All the presentations from the SC07 tutorial are available at:
This looks interesting. Thanks for posting this!
BTW, URL typo on the last talk, which should be http://www.gpgpu.org/sc2007/SC07_CUDA_8_Oi…Deschizeaux.pdf
Thanks, I fixed the error.
I think there is an mistake in SC07_CUDA_5_Optimization_Harris.pdf on slide 19.
the index should be
int index = blockIdx.x * blockDim.x * 3 + threadIdx.x;
int index = blockIdx.x * blockDim.x + threadIdx.x;
since we access an array of type float afterwards.
or do I get something wrong?
I think you’re correct, nice catch. That was my error in the code, perf is still the same. We’ll see about correcting the slide.
I’ve posted a corrected PDF. Thanks!