Hi Per, woh, take it easy. I don't claim to be an expert. Thanks for showing me that though. It certainly didn't seem right at first, but I had trouble figuring out the laziness in clojure, me being new to it.
Anyway, have a good weekend! Carson On Jul 17, 10:02 pm, Per Vognsen <per.vogn...@gmail.com> wrote: > How about some common sense? The convolution algorithm everyone is > discussing runs in O(n m) time. So your measured time would imply that > a single of your computer's core is capable of 10 * 3,000,000^2 = 9 * > 10^14 FLOPS, or about one peta-FLOPS. That should have tipped you off > you aren't actually measuring anything useful. > > Wrap your (convolve dxs dys) calls with (doall ...). And you will want > to reduce the three million number to something much smaller, or > you'll be waiting for a very long time. > > By the way, no-one actually uses this quadratic-time algorithm for > very wide kernels. For that you use FFTs and pointwise multiplication > (look up the convolution theorem), which yields an O(n log n) > algorithm. > > -Per -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
