Has any one use the parallelization capability of Mathematica with multiple cores and/or CUDA Compliant GPUs and/or TPUs?
I would like to know before we invest big $ in more hardware.
The Barkley Group
I use multiple cores all the time. The speedup is impressive.
Combining Mathematica and CUDA works, but most of our GPU programming was directly with CUDA and C++. The speedup is enormous, however, one needs to specialize in CUDA and be familiar with the specific problem and how to represent the solution efficiently for the GPU. Mathematica, in this case, will serve as a "glue" language (if the problem is GPU intensive).
I'm not aware of TPU's support, if you mean Google's TPU, but I suspect one of Wolfram employees can answer this much better than me.
I expect to buy a computer with intel i7 Processor which has 9 cores. In addition, I will add a GPU with almost 1000 cores. The difficulty is that memory management for my problem is ugly. However, Mathematica interfaces with a CUDA compliant GPU and does the memory management for you behind the scenes.
Will let you know how it pans out after I get the project.
Any update on this? I have the same question, about a good hardware set-up.