On Wednesday, 16 May 2012 at 09:26:45 UTC, Tiberiu Gal wrote:
himany claim their code solves the problem in order of ms ( c/pascal/haskell code)
I used the blockwise parallel sieve described here, and measured nice speed-ups as described in his blog. It completes calculations within an L1 cache block before moving on. Perhaps you can redesign you code to do the same...
http://create.stephan-brumme.com/eratosthenes/
