Author: Aaron Gordon
Date: 12:55:37 07/04/03
Go up one level in this thread
On July 04, 2003 at 12:57:54, Slater Wold wrote: >On July 04, 2003 at 12:52:31, Brian Richardson wrote: > >>>>Opterons are *much* faster than Athlons. >>>> >>>>My 2x1.73s got about 1.7M nps. >>> >>> >>>Opps! >>> >>>From the exe offered on Hyatt's website. >>> >>> >>>EPD Kit revision date: 1996.04.21 >>>found computer opening book file [./bookc.bin]. >>> >>>Crafty v18.15 (1 cpus) >>> >>>White(1): mt 2 >>>max threads set to 2 >>>White(1): bench >>>Running benchmark. . . >>>...... >>>Total nodes: 106580575 >>>Raw nodes per second: 1440278 >>>Total elapsed time: 74 >>>SMP time-to-ply measurement: 8.648649 >>>White(1): end >>>execution complete. >>> >>> >>>1.73M with a profiled exe using Intel's compiler. >>> >>>I'd be interested to see what he gets with the FTP exe of 18.15. >> >>FTP binary, no profile optimization, I think: >> >>hash table memory = 384M bytes. >>pawn hash table memory = 24M bytes. >>5 piece tablebase files found >>20370kb of RAM used for TB indices and decompression tables >>max threads set to 2 >> >>Crafty v18.15 (2 cpus) >> >>White(1): bench >>Running benchmark. . . >>...... >>Total nodes: 86094416 >>Raw nodes per second: 1721888 >>Total elapsed time: 50 >>SMP time-to-ply measurement: 12.800000 >>White(1): mt=1 >>max threads set to 1 >>White(1): thread 1 exiting >>bench >>Running benchmark. . . >>...... >>Total nodes: 84497760 >>Raw nodes per second: 1069591 >>Total elapsed time: 79 >>SMP time-to-ply measurement: 8.101266 >>White(1): > > >So here we see that an Opteron @ 1.7Ghz is a *lot* faster than an Athlon 1.7Ghz >using the same code. Looks like the Opteron excels at 32-bit code, Bob. ;) > >To the tune of about 20%. My old Athlon 1900+ (1.6GHz) @ 1.8GHz got about 1.19 million, Slate... I hope Hyatt will fix up the SMP stuff in Crafty soon too. Seems Craftys speedup gets worse when you throw faster & faster duals at it. Crafty 19.3 Dual Athlon 2.4GHz: (1.54x speedup, c'mon Hyatt.. fix it please:) Total nodes: 74470910 Raw nodes per second: 2256694 Total elapsed time: 33 SMP time-to-ply measurement: 19.393939 Dual Athlon 2.4 w/ 1 thread Total nodes: 62767026 Raw nodes per second: 1459698 Total elapsed time: 43 SMP time-to-ply measurement: 14.883721 Single Athlon XP 2.5GHz Total nodes: 62767019 Raw nodes per second: 1609410 Total elapsed time: 39 SMP time-to-ply measurement: 16.410256 Single Athlon XP 1.7GHz (1711.57MHz, 213.95fsb (427.89DDR)) Total nodes: 62767091 Raw nodes per second: 1101177 Total elapsed time: 57 SMP time-to-ply measurement: 11.228070 These runs were with my optimized binary. This is available at: ftp://newageoc.com/pub/crafty/c193smp-k7sse.zip Brian contacted me about me making an Opteron dll via PGO. When he gets the .dyn files back to me I'll recompile it and have a 32bit PGO version for the Opteron on the FTP. It should beat my 1.7GHz Athlon XP scores. Once everything is optimized in 64bit mode the Opteron should absolutely smoke it. :)
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.