Experiments with Null Hypothesis Generation
July 7th, 2009
Since the Null Hypothesis generation is the bottleneck currently, and specifically the generation of random numbers, I’ve tried a few things. First off, I changed the kernel that uses the random numbers so that it is based off probability. Secondly, I moved the probability calculation off the CPU (and onto the GPU). Lastly, I’ve tried generating random numbers on the GPU using cudpp and on the cpu using SFMT. The SFMT appears to be just slightly faster by tenths of a second.
Null hypothesis generation is still the bottleneck!
Also, I fixed the bad p-value output bug from yesterday. Turned out I had to seed the cudpp random generator every useage.
Categories: SURF