Author: Slater Wold
Date: 12:18:45 02/11/04
Go up one level in this thread
On February 11, 2004 at 14:51:42, enrico carrisco wrote: >On February 11, 2004 at 12:12:19, Slater Wold wrote: > >>On February 11, 2004 at 05:01:33, enrico carrisco wrote: >> >>>On February 11, 2004 at 02:17:46, Slater Wold wrote: >>> >>>>On February 11, 2004 at 02:14:41, enrico carrisco wrote: >>>> >>>>>On February 11, 2004 at 00:46:02, Slater Wold wrote: >>>>> >>>>>>On February 10, 2004 at 23:42:02, Russell Reagan wrote: >>>>>> >>>>>>>On February 10, 2004 at 13:00:34, Bryan Hofmann wrote: >>>>>>> >>>>>>>>Here are the results of the different compiled version of Crafty. Your's is just >>>>>>>>a hair (1.4%) faster then all of the others. >>>>>>> >>>>>>>Give these two a try and see how they do on your system. I compiled them with >>>>>>>MSVC .NET 2003. They were a hair faster than Aaron's on my Athlon 2400+ (after >>>>>>>hours of tinkering with compiler options). >>>>>>> >>>>>>>Without FUTILITY: >>>>>>>http://home.comcast.net/~r.reagan/crafty19.10.zip >>>>>>> >>>>>>>With FUTILITY: >>>>>>>http://home.comcast.net/~r.reagan/crafty19.10f.zip >>>>>> >>>>>>Aaron's: >>>>>> >>>>>> >>>>>>Crafty v19.10 >>>>>> >>>>>>White(1): bench >>>>>>Running benchmark. . . >>>>>>...... >>>>>>Total nodes: 102625951 >>>>>>Raw nodes per second: 1603530 >>>>>>Total elapsed time: 64 >>>>>>SMP time-to-ply measurement: 10.000000 >>>>>>White(1): >>>>>> >>>>>> >>>>>>Russell's: >>>>>> >>>>>>C:\crafty>crafty19.10f >>>>>>unable to open book file [./book.bin]. >>>>>>book is disabled >>>>>>found computer opening book file [./bookc.bin]. >>>>>> >>>>>>Crafty v19.10 >>>>>> >>>>>>White(1): bench >>>>>>Running benchmark. . . >>>>>>...... >>>>>>Total nodes: 89942714 >>>>>>Raw nodes per second: 1697032 >>>>>>Total elapsed time: 53 >>>>>>SMP time-to-ply measurement: 12.075472 >>>>> >>>>>This can't be the FX system.. What did you run this on? >>>>> >>>>>-elc. >>>> >>>>LOL - My FX51 @ 2.4Ghz... >>>> >>>>I promise. >>> >>>Did new additions to Crafty slow down its nps? Mid 19.xx's I was at 1.65m on >>>the XP @ 2.5GHz. I'll have to ask Aaron what his box runs his executeable at, >>>to compare with your results (for 19.10) >>> >>>-elc. >> >>We've already established that the FX51 @ 2.4Ghz is only 4% faster than his >>XP2.5Ghz. > >It was my understanding that when you were 4% faster you were still at stock or >near stock speed of the FX (2.2GHz). That might be correct. I don't honestly remember. This week has sucked. No sleep, almost at all... >In my tests -- Hiarcs has shown a 19-20% increase on AMD 64-bit even though it >has no optimizations. (This, of course, is clock for clock, vs. Athlon XPs.) > >Essentially, 2.2GHz FX/Opteron = 2.618GHz Athlon XP (at the minimum) for Hiarcs. > The tests I ran were on an Opteron 246 and the XP also at 2GHz. Crafty is not the biggest gainer. I believe that title would go to F7(no mmx). It gets over 2M nps, in the start position. (This compared to CCC archives.) DF8 gets only a ~30% speedup over a P4 3.0Ghz! >I'll have to get in touch with Aaron and see what his take is on all of this. I >can't imagine an unoptimized Hiarcs to be drastically different than an >unoptimized Crafty -- when comparing their 32-bit performance. Agreed. >-elc.
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.