On Feb 22, 2010, at 10:37 , David MacMahon wrote: > On Feb 22, 2010, at 6:42 , Andrew Ross wrote: > >> I would like to have a thorough comparison of the time difference. >> This should include a "large data" case as well where timings >> might be >> more important. The lena image might be one suitable case. > > As I said earlier, I share this concern. My 100x test was done using > the Lena image, but I agree that further analysis would be reassuring > (or not as the case may be).
I have done some further performance testing using x20c (the test using the Lena image). The results are slightly different, but still very encouraging. I think my previous x20c loop tests were not a valid comparison because it turns out I was looping on a call to plimage, which I had not yet changed to call the plf2ops version named "plfimge". I had created plfimagefr, but I had not created plfimage. So I think my big x20c loop test was not really exercising new code. I have since created plfimage (patch forthcoming) and modified x20c.c to loop 2000 times on each of the two calls to plimage that plot only Lena's eyes (one call plots only the eyes in their normal size/ position, the other call plots them "zoomed in"). This "subset" plotting involves copying parts of the 2D data array so it definitely makes use of the new plf2ops code. I ran the modified x20c test three times using an unpatched libplplot with NO plf2ops stuff added and three times using a patched libplplot with the plf2ops stuff added. Here are the results from my (now old) MacBook Pro laptop... WithOUT plf2ops --------------- 2m 22.1s 2m 21.8s 2m 21.9s WITH plf2ops ------------ 2m 23.2s 2m 23.1s 2m 23.2s So the plf2ops version took about 1.2 seconds longer than the non- plf2ops version or approximately a 1.5% performance hit. It would be great is someone else could repeat this test and corroborate (or not) the results. Dave ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
