Basics

Hi All,

I am new to OpenCL and I have few doubts

  1. i used clGetDeviceIDs to get the device IDs of type CL_DEVICE_TYPE_CPU
    but did not get any IDs. I tried running on different PCs still got the same result.
    How can i run my OpenCL kernals on my multicore CPU?
  2. Can we call a Kernal inside a Kernal?
  1. i used clGetDeviceIDs to get the device IDs of type CL_DEVICE_TYPE_CPU
    but did not get any IDs. I tried running on different PCs still got the same result.
    How can i run my OpenCL kernals on my multicore CPU?

Install AMD’s OpenCL SDK.

  1. Can we call a Kernel inside a Kernel?

Yes, you can call a kernel function inside a kernel function. The behavior is the same as calling any other function.

Install AMD’s OpenCL SDK.

My CPU is Intel i5 and other pc is Intel i7 will the AMD’s SDK runs for Intel also.

Thanks for the quick reply

My CPU is Intel i5 and other pc is Intel i7. Will the AMD’s SDK runs for Intel also?

Yes, it will work on any x86/x64 processors.

Hi All

i have few more queries

  1. how to achieve task parallelization? in the specification it is mentioned that Opencl supports task parallelization please provide any links or documents.

  2. how can we optimese a multiply and accumulate operation or just accumulation
    I just tried somthing and i was not getting proper values

  3. How to use image structure supported by opencl? any sample code will be helpful

4)I there a limit on how many kernels we can write?

Regards

OpenCL stands for Open Computing Language, and you can read more about it here OpenCL Overview - The Khronos Group Inc and herehttp://en.wikipedia.org/wiki/OpenCL

Hi,

Does opencl have any limitation on the amount of code that can go on a kernel?

Regards

Does opencl have any limitation on the amount of code that can go on a kernel?

In theory you are only limited by the amount of memory available in your system. In practice I imagine that for very large kernels you may hit other implementation-specific limits.