HPC - The way all computing will look...

Thinking about HPC Infrastructure

Port 25 — Fri, 01 Dec 2006 22:21:18 GMT

I started the first HPC blog with an understanding that HPC is an area where there has been a surge of activity from a development/investment standpoint. So after getting a basic understanding behind the importance of why HPC matters, the next logical

HPC - The way all computing will look...

Search and High Performance Computing — Wed, 08 Nov 2006 13:08:02 GMT

So I have blatantly just taken that headline from a post on Port25 by Michael Freedman a Distinguished

re: HPC - The way all computing will look...

farhan — Tue, 07 Nov 2006 21:28:52 GMT

I guess this is the price we pay for making applications "portable" and not tied to a specific architecture or infrastructure. Like you said, highly specialized applications that are not subject to architectural changes on a regular basis (even the 3 year OS release cycle can prove to be costly) are best suited for such environments.

re: HPC - The way all computing will look...

kishi — Tue, 07 Nov 2006 19:54:26 GMT

Farhan, I do agree, the OS can only go up to a point, they key is in utlilizing the architecture to enhance the "processing experience" of the application. There are bleeding edge applications out there that do give some architechtures a run for their money but by-enlarge this is an area of great potential and very few apps actually do take advantage of the underlying architechture

re: HPC - The way all computing will look...

farhan — Tue, 07 Nov 2006 19:03:16 GMT

But won't that architectural firm or the accounting firm need to update their software to take full advantage of the parallel/multi-core computing infrastructure? A lot of software out there doesn't even take advantage of the multiple cores that are becoming mainstream now. The OS can only do so much.

I suppose the software industry and the HPC industry need to progress in tandem and keep in sync.

re: HPC - The way all computing will look...

hjanssen — Tue, 07 Nov 2006 16:06:21 GMT

I think HPC will fill a space in areas where people have not traditionally thought of HPC computing. It seems that the market for large HPC systems is not that large (with oodles and oodles of nodes) but the market for smaller ones has not really been tapped yet.

Would it not be cool for an small consulting or architectural firm to have an HPC system to do engineering on. Or an accounting firm that wants to run some scenario calculations. They will be able to do it for not that much money. With AMD and Intel about to release (in Intel's case they already have) quad core processors, it is becoming way cheaper for smaller companies to have 4, 8 or 16 single quad core machines set up as an HPC.

As to adoption statistics for Windows Server Computer Cluster Edition, I will let Kishi speak for that. He is doing the HPC stuff in our Labs.

Hank.

re: HPC - The way all computing will look...

farhan — Tue, 07 Nov 2006 07:30:04 GMT

I cannot help but think about the Sun Grid Compute Utility that Sun launched a few months ago. It came out with a big bang but has now silently slipped back into obscurity. Perhaps they over-estimated the demand for massive computing power in the sense that people who need it already have it (think research universities, investment banking industry, drug research corporations etc.). But it was an interesting service in any case, and it should be worthwhile to track its progress. (http://www.network.com).

I am also quite happy that Microsoft has finally ventured into the HPC area through its introduction of the Windows Server Computer Cluster edition. Does anyone have any statistics with regards to its adoption in the mainstream?