Fluent6.0.12 : How to increase the speed of reading the casefile on a Windows cluster?
Two customers complain about the low speed of file i/o in the parallel Fluent solver on a Windows cluster. Is there any way to increase the speed with given system configuration? Decreasing the chunk size to 16 kBytes has reduced the reading time of a given case from 8 hours to 20 Minutes. The scheme command is the following: (mp-set-tcp-chunk-size 16) The hardware was a 4-PC cluster running W2000 with 100 Mb/s network and Intel 82562 ET network cards. |
||
![]()
|