 |
Everybody likes benchmarks, so here they are. All simulators were compiled
with the same compiler and compiler options. The commandline was the same.
Needless to say, the results were the same too.
- Athlon-XP 2000+, Times in seconds, command line e.g. time ./exhaust -bkF 4000 -r 1000 fixed.rc fixed.rc
Compiler and options: gcc 3.3.2 -O2 -fomit-frame-pointer -march=athlon-xp
These options produce faster code than -O2 or -O3
pmars-0.9.2 exhaust-1.9 exhaust-ma W L T
Fixed vs Fixed 4,775 3,981 3,026 277 324 399
Jaguar vs Fixed 6,222 5,523 4,059 248 219 533
Jaguar vs Jaguar 8,382 7,795 5,427 125 143 732
Stalker vs Fixed 4,523 4,245 2,974 300 525 175
Stalker vs Jaguar 4,806 4,649 3,255 366 407 227
Stalker vs Stalker 3,850 3,759 2,630 470 458 72
nPaperII vs Fixed 7,102 5,976 4,634 215 173 612
nPaperII vs Jaguar 8,815 7,804 5,781 142 90 768
nPaperII vs Stalker 5,552 5,316 3,825 401 415 184
nPaperII vs nPaperII 9,672 8,142 5,868 86 94 820
----------------------------------------------------------------------
total: 63,699 57,190 41,479
relative time: 100% 89,8% 65,1%
faster than pmars: 0,0% 11,4% 53,6%
- AMD K6-2 400, Times in seconds, command line e.g. time ./exhaust -bkF 4000 -r 1000 fixed.rc fixed.rc
Compiler and options: gcc 2.95.4 -O -fomit-frame-pointer -march=k6
These options produce faster code than -O2 or -O3
pmars-0.9.2 exhaust-1.9 exhaust-ma W L T
Fixed vs Fixed 30,51 23,92 20,23 277 324 399
Jaguar vs Fixed 28,51 26,80 19,66 248 219 533
Jaguar vs Jaguar 32,05 32,94 23,79 125 143 732
Stalker vs Fixed 17,89 17,98 13,64 300 525 175
Stalker vs Jaguar 18,78 19,34 14,94 366 407 227
Stalker vs Stalker 14,83 15,23 12,05 470 458 72
nPaperII vs Fixed 42,90 34,11 27,93 215 173 612
nPaperII vs Jaguar 41,23 37,50 27,62 142 90 768
nPaperII vs Stalker 21,31 21,55 16,78 401 415 184
nPaperII vs nPaperII 51,48 42,70 32,75 86 94 820
----------------------------------------------------------------------
total: 299,53 272,07 209,86
relative time: 100% 90,8% 70,1%
faster than pmars: 0,0% 10,1% 42,7%
|
|
.