There have been many new products launched in the personal supercomputing space over the past couple years. There is HP’s Cluster Platform Workgroup system from Nov 2007, NVIDIA’s personal supercomputer specification (offered by a variety of vendors) from Nov 2008, Cray’s CX-1 from Sep 2008, SGI’s Scalable Workgroup Cluster from Aug of this year, and probably one or two others that I’m missing. And then of course there were the deskside units that SiCortex had started shipping before they closed up shop.
So, do you have one of these? I hear from the vendors about how great these products are, and each of them has a case study or two extolling the virtues of their machine. But I’d like to hear from users who have these things in the real world about what uses you are finding for them. What works? What doesn’t work? Are you using them as a primary workhorse, or are they development platforms for your larger machines?
If you like, send me an email with your experiences, or leave a comment. If I get enough response, I’d like to put together a feature article on real world experiences from real folks, just like you.
Hi,
I’ve been using CUDA and GTXs/Tesla for the last year. The speedup over our CPU
code (a few similar algorithms) is ~x60. Its been the most challenging stuff I’ve ever done (in my 13 years of experience) and for sure the most fun (altough frustrating at the same time 😉 )
I’m now working on converting another algorithm to CUDA and it seems promissing as well.
I’ve been using both Linux and Windows with C++ and Cuda 2.0 through 2.3.
Let me know whether I can offer some more information.
Eyal.
eyalhir74@gmail.com