My company is testing / comparing various grid + cloud alternatives to do scientific calculation.
I read David Chappell ’s interesting HPC / Azure white paper , which is different from the perspective of “concepts,” but since it’s a fairly new and cutting edge, it was difficult for me to find a complete “dumb guide” for setting up an HPC server by putting nodes in the clouds, with the multi-threaded program "Hello HPC-Cloud World".
I read / tried to try a couple of days, and I have to admit that this is a dense forest, so I'm a little lost. We cannot afford to set aside a resource for an entire month just to give HPC a chance. If anyone can help with this, it will save valuable time for the first testers and adventurers like us :-)
Peace.
NB: A premonition of comments / questions requiring further guidance here: What I am ideally looking for is a guide to getting started from scratch on how to mount the HPC server in place, allocate nodes on Azure Nodes (or some other Windows virtual machines Ideally, Server R2) (which is the cloud part), and deploy the dummy hi-world .Net application, in which there are hundreds of threads that will be assigned to different nodes (this is part of the grid)
Edit 1: There is a very good online HPC lab offered by MS. Unfortunately, he continued to lament and so on, so we find a simple "paper" guide.
Edit 2: Listen ... are you here this sound ??? Listen closer ... This is the sound of a blowing wind in the desert. "Feeeeeeeeeeeeeeeeeeewwwwwwwwwwwwwww" ... See what Tumbleweed rides there? This is my sad and lonely question, awaiting an answer to someone :-)
source
share