Super User is a question and answer site for computer enthusiasts and power users. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

What is needed to make a GPU fully fed by CPU?

Comparing their flops/s is enough?

For example, if i could manage to make a very old(pentium-3) CPU with one of Nvidia-Fermi GPU, it would not be able to fed the gpu with data per sec. What is the criteria to fit CPU to GPU exactly when OpenCL or some similar work needed?

Of course RAM and BUS will be chosen in a similar way but how exactly?

Assume each GPU-core will calculate a sqrt and a division and an adding for 100 times for every itearation.


share|improve this question

closed as not constructive by studiohack Sep 15 '12 at 19:10

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.If this question can be reworded to fit the rules in the help center, please edit the question.

up vote 2 down vote accepted

As Lucas Kauffman noted, it depends on your application.

If you need to feed GPU big amount of data from hard drive without any pre/post-processing on CPU, then CPU is nearly irrelevant. If before/after/during processing data on GPU it needs heavily processing on CPU, then CPU might become the limiting factor.

Several rules-of-thumb:

  • Your CPU should have at least N+1 cores, where N is number of GPUs.

  • You should have at least as much RAM as have all your GPUs in sum.

  • Your PCI-Express bus should be the latest version supported by your GPU (note, that on some motherboard with e.g. 2 PCI-Express slots each might function at x16 speed, but when used together, each is limited to x8). After all, PCI-Express is most common bottleneck for GPUs.

  • If you need video output from your box, consider buying some cheap card for this sole purpose. To avoid any complications with drivers, stick to the same vendor as of your number-crunching cards.

share|improve this answer
Just use PS3's. Japan used I believe 1000+ for a super computer – Cole Johnson Sep 15 '12 at 18:02

There is not one single answer, there are a lot more factors involved as just feeding information will also depend on what the task at hand will be doing.

But just in short, preferably multiple cores with a high clock-rate.

share|improve this answer
I just want to fit different pieces as exactly as possible. Would an openCL library of C++ make true decisions when i use opencl-capable CPU + GPU? – huseyin tugrul buyukisik Sep 15 '12 at 16:56

Not the answer you're looking for? Browse other questions tagged .