|
From: Florian K. <br...@ac...> - 2012-06-02 04:04:07
|
>From 40 bytes to 32 bytes on LP64. From 20 to 16 bytes on ILP32. Same procedure as in previous patch for Qop. This time putting the Triop bits into a separate structure IRTriop. Here are some numbers showing the memory allocated by VEX in bytes for amd64/ppc64/s390x. The net is: it's always a win and the savings are pretty consistent 4-5% across platforms. amd64 before after bigcode1 1172040488 1114079872 -5% bigcode2 4428964792 4207718136 bz2 221793528 212074448 -4% fbench 109699680 105168664 -4% ffbench 100487672 96331128 -4% heap 82027160 78687008 -4% heap_pdb4 83288176 79893360 many-loss 88157280 84560800 many-xpts 79706304 76467064 sarp 76674560 73546416 tinycc 248763568 238678808 -4% ppc before after bigcode1 109679144 104873888 -4% bigcode2 109886664 105045848 bz2 276959560 263978936 -4% fbench 136637008 130500232 -4% ffbench 126413128 120634448 -4% heap 104295184 99522984 -4% heap_pdb4 105518000 100948824 many-loss 111422944 106495912 many-xpts 100370888 95973728 sarp 96301576 91938296 tinycc 305117568 292018544 -5% s390x before after bigcode1 878161120 843154104 -4% bigcode2 3256335832 3126573824 bz2 212416744 203591736 -4% fbench 113401536 108842224 -4% ffbench 98848776 94917216 -4% heap 81754568 78484056 -4% heap_pdb4 83213640 79884368 many-loss 88412768 84879128 many-xpts 78987296 75834040 sarp 75336264 72318352 tinycc 261218408 250868952 -4% Florian |