Saturday, July 21, 2007
Blade and HPC (High Performance Computing)
What has been happening in the world of HPC?
A list of the Top Ten Highlights in HPC for 2007 Jan-Jul is Posted Here for a quick look.
Having added a Live and RSS feed button in Firefox lets me view tech/news headlines from hundreds of feeds.
Great plug-in for Firefox.
The plug-in lead me to great new Podcast about HPC called: Meat Cast.
The webpage is at Semi-Coherent Computing on The Register.
The first podcast of Meat Cast had an interview with Chris Hipp, co-founder and CTO of blade server pioneer RLX Technologies..
My simplified explanation of blades is the ubiquitous PCI card of the PC world.
The 1U box compressed the parts of a server into a stackable short box. Need more power? Stack more boxes. The Blade Idea turned the 1U boxes on it's side, compressed them a bit more and plug them into a server box that had multiple slots for power and I/O connections.
The big win for this new form-factor was more computing power for the same space, at a greatly reduced power outlay to power the units.
But, of course, the racks keep getting bigger and bigger.
Power, and cooling seem to be the big road blocks to Supercomputer design.
Some Supercomputers use more POWER than is consumed by some small states.
That is dreadful, and so the door is opening for Green-HPC designs.
The Top 500 Supercomputer List:
(Some quick highlights:)
Clusters remain the most common architecture in the Top500 list. The list has 373 clusters, representing just under three-quarters of all systems. HP Cluster Platform 3000BL and 4000BL systems, based on HP BladeSystem c-Class servers, accounted for 152 entries on the list. HP has the highest number of systems on the list with 40.6 per cent, IBM now has 38.4 per cent, and Dell holds 4.8 per cent, with no other manufacturer having more than five per cent of the systems listed.
I discuss the problems with using GPU's as secondary CPU's and the software snag in utilizing virtualization (the Next Frontier) in HPC and Supercomputers.