Testing Cuda

Hi everybody,

I am searching for a code that implements simple algorithmes for testing GPU Vs CPU performance.
I went through the GPU SDK and studied the matrix multiplication code, but it seems to work for small matrices.
Does anybody have a code , or do you know how to modify the matrixmul code (on the sdk) in order to be able to multiply large matrices and to implement a cpu timer?

Thanks for your help!


Have you tried to change BLOCK_SIZE in the header file?