Author: Kerry McDermott
Date: 20:14:03 08/16/02
Go up one level in this thread
On August 16, 2002 at 01:12:21, Matthew Hull wrote: >On August 16, 2002 at 00:13:19, Kerry McDermott wrote: > >>I've read in old SSDF rating lists that Psion Atari is rated about 2080 USCF and >>CM2100 Amiga is about 1900 USCF. If these are versions of Mephisto Amsterdam >>and Fidelity Mach II LA respectively(as I have read), then why aren't their >>ratings closer? The Mach II LA and the Amsterdam are nearly identically >>rated... both around 2125 USCF. Why are the ratings of the programs on the >>personal computer platforms nearly 200 pts. apart? > >First of all, the old (and I mean really OLD) USCF ratings for these machines >were derived against relatively few _human_ opponents a long time ago. Back >then, most club players didn't know how to play against computers and their >known weaknesses. So I think their ratings tended to be overstated. Since >then, we know that these old machines are not as strong as their "official" USCF >ratings indicate. > >Secondly, if you remember even longer ago, the old Novag Super Constellation was >known for its human-like style of play and it tended to do better against humans >than other computers. In other words, computers can have different ratings >versus humans than they would against other computers. SSDF is strictly >computer-vs-computer ratings. That's one reason why the Mach II LA is so >different vis-a-vis the Amsterdam in SSDF than in USCF. > >I have a Mach II LA, and from my experience, it plays just under expert (2000). >And in truth, it never really was competitive with the Amsterdam. > >> >>Is it safe to assume that Psion Atari = Psion Macintosh in rating? >>Is it safe to assume that CM2100 Amiga = CM2100 Macintosh in rating? >>(All three use the 6800 chip) >> >>I know that CM2100 is nearly identical to Sargon IV on Macintosh (through >>various little tests and reading I've done in old CCR's) only a little slower. >>Can I assume that Sargon IV is a little higher than CM2100... say 1925? > >Also remember that with software, it is possible to have different hardware >running it. For these programs, you can have 68020, 030, 040 or even PPC >running in 68K emulation. The software will have different performance on each, >in addition to which, more RAM might be available for hashing. On a Mach II LA, >the hash table was fixed at 1024k. (I think it was the first dedicated machine >to use hash tables.) Software versions might have been able to adjust this >amount up or down based on installed hardware on the Mac or Atari. This could >affect playing strength. > >This is one reason why some disregard comp-vs-comp testing, while others think >its the best measure. In the old days, the programmers would gun for their >commercial competition so that their machine would seem like the stronger, when >in fact it was weaker against human competition. This probably continues today >to some extent, which is another reason why some take SSDF ratings with caution. > >> >>Help! >> >>Thanx, >>Kerry Thanx for the info. I realize the difference in rating for different processors (68000 vs. 68020 etc.). I am running all programs on a 68000 Macintosh SE (5-10% faster than original Mac Plus) and theoretically the same as the Amiga and the Atari. Ultimately the question for me is... did they port the programs well or not? If they are identical then I have no question really, and I'll just go w/ the ratings on SSDF list with corrections for USCF. Also I have noticed something about Sargon IV. It simply does not recognize when it will have a piece trapped (at least at 15 seconds per move). An example would be: having a bishop trapped and cornered on b6 by pawns at a3, b4 and c5. In a recent 10 game match with CM5500 playing 3ply vs. Sargon IV at 15 seconds a move, Sargon lost the match .5 - 9.5! It just couldn't keep up with the capture extensions and the stupid trapped piece flaw in the program. Does the Mach II LA also have the same trapped piece flaw? Why do you believe that it is around 2000 instead of 2125ish as SSDF indicates (with USCF correction of course)? Incidentally in that same match Psion is leading CM5500 3ply by a score of 4-2. I realize that 15 seconds per move is not exactly 40moves/2hrs, however it should give some idea of the relative strength.... shouldn't 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.