Author: John R. Menke, Sr.
Date: 06:55:08 12/20/99
Go up one level in this thread
Hi, Simple question, complex answer, because it varies considerably on my machine -- an upgraded 80486 (AMD5x86 133MHz), originally 8MB RAM upgraded to 64MB RAM, originally with no L2 cache now upgraded to 256K. The AMD5x86 133MHz is reportedly equivalent to a Pentium 75, but my McAfee Nuts & Bolts benchmark gives it an equivalency of about a Pentium 69Mz (if there were such a beast). To complicate things further, I only have a 0.5G hard drive, and CM6000 and CM7000 are now installed on Zip disks, not on my hard drive. Further the Zip drive is external on my parallel port, which may run at normal speed or a special accelerated loading speed if I tweak the chipset. Today with CM6000 I got 2 samples at 2424 with 32MB hash, and two samples at 2426 with 1MB hash. At the same time with CM7000 I got 2 samples at 2431 with 32MB hash and 2 samples at 2431 with 1MB hash. Tweaking the chipset to give accelerated loading from the Zip disks had no affect. So whatever the benchmark is based on, it is apparently not based on loading speed. Just a few days ago, when I had these installed on my hard drive, I recall a CM rating of 2501. I can't remember if it was CM6000 or CM7000. Further, in the past with CM6000 installed on my hard drive, I believe I have seen somewhat higher ratings, but can't recall the exact numbers. However usually the rating comes in somewhere in the 2400-2499 range, if I recall correctly. Why is it a variable, not a constant? I'm not sure, but believe it may have something to do with how the program gets paged into RAM. I find similar or greater variance in speed with Hiarcs 7.32 and Fritz 5.32. If I reset my system just before loading them, my chances for a better speed seem improved. I can't believe I said all that just to get a point on a graph!? Cheers... --JRM
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.