what a mess if it is a function it's huuge did you try split it to useful chunks ? it's just unreadable and that's why you cann't spot anything i guess..
пятница, 9 ноября 2012 г., 0:48:20 UTC+4 пользователь Cedric Greevey написал: > > I have the following code to perform a complicated image convolution. It > takes 10-15 seconds with output dimensions 256x256 and samples 6. No > reflection warnings, and using unchecked math doesn't speed it up any. I've > tried to ensure it uses primitive math inside the loops, aside from > generating the outer loop's values. What cached-load-chunk does shouldn't > matter much, but in most cases it should boil down to a map lookup inside a > swap! and a couple of atom derefs and function calls. The bottleneck is > likely in the math somewhere, and likely something is being boxed, though > I've primitive-coerced every numerical let and loop value and avoided more > than two arguments per arithmetic op. > > Can anyone spot anything I haven't that could be causing boxed arithmetic > inside the loops? > > (let [^BufferedImage out (BufferedImage. width height > BufferedImage/TYPE_INT_BGR) > half-w (double (/ width 2.0)) > half-h (double (/ height 2.0)) > lnk (double (- (* (Math/log 0.1) (inc l10-mag)) > (/ (Math/log (+ (* half-w half-w) (* half-h > half-h))) 2))) > cconv (double (/ chunk-width (* 2.0 Math/PI))) > c-base-ln (double (* chunk-base-mag (Math/log 0.1))) > x-offset (double (* chunk-width (+ 0.5 (/ deg-rot 360)))) > fmtstr (str "%0" chunk-name-digits "d") > cached-load-chunk (cached-objects > (fn [chunk-num] > (maybe-load-image > (str chunk-name-base > (format fmtstr (inc chunk-num)) > ".png"))))] > (doseq [oy (range height) ox (range width)] > (let [ox (int ox) > oy (int oy) > ix (double (- ox half-w)) > iy (double (- oy half-h)) > samples (double samples) > smo (int (dec samples))] > (loop [sx (int 0) sy (int 0) ns (double 0) > r (double 0) g (double 0) b (double 0)] > (if (== sy samples) > (.setRGB out ox oy (int (+ > (int (/ b ns)) > (+ (* 256 (int (/ g ns))) > (* 65536 (int (/ r ns))))))) > (let [nsx (int (if (== sx smo) 0 (inc sx))) > nsy (int (if (zero? nsx) (inc sy) sy)) > ix (double (+ ix (/ (+ sx (rand)) samples))) > iy (double (+ iy (/ (+ sy (rand)) samples)))] > (if (and (== ix 0) (== iy 0)) > (recur nsx nsy ns r g b) > (let [lnz-P (double (+ lnk (/ (Math/log (+ (* ix ix) (* iy > iy))) 2))) > argz-P (double (+ Math/PI (Math/atan2 iy ix))) > ; atan2 output range is -pi...pi and oy increases > down the screen > x (double (* argz-P cconv)) > y (double (* (- c-base-ln lnz-P) cconv)) > ; > chunk-num (int (/ y chunk-height)) > chunk-x (int (rem (int (+ x x-offset)) chunk-width)) > chunk-y (int (- y (* chunk-height chunk-num))) > ^BufferedImage chunk (cached-load-chunk chunk-num) > pixel (int (if chunk (.getRGB chunk chunk-x chunk-y) > 0)) > r (double (+ r (/ (bit-and pixel 0x00ff0000) 65536))) > g (double (+ g (/ (bit-and pixel 0x0000ff00) 256))) > b (double (+ b (bit-and pixel 0x000000ff)))] > (recur nsx nsy (inc ns) r g b)))))))) > out)) > > -- 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
