Is there any way to disable the video of the Nvidia video card and use it only for CUDA and use the video card’s onboard video in windows 7? I can put sleep the kernel for the video can be update, and after the kernel continue execute?
When i divide the kernel, i use a lot memory for vector in parameters, so i get the message “out of memory”. I try divide the kernel and the image volume, but i do many copys the host memory for device memory and vice versa, so this use a lot time that i think no is good for final time the application.
While i implements the kernels, the QR method for get eigensvalues and eigenvectors is reason for message “cuda the launch timed out and was terminated” when the loop have more than 4 iterations.