Author: Paul Doire
Date: 18:57:37 12/10/00
Go up one level in this thread
On December 10, 2000 at 20:20:40, Jason Williamson wrote: >On December 10, 2000 at 19:42:15, Paul Doire wrote: > >>On December 10, 2000 at 18:27:43, Jason Williamson wrote: >> >>>On December 10, 2000 at 18:00:20, Paul Doire wrote: >>> >>>>The utility I used for this was Binary Work CPU Controller 1.0 >>>>it is available at www.tweakbase.com,as I mentioned in my previous post. >>>>My hardware is Celeron 541(overclocked 433), 128MB RAM, Win98SE. >>>>*This test was to see if CM6000 and CM8000 CPU control could be limited >>>> to allow fair, same PC matches.I tested it against Gambit Tiger 1.0 in this >>>> instance* >>>> >>>>Settings Gambit Tiger1.0-NoBook-Brain ON-1 MB Hash >>>> Set At 12 ply Fixed Depth >>>> CM8000-NoOBK-PonderOn-Default 1MB Hash >>>> Infinite Time >>>> CM6000-NoOBK-PonderOn-Default 1MB Hash >>>> Infinite Time >>>>Objective-Note time Gambit Tiger takes to solve 12 ply running alone,>>> then with CM8000 in background,then CM6000 in background. >>>> >>>>All tests were run from inside utility! >>>> >>>>Gambit set to "normal" alone time= 44.5 seconds >>>>CM8000 set to normal in background= 52.8 seconds >>>>CM6000 set to normal in background= 50.5 seconds >>>> >>>>Gambit set to "normal" alone time= 44.5 seconds >>>>CM8000 set to "minimum"in background= 46.5 seconds >>>>CM6000 set to minimum in background= 45.5 seconds >>>> >>>>Gambit set to "high" alone time= 44.5 seconds >>>>CM8000 set to minimum in background= 45 seconds >>>>CM6000 set to minimum in background= 44.5 seconds >>>> >>>>Conclusion: Using this utility it is possible to stop CM's >>>>influence on Gambit's thinking. >>>>I will play some matches to verify results. >>>> >>>>Regards,Paul >>> >>>Chessmaster 8000 doesn't use CPU outside of its thinking. That problem is with >>>CM6000 and CM7000. >> >>Based on these results IMHO it does ,and it clearly shows in this >>test suite. >>Cordially Yours, >>Paul > >I run windows 2000, and I have my task manager running when I do engine tests. >The Chessmaster.exe (which is the gui, as the engine is THe King.exe) ONLY uses >cpu power during a screen update, that is if it has move or such. This lasts >less then a second, and doesn't go above 10% of one of my 2 cpus. > >JW I understand what you are saying,but both CM's influence GT1's solve time when they run in the background running the same test.As in a real game. I.E. when running no ponder there is minimal impact on GT1 thinking time, but as soon as you set ponder on,such as in a real game CM8000's influence on GT1's thinking jumps to higher than CM6000's. Also if CM8000 isn't using cpu time outside of its thinking,why such profound differences in GT1's thinking? I believe this utility has clearly indicated the effect of CM6K & CM8K on GT1 and I will be posting 4 pgn's using the settings that equalize GT1's thinking time.When GT can solve the test at the same time running alone and then at the same time with another engine running,then IMHO it is equal ground for a match. Regards, Paul
This page took 0 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.