GPU Computing on Business
Overview
GPU Computing enables us to get Massively Parallel Processing power on a personal desktop computer. GPU Computing becomes popular among HPC recently. But in business area, it seems to be far less popular.
When it comes to academia, GPU Computing is something requires smart people to have hard trainings. There is no problem in academic industry because estimation of low investing cost encourages investments on GPU Computing. On the other hand, there are walls preventing the growth of GPU Computing in business. For instance, GPU Computing requires high-skilled engineers and in order to get them you must expect much cost for recruiting, training or outsourcing… After all, in many business aspects, GPU Computing costs too much to step in.
We introduce one of idea for this problem, “NtParallel”.
It wraps some difficult procedures using GPU Computing by CUDA, which enables us to use GPU Computing from any program language you like. You can use GPU Computing even if you are only familiar with Microsoft Excel or VBA. (Surely for some who are familiar with ruby, python, C# and so on, too.)
Of course it has some restrictions, but it may cover almost all your needs.
In the document you can download on this page, we showed some performance benchmarks by NtParallel. The result differs depending on target problems, however, acceleration was acknowledged from 5 times to 100 times.
We know there are many who hesitates to try GPU Computing because of its difficulty, but now you may feel like trying easy-to-use tool like NtParallel for starter. Once you have more interests in GPU Computing, of course you can start to learn CUDA!
We are
We are a cutting edge financial risk management software company as well as a consulting firm specialized in HPC. Financial risk management system requires huge calculation power so that our software support large, over 100 nodes, grid computing environment.
GPU Computing has possibility to get much faster processing speed, which leads us to solve larger scaled problems. We continue our research and development of GPU Computing as a key technology in next generation.
Following document is the handout provided at “GPU Computing 2010,” a seminar held by NVIDIA Corporation on July 16, 2010 in Tokyo.
GPU Computing on Business (.pdf) [3.9 MB]F.Y.I. “What is GPU Computing?” on NVIDIA website.