Author: Matt Taylor
Date: 23:37:54 12/21/02
Go up one level in this thread
On December 21, 2002 at 02:40:42, Matt Taylor wrote: >OK, after a bit of work cleaning up an undergrad project (the buggy source >highlighter that I used to convert the code -> html), I've got the code up. > >Timings: >http://65.186.75.165/~mtaylor/bitscan_athlon.html > >Code: >http://65.186.75.165/~mtaylor/ipcbench.c.html >http://65.186.75.165/~mtaylor/bitscan.c.html > >I'll run results for an original K6 tomorrow just as a matter of curiousity. I >don't have access to -ANY- Intel machines right now, so I can't get Intel data. >If anyone wants to help out, I can toss the binary up on FTP. Offhand I don't >think it will run terribly well on a P4 and possibly not very well on previous >Intel processors. > >-Matt Ok, I consolidated the latest source and results at: http://65.186.75.165/~mtaylor/bitscan/index.html I incorporated both Gerd's and Walter's suggestions for most of the tests; in some cases, the new tests were slower on my Athlon, so I tweaked them until they ran faster again. It's all accessible from that link. I also appologize for my color schemes. :-) I have timings for K6 and K7 posted now. A friend of mine ran the tests on his Celeron-2 (P6 core), but I forgot to re-enable some tests. As a result, the P6 data is not as complete as the rest of the data. I'll try to get him to re-run the benchmark tomorrow. I ran into some bizarre issues with Intel processors. I had a "null" function to measure the bias from the timing part of the code. I ended up disabling it because short functions would execute FASTER than the null function. I had one test where every bitscan function completed faster than the null function, and it was giving me negative timings. Anyway, I ended up disabling it because I can't figure out how ADDING instructions makes it faster. BTW, you can draw some interesting conclusions about K6 implementation. Very interesting how walking the bits takes progressively longer to bsf scan it. -Matt
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.