Author: Djordje Vidanovic
Date: 17:51:27 05/09/98
Go up one level in this thread
>On May 09, 1998 at 14:16:31, Ernst A. Heinz wrote: >>After updating my "old" Fritz 5.00 with the CBM63 version (a.k.a >>Fritz 5.03) both short and long FritzMark results on my AMD >>K6/233MHz with 48MB hash increased by roughly 20%. I now get >>142 short FritzMarks and 180 long FritzMarks vs. 125 short and >>150 long before. The speed in nodes per second did not shoot up, >>however -- it even dropped slightly. On May 09, 1998 at 19:07:58, Moritz Berger wrote: >[This has been posted here before. The LCT solution times might be more >reliable than the Fritzmark scores, the Fritz 5.01 engine ought to be >the same you get in the CBM63 update, so maybe you could post some >scores for your AMD to check the correlation LCT II / FritzMark? I have >used only 12MB hash tables in case your machine doesn't have 128MB RAM >...] > >Fritz 5 gets 241000 Nps in the long Fritzmark benchmark on my P233MMX >(with 98304KB hash tables). That translates to a long/FM of 202. > > > Prog Fritz 5.01 (fritz5.eng size: 16496 bytes) > CPU P233MMX > Hash 12288KB Didn't have enough time to check on LCTII, but would like to report that the update does enhance the performance on FM on the AMD -- on my K6/233 with 48MBs of hash the short FM is now 142 (217 Kns) (used to be 128), and the long FM 183 (242 Kns)(used to be 151) which is more than the P233MMX in terms of nodes, but less in terms of FMs (??) Now, how does one interpret this? The engines (be it Fritz 5.03 or 5.01) seem to be the same -- 106496 bytes, date March 14, 1998. Djordje
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.