Author: Chris Carson
Date: 18:01:31 01/27/00
Go up one level in this thread
SSDF SSDF
Mp MHZ Date SI95 Rating Program
386 33 Apr-89 0.21 2126 Mchess
486 33 May-90 0.50 2240 Genius
486 66 Aug-92 1.00 2367 Genius
P5 90 Mar-94 2.74 2486 Tiger
P5 MMX 200 Jan-97 6.44 2575 CM
AMD 450 Jun-98 16.90 2696 Tiger
I am not sure this is what you wanted,
but here are the number #1 rated program
for each hardware on the SSDF list.
I have done a regression analysis with NPS
and PLY if you are interesetd. IMHO this
gives a good point of reference for what
good programmers can do as hw increases.
I think the DarkThought experiment will show the
effects of HW speed increase only. Here is
a similiar chart with experiments Thompson
did using BELLE (one of the Darkthought References):
Ply % diff move Estimated Estimated
from BELLE (ply-1) Rating Rating
R(4)=1320 R(4)=1300
R(5)=1570 R(5)=1570
------------------------------------------------
4 33.1 1320 1300
5 33.1 1570 1570
6 27.7 1779 1796
7 29.5 2002 2037
8 26.0 2198 2249
9 22.6 2369 2433
10 17.7 2503 2577
11 18.1 2639 2725
------------------------------------------------
Source: M. Newborn (1996), "Kasparov versus Deep Blue",
Springer, New York, New York.
IMHO, programs on the SSDF on AMD 450's are hitting in
the 11 ply range. Monty Newborn also states that
270 points between plys is more likley correct than
250 points, based on the 2 Thompson Experiments.
Please let me know if you would like to see my
expanded chart with estimated NPS, PLY, and SSDF
ratings for different HW (I have posted it once
earlier). :)
Best Regards,
Chris Carson
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.