Developing a Plan for Cloud Based GPU Processing

For some applications, cloud based clusters may be limited due to communication and/or storage latency and speeds. With GPUs, however, these issue are not present because application running on cloud GPUs perform exactly the same as those in your local cluster -- unless the application span multiple nodes and are sensitive to MPI speeds. For those GPU applications that can work well in the cloud environment, a remote cloud may be an attractive option for both production and feasibility studies.

The post Developing a Plan for Cloud Based GPU Processing appeared first on insideHPC.