Author: enrico carrisco
Date: 00:23:02 09/30/04
Go up one level in this thread
On September 29, 2004 at 15:41:43, Aaron Gordon wrote: >All engines used 64mb hash (except CT14/15, which used 48mb). I used the >starting position and did infinite analysis to a specific depth and recorded NPS >at the end of the finished ply. Both cpus are exactly identical in design except >the addition of the extra 256k L2 on the Barton core. Everything else on the >system was the same between test runs. > >The system: >Athlon XP 2400MHz, 200fsb (400DDR) 256K L2 Thoroughbred-B core >Athlon XP 2400MHz, 200fsb (400DDR) 512K L2 Barton core >Motherboard: Epox 8RDA / Nforce2 chipset >Ram: Corsair PC2700 XMS running PC3200 @ 3-4-4-8 > >ChessTiger 14.0: >256k L2: 496.69307kn/s >512k L2: 501.66kn/s >Benefit: 0.099% > >Chesstiger 15.0: >256k L2: 503.23611kn/s >512k L2: 510.32394kn/s >Benefit: 1.408% > >Deep Fritz 8: >256k L2: 1103.5846kn/s >512k L2: 1120.8281kn/s >Benefit: 1.562% > >Deep Junior 8: >256k L2: 1843.9153kn/s >512k L2: 1843.9153kn/s >Benefit: 0% > >Hiarcs 9: >256k L2: 269.144kn/s >512k L2: 269.144kn/s >Benefit: 0% > >Shredder 8: >256k L2: 409.28049 >512k L2: 424.82278 >Benefit: 3.797% > >Crafty 19.10: >256k L2: 1425360 nodes/second >512k L2: 1445435 nodes/second >Benefit: 1.408% Hiarcs fits inside the 256kb cache, obviously. I am surprised that DF8, CT and S8 do not, however... -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.