On Sun, 2010-01-17 at 16:06 +0100, gabor papp wrote: > hi Dave, > > > It's a long time since I looked at this code, but could it test if all > > the indices were the same for a vertex and skip the processing? > good idea! i did this and the loading time dropped to 1.4 seconds. the > file size increased from 1.2mb to 4.8mb, so now only basically line > processing and tokenization take time. with a 3.8mb obj file the loading > time was about 30 seconds, with the optimized version it went down to 16 > seconds, and now it's about 4.3 seconds if all the indices are the same > (file size increased to 15megs).
How does this compare to the speed that (for example) blender takes to load the same file? It's pretty quick isn't it, from what I remember? Do you think we are doing something wrong... cheers, dave
