0

My computer has an ATI graphics card, but I need to code an algorithm I already have in CUDA, to accerelate the process. Is that even possible? If yes does anyone have any link or tutorial from setting up my IDE to coding a simple image processing or passing an image. I also considered OpenCL but I have not found any information how to do anything with it.

DogDog
  • 4,820
  • 12
  • 44
  • 66
  • you can use the device emulator in the CUDA SDK to run it on your CPU or you can take a look at: http://code.google.com/p/gpuocelot/ if you are on linux – PeterT Dec 02 '11 at 02:43
  • 2
    There is an experimental windows build of gpuocelot. See http://code.google.com/p/gpuocelot/wiki/Installation#Windows_Build_%28Experimental%29 – flipchart Dec 02 '11 at 05:50
  • 2
    @PeterT Correct me if I'm wrong, but for as far as I'm aware there is no longer a device emulator in the SDK (since a 3.x version of CUDA I think). – Bart Dec 02 '11 at 09:32

2 Answers2

4

This answer is more directed toward the part

I also considered OpenCL but I have not found any information how to do anything with it.

Check on this NVIDIA site: http://developer.nvidia.com/nvidia-gpu-computing-documentation

Scroll down and you find

  1. OpenCL Programming Guide

    This is a detailed programming guide for OpenCL developers.

  2. OpenCL Best Practices Guide

    This is a manual to help developers obtain the best performance from OpenCL.

  3. OpenCL Overview for the CUDA Architecture

    This whitepaper summarizes the guidelines for how to choose the best implementations for NVIDIA GPUs.

  4. OpenCL Implementation Notes

    This document describes the "Implementation Defined" behavior for the NVIDIA OpenCL implementation as required by the OpenCL specification Version: 1.0. The implementation defined behavior is referenced below in the order of it's reference in the OpenCL specification and is grouped by the section number for the specification.

On AMD/ATI you have this site for a brief introduction: http://www.amd.com/us/products/technologies/stream-technology/opencl/pages/opencl-intro.aspx

And for more resources check: http://www.amd.com/us/products/technologies/stream-technology/Pages/training-resources.aspx

Unless CUDA is a requirement you should consider OpenCL again as you can you use it on both platforms and you state have one and want to develop for the other.

jmsu
  • 2,033
  • 21
  • 17
1

You might also want to take a look at these:

http://blogs.nvidia.com/2011/06/cuda-now-available-for-multiple-x86-processors/

http://www.pgroup.com/resources/cuda-x86.htm

I haven't tried it myself, but the prospect of running CUDA code on x86 seems pretty attractive.

Vlad
  • 18,195
  • 4
  • 41
  • 71