Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is Sapphire 2 playing low Master Strength at 40/2hrs?

Author: Terry Ripple

Date: 13:49:19 08/13/99

Go up one level in this thread


On August 13, 1999 at 16:27:01, James Robertson wrote:

>On August 13, 1999 at 14:23:47, Terry Ripple wrote:
>
>>On August 13, 1999 at 10:56:21, James Robertson wrote:
>>
>>>On August 13, 1999 at 03:18:37, Miguel A. Guzman wrote:
>>>
>>>>Sapphire II plays at the FICS. This are its records:
>>>>
>>>>Statistics for SapphireII(C)    On for: 5 hrs, 20 mins, 55 secs   Idle: 37 mins
>>>>
>>>>          rating     RD      win    loss    draw   total   best
>>>>Blitz      1926     25.7    3035    1110     324    4469   2210 (06-Oct-98)
>>>>Standard   2053     40.9    1330     383     134    1847   2215 (02-Mar-99)
>>>>Lightning  2014     40.9     561     294      52     907   2287 (06-Apr-99)
>>>>
>>>>You be the judge. ;)
>>>
>>>Oops, missed your post. I have never played SapphireII, but after watching many
>>>games it has played, it seems very vulnerable to the 'computer-playing human'
>>>rated 1750 and up you find so often on FICS. My personal opinion is that it is
>>>not master strength, at least not against the FICS crowd.
>>>
>>>James
>>
>>Hi James,
>>  The figures seem artificial because Larry Kaufmann's review praises it to be
>>about a solid 2250 USCF at 40/2hrs, which come from his personal tests and-
>>analysis of it's games played. Like you said, there is alot of cheating on the
>>chess sites and i played several people who claimed their ratings were in the
>>1500 to 1600's but played more like strong expert at times and like masters
>>other times. Figure that one out!!!
>>  Some of the posts to this post give SapphireII the same strength as Larry did!
>>
>>Regards,Terry
>
>I have an incredibly hard time believing it to be anywhere near 2250 at 40/2. I
>will have to take all of Kaufmann's reviews with a grain of salt....
>
>James

----------------------------------------------------------
Hello again James, and interested parties of SapphireII
 Here is some interesting reading material!

The original Novag Diamond went through a C.R.A. (Computer Rating Agency) Action
Chess (game in 30 minutes) test and received an Action rating of 2383, the
highest rating ever obtained on any model priced below $500. Although the rating
was a bit higher than expected, it was earned fair and square. It should be kept
in mind that computers generally earn ratings at game/30 about one hundred
points higher than they earn at the traditional tournament level of 40 moves in
two hours, so if this rating was accurate it would imply that the Diamond would
rate about 2283 at 40/2. The new model, Diamond II, is an upgrade that has not
been C.R.A. rated; Novag estimates it at 2420 Action which implies a 37 point
improvement and a 2320 rating at 40/2. While I have not done a direct comparison
between this model and the older one, I ran the Diamond II through my CCR
"One-Hour" test. It scored 37 points out of a possible 100, which works out to a
predicted rating of 2250 at 40/2, some 70 points below Novag's estimate. My
expert-rated son scored one draw in three games at game/30, the draw occurring
when the Diamond lacked the sophistication to win a pawn up endgame. My feeling
based on watching its play and analysis is that the truth lies between these
figures; we can say that the Diamond II is a very solid master, perhaps close to
2300, at 40/2, and a very strong master, knocking on the door of Senior Master,
at game/30. This is really quite an achievement for a unit powered by the
inexpensive h-8 processor, for although Novag likes to call it a "RISC style"
processor, it is not what most people in the computer industry would call a true
RISC processor. The processor runs at 16 MHz (a couple years ago most h-8 models
ran at 10 MHz), the Program size is 160K Byte ROM (enough for a sophisticated
program and a large opening book), and the RAM size is 129k Byte, most of which
(118k) is used for Hash Tables, which aid mostly in the endgame by avoiding
duplicate analysis of transposed lines, and probably add a hundred points or so
to the strength as compared to similar models without them (such as Novag's new
"Amber" and "Turquoise"). My "One-Hour" test does not cover the endgame and so
tends to under-rate models with hash tables, so this may partly account for the
slightly low score on this test. Novag models all have an aggressive,
interesting style, and the Diamond is no exception. If you are looking for the
strongest model available without spending several hundred dollars, you need
look no further. The strength vs. price ratio is excellent.

Regards,Terry



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.