Beowulf Cluster configuration?
My fluid department wishes to buy a Beowulf Cluster with 32 or 64 nodes to use it for solving large computations problems in OpenFOAM and/or Fluent. They want to have a linux platform for this server. My questions are the following:
Want is the best configuration for CPU, RAM etc. (for solving the CFD problems)?
It is possible to use also the new technology of Tesla Nvidia GPU for this server with linux (Ubuntu server)?
from my experience the most important point will be the connection between your nodes, you will need something faster (with less latency) than Ethernet, e.g. Infiniband...
|All times are GMT -4. The time now is 02:27.|