Page 11 of 11

Optimized math functions

Posted: October 30th, 2010, 2:24 pm
by Cuchulainn
QuoteOriginally posted by: outrunQuoteOriginally posted by: CuchulainnQuoteOriginally posted by: outrunYes I'm going to use threads as well, because I have 8 cores in my machine that I want to utilize. I'll look up the gcc equivalent, thats my platformSo, loop parallelism? This will give almost linear speedup, if there are no dependencies.Yes, thats what I thought. SSE can work on 4 functions in parallel, and the concept is easily adoptable to GPU's with hundreds of kernels..Something like this!