Need Help Putting Computer Together?

[Hellow]

Mathematical stats:
1280 x 8 = 10,240 total processing cores

That's a lot of processing power.

[RICHARD]

Mathematical stats:
1280 x 8 = 10,240 total processing cores

That's a lot of processing power.

Can do do the comp in layman's terms?


How many bags of M&M's and cans of Mountain Dew?

[blue]

Dont forget the Hot Pockets.

[kokopup]

While I'm sure they have developed a great cluster for doing campus computing task I am at a loss as to how this can be called a supercomputer that works as a one big machine. Unless these guys have developed some super optical data bus I don't know how they can harness all or those processors in a parallel fashion to harness the computing power of all those processors. All those processors need a lot of databus bandwidth

[Hellow]

A lot of hot pockets, that can be sure.

While I'm sure they have developed a great cluster for doing campus computing task I am at a loss as to how this can be called a supercomputer that works as a one big machine. Unless these guys have developed some super optical data bus I don't know how they can harness all or those processors in a parallel fashion to harness the computing power of all those processors. All those processors need a lot of databus bandwidth

Clustering software can split work up between the independent machines, making them all function as one big computer. Most of the software for Linux/*nix computers is free, search for "Open Source clustering software", you are sure to find a lot on that topic.

[Lady's Human]

Koko,

The USPS has a small distributed processing setup in every mail processing plant for image processing made up of 20-200 pizza boxes, each box being a quad core linux box with a master parceling out work based on response time and load monitoring. They communicate through a gigabit copper network.

Required result response time back to the originating sorting platform is 320 ms, and the system meets that mark 95-96% of the time. Our site is tiny, it's only set up to process about 150,000 images per hour. It's amazing what distributed processing can accomplish.

[kokopup]

I'm well aware of the work that can be accomplished by distributed computer systems and clusters. I worked for GE Engineering Services for 25 years and we had a computer system called the DMC ( distributed Micro Computer) We had Hundreds of these all sharing work over a reflective memory network with Vax computer clusters suppling man machine interface and recipes. By your definition these would be supercomputers. These would be capacity computers or quasi-supercomputers.

My problems is the use of the term Supercomputer to describe a cluster of independent computers preforming complex task. These are Quasi-Supercomputers and do not come close to the true computing power of a true supercomputer that is a Capability computer manufactured by the likes of Cray, IBM, or HP. I guess the word Supercomputer is a little over worked in todays world because even Google uses a quasi-supercomputer for their search engine. I may be a little sensitive about this subject because I have worked around a
true Supercomputer and the one described in this writeup does not meet my criteria. Compared to some of the computers I have worked with, the PC I am posting with is a Supercomputer.