Is there is any company has super computers to run my intensive numerical codes using parallel computations

I have intensive computations to do it with matlab. We already have a cluster (32 worker), but still is not enough of waht are going to do.

I wonder if threre are any competent authorities about that?

Thnk you in advance

Answer 1

[John D'Errico]

As I recall there are services which can do things like that. A quick Google search gave me some hits.

http://www.sabalcore.com/?gclid=Cj0KCQiA28nfBRCDARIsANc5BFArQIrdxhG7z7uH4BTldIaHqSGaKj-gAE5O2nAlm7ktD77vCjnrWp8aAjZ-EALw_wcB

https://www.extremetech.com/computing/96829-rent-the-worlds-30th-fastest-30472-core-supercomputer-for-1279-per-hour

https://www.microsoft.com/en-us/research/blog/supercomputing-on-demand-with-microsoft-azure/

It looks like even companies like Amazon will do what you want.

https://aws.amazon.com/hpc/

Don't expect anything to be cheap.

Far better? Write better code. Use the profile tool to find the bottleneck(s) in your code. Or find better algorithms to do what you want to do. Almost always, when someone writes code that runs too slowly for them, the solution is to write better code, not to throw it at a super computer.

[Ahmed Abdalazeez]

Thank you for informations

and for advise as well :)

[John D'Errico]

Long ago, I wrote some code that was rather computationally intensive. I recall calculating that to solve a moderately sized problem would require hours of computing time on the fastest computer we could find. We even discussed whether renting super computer time was worth the investment. We never did that.

Instead, over the years computers got faster and more powerful, so we gained a lot that way. As important, and possibly of more importance was algorithm development. Once I learned to use better algorithms to solve the problem, I gained orders of magnitude in compute performance, to the extent that a super computer was never necessary at all. So faster computers will never hurt, and sometimes you really need them. But of more importance was just better algorithmic performance, thus writing better code on my part.