] Hostname: pcc-gpu.jaist.ac.jp -System 9nodes / 144 cpu cores and 18 gpus -Node Spec --CPU ---AMD opteron 6136 (Mangny-Cours 8Core) * 2sockets --Memory ---Memory 32GB DDR3 --GPU ---Nvidia Tesla M2050 * 2 *Queue CLASS [#n74c4f5e] |~CLASS|~NODE/CORE|~TIME|~EXECUTABLE JOB|~EXECUTABLE JOB/USR|~Default Node/core| |SINGLE|1node/16core,2GPUS|168hours|n/a|n/a|1node/1core| |SMALL|2nodes/32core,4GPUS|96hours|n/a|6jobs/usr|1node/1core| |MEDIUM|4node/64core,8GPUS|48hours|6jobs|3jobs/usr|4nodes/64cores| |LARGE|9node/144core,18GPUS|24hours|2jobs|1job/usr|9nodes/144cores| *SAMPLE OF PBS [#ib2d057b] ""CUDA PROGRAMING **CUDA CASE [#p9e5042d] #!/bin/bash #PBS -j oe #PBS -l select=1:ncpus=8:ngpus=2 #PBS -q SINGLE #PBS -N JobName cd $PBS_O_WORKDIR NPROCS=`cat $PBS_NODEFILE | wc -l ` ./JobName