Is there any way to use GPU memory directly ?!

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi everyone!

I’m currently working with image processing!

I have to do the follow steps when using CUDA:

  1. Allocate memory and Load image (in RAM)
  2. Using “cudaMalloc” and “cudaMemcpy” to allocate memory and copy that image into GPU memory
  3. Using GPU to process that image
  4. Using “cudaMemcpy” to copy that image backward into RAM
  5. Using “SetDIBitsToDevice” (Windows API) to display image on screen!

My question is:
Is there any way to display the image directly (without step 4 and 5)
(Can i use OpenGL or DirectX for this)

Thank!!!

2

Binding device memory to OpenGL or DirectX textures is possible. See the corresponding SDK samples