Dedicated Chess Computer Test Scores

[Steve B]

Yes that is true that is our difference.

Good
We can Finally Agree On Something!

besides you won the Tournament with the TM Porto so you should respect it a bit more after that

Winning is Everything Regards
Steve

[spacious_mind]

Yes that is true that is our difference.

Good
We can Finally Agree On Something!

besides you won the Tournament with the TM Porto so you should repesct it a bit more after that

Winning is Everything Regards
Steve

Hey I love TM Porto! As I love them all. BTW. did you know that my very first extravagant purchase was 68020 Porotorse with Bavaria board. So Portorose has a special place in my heart.

Best regards

[paulwise3]

Hi Nick,

Just received the Saitek Executive, which is supposed to be a GK 2000 clone. Since it seems the GK 2000 is not yet on the list, I started testgame 1 with it, along with the Novag Emerald. Both with interesting results for game 1: The Executive scored about the same as the Cougar, and as for the Emerald, it scored an amazing 3032 for white!
See:

Code: Select all

    Executive | Emerald
T1W   2564       3032
T1B   2208       2312
T1T   2386       2672

Tomorrow testgame 2, and maybe some more.
Will be interesting to see which games are their weak spots

Best regards,
Paul

[spacious_mind]

Hi Nick,

Just received the Saitek Executive, which is supposed to be a GK 2000 clone. Since it seems the GK 2000 is not yet on the list, I started testgame 1 with it, along with the Novag Emerald. Both with interesting results for game 1: The Executive scored about the same as the Cougar, and as for the Emerald, it scored an amazing 3032 for white!
See:

Code: Select all

    Executive | Emerald
T1W   2564       3032
T1B   2208       2312
T1T   2386       2672

Tomorrow testgame 2, and maybe some more.
Will be interesting to see which games are their weak spots

Best regards,
Paul

HI Paul,

Wow that is a big score for Emerald It will be interesting to see how it finishes.

Best regards

[lexman]

I recently began running the Radio Shack Chess Master a 1998 Morsch program through the tests. I also began with the assumption that it may be a gk2000 clone although its identity has been the subject of some debate. In the first game its white score was a very creditable 2764. Its black score was 2208 to give a combined score of 2486. I would note that both the Barracuda and now the Senator, also tagged as gk2000 clones scored 2386 n this test, so the Chess Master does not appear to be a GK 2000 at least on this game. I will post the further tests in due course.

[spacious_mind]

I recently began running the Radio Shack Chess Master a 1998 Morsch program through the tests. I also began with the assumption that it may be a gk2000 clone although its identity has been the subject of some debate. In the first game its white score was a very creditable 2764. Its black score was 2208 to give a combined score of 2486. I would note that both the Barracuda and now the Senator, also tagged as gk2000 clones scored 2386 n this test, so the Chess Master does not appear to be a GK 2000 at least on this game. I will post the further tests in due course.

Thanks! keep publishing your scores and game moves and I will update the spreadsheets and ranking list accordingly and publish the completed game spreadsheets periodically.

ps. Mephisto Montana and Radio Shack 2050X also finished game 1 with 2386. So it will be interesting how yours and Paul's tests progress. Regarding Mephisto Senator I would imagine it will end up different as it should be a GK2100 follower and no GK2000.

Clone hunting regards

[lexman]

In game 2 the Radio Shack Chess Master scored 2085 for white and 1881 for black to give a combined score of 1983. This is very close to the Barracuda score of 1980 and was trendiing back to GK 2000 territory. However in game 3 it scored 2570 for white and 2255 for black to give a combined score of 2412 as opposed to 2215 for the Barracuda. Interestingly the only Morsch to match this score exactly in game 3 is Fritz 2 on a 386.

Game 2
Radio Shack Chess Master

Morsch
30 sec av
Radio Shack Chess Master
WHITE BLACK
MOVE PLAYED PT MOVE PLAYED PT
START TEST 6. ... Bb4+ 30.00
7.0-0 30.00 7. ... Nf6 28.80
8.Ng5 0.00 8. ... Qf5 30.00
9.Bd3 30.00 9. ... Bc5 13.80
10.Nd4 30.00 10. ... Qf6 30.00
11.Nxe6 0.00 11. ... Bh3 6.60
12.Nxe6 0.00 12. ... Bd7 30.00
13.Nxd5 0.00 13. ... 0-0-0 30.00
14.Bxc6 0.00 14. ... Bxc6 30.00
15.Nb5 0.00 15. ... Nb4 0.00
16.Be3 30.00 16. ... Bxe3+ 0.00
17.Nbxc7 30.00 17. ... Bb4 0.00
18.Rxe3 30.00 18. ... Bf7 27.60
19.Nxd5 30.00 19. ... Rd6 0.00
20.Rd3 0.00 20. ... g5 0.00
21.Rxe8 26.40 21. ... Re4 1.80
22.Rd3 30.00 22. ... Re4 30.00
23.b4 30.00 23. ... b5 0.00
24.Qb3 0.00 24. ... Nc6 0.00
25.Rxb7+ 30.00 25. ... Kc8 30.00
26.Rxa7+ 30.00 26. ... Kc8 30.00
27.Qb6+ 30.00 RESIGNS
PERFORMANCE EVALUATION
386.40 348.60
61.33% #REF! 55.33% #REF!
2085 0 1881 0
COMBINED RATING: 1983 0

Game 3
MFR NAME
Radio Shack Chess Master Saitek
morsch
av 30sec
H8?
WHITE BLACK
MOVE PLAYED PT MOVE PLAYED PT
START TEST 3. ... d6 30.00
4.Qg4 17.40 4. ... d6 30.00
5.Nf3 23.40 5. ... exf4 30.00
6.Nf3 30.00 6. ... Qb6 0.00
7.c3 0.00 7. ... Qxb2 30.00
8.Bxf7+ 30.00 8. ... Kd7 19.20
9.Be6+ 0.00 9. ... Qxa1 30.00
10.Bc1 0.00 10. ... Qf6 30.00
11.c3 30.00 11. ... Bxc3+ 0.00
12.Nxc3 30.00 12. ... Qxh1 30.00
13.Qg4+ 30.00 13. ... Kc7 30.00
14.Qxg7 30.00 14. ... Ne7 0.00
15.Bxg8 30.00 15. ... Ngf6 28.80
16.Bxd6+ 0.00 16. ... Kd8 0.00
17.Bxd6+ 30.00 17. ... Kb7 30.00
18.Bd5+ 30.00 18. ... Ka6 30.00
19.d4 30.00 19. ... b4 0.00
20.Bxb4 30.00 20. ... Ne5 0.00
21.c4+ 30.00 21. ... Kxb4 30.00
22.Qb3+ 30.00 END TEST
PERFORMANCE EVALUATION
430.80 378.00
75.58% #REF! 66.32% #REF!
2570 0 2255 0
COMBINED RATING: 2412 0

[spacious_mind]

Thanks for the game posts. Also compare RS Chess Master against Saitek Cougar which is also a GK2100 program but with 16 MHz instead of 10 MHz. It had scored ELO 2233.

129 SAITEK COUGAR 30S AV. 2382 2087 2403 2250 2089 2233

It seems that Morsch's like these tests and their results trend towards higher scores than their real game performances against other dedicated computers. Maybe these tests show a little of why human's have a lot of difficulty of beating these programs compared to other dedicated programmers.

Best regards

[paulwise3]

Here the results with testgame2 added

Code: Select all

    Executive | Emerald
T1W   2564       3032
T1B   2208       2312
T1T   2386       2672

T2W   2085       2377
T2B   1817       1778
T2T   1951       2077

It seems that specially the Emerald is much better in attacking then in defending .
Also I have to mention that the Emerald uses at least twice as much time as the Executive. In game 1 there were a lot of moves with at least 2 minutes, and also in game 2 that was clear: I started the test for both at the same time, and the Executive finished at the point where the Emerald was only at half the test. It makes me think of the situation where the computer has to make his first move after he is out of opening book. Many computers take their time at that moment. Possibly the Emerald resets its clock every time you take back a move?

Wondering regards,
Paul

[lexman]

So on to game 4. In this game the Barracuda scored 2026, while the Radioshack Mega 2050x, probably another GK2000 clone scored 2039. The Cougar widely regarded as a GK 2100 clone scores 2250 for this game. Where would the Radio Shack Chess Master place?
Well for this game it scored a combined total of 2115 approximately between
the two groups and lending some credence to it being a GK2100 but with a slower processor speed and less hash.

Game 4

Radio Shack
Chess Master
Morsch
Av 30sec
H8 10mhz
WHITE BLACK
MOVE PLAYED SCORE MOVE PLAYED SCORE
START TEST
13.Rfd1 16.80 13. ... Bb7 30.00
14.Nb3 22.80 14. ... cxd4 26.40
15.b3 25.20 15. ... cxb3 7.80
16.axb5 0.00 16. ... Bb7 0.00
17.Bc2 0.00 17. ... Be8 0.00
18.Bf2 0.00 18. ... Rac8 0.00
19.Bf2 0.00 19. ... Rac8 0.00
20.a5 30.00 20. ... bxa4 30.00
21.Bxa4 30.00 21. ... Nb5 30.00
22.Rac1 0.00 22. ... Bxb5 30.00
23.Rfb1 24.00 23. ... a5 30.00
24.Bf2 30.00 24. ... a5 30.00
25.bxa5 30.00 25. ... Rxa5 30.00
26.Rxa5 16.80 26. ... Rga8 25.20
27.Rxa5 30.00 27. ... Rxa5 22.00
28.Rc1 30.00 28. ... Bc6 24.00
29.Nf1 30.00 29. ... Ba4 18.00
30.N1d2 0.00 30. ... Ra2+ 30.00
31.Bd2 30.00 31. ... Bc6 15.60
32.Rxd1 30.00 32. ... Ba4 30.00
33.Rb1 30.00 33. ... Bc2 0.00
34.Kg3 30.00 34. ... fxg4 0.00
35.Ne3 30.00 35. ... Kg8 30.00
36.Bc1 0.00 36. ... Kg7 15.00
37.Bxa3 30.00 37. ... Bxc1 30.00
38.Rxc1 30.00 38. ... Kg7 0.00
39.Rg1 0.00 39. ... gxh5 7.20
40.Re1 1.20 40. ... Rxe1 30.00
41.Nxe1 30.00 41. ... gxh5 30.00
42.Kxg4 16.80 42. ... Nf5 16.80
43.Nxf5 0.00 43. ... gxf5+ 30.00
44.Kg3 30.00 44. ... Kg7 30.00
45.Nf3 0.00 45. ... Bc2 0.00
PERFORMANCE EVALUATION
603.60 628.00
60.97% 0.00% 63.43% 0.00%
2073White 2157 Black
combined rating 2115

[lexman]

And so to game 5 for the Radio Shack Chess Master. In this game it scored 2175 to give an overall combined rating of 2208 for the five games. This compares to 2233 for the Cougar and the GK2000 clones of RS Mega 2050 and the Barracuda scored in the 2040,s. So it seems likely that it is a GK2100 clone of some description. Sorted?
Well perhaps not. A friend who also owns an RS Master also ran the tests at the same times of 30s av and his scored somewhat worse with an overall rating of 2115 being particularly worse in the 5th game?? So rather like the Obsidian the question arises are there different versions with perhaps some running at 10mhz and others 16mhz or program tweaks given that Radio Shack had some input with Saitek? The enigma continues.
However given that it cost me 25 euro from a thrift shop I am reasonabley
happy with the purchase!
radio shack
Chess Master
Morsch
av 30sec
H8
WHITE BLACK
MOVE PLAYED SCORE MOVE PLAYED SCORE
START TEST
7.Bb3 30.00 7. ... Bb4 0.00
8.a3 19.20 8. ... Bc5 28.80
9.Qf3 30.00 9. ... d6 12.00
10.h3 22.80 10. ... Bg4 30.00
11.Bxe6 30.00 11. ... Qxe6 8.40
12.0-0 19.20 12. ... 0-0 14.40
13.Qe2 25.20 13. ... 0-0 0.00
14.Ne2 25.20 14. ... 0-0 0.00
15.Rae1 0.00 15. ... Bd4 25.80
16.Rae1 10.80 16. ... Raf8 3.60
17.Nd1 6.00 17. ... Kd8 10.80
18.Ne2 30.00 18. ... Kd8 4.80
19.Qc3 24.00 19. ... Kd8 0.00
20.b4 28.80 20. ... Bb6 0.00
21.a4 30.00 21. ... Nxe5 30.00
22.Nd4 30.00 22. ... Kd7 30.00
23.exd5 30.00 23. ... Bc7 30.00
24.f4 30.00 24. ... hxg3 30.00
25.exd5 0.00 25. ... hxg3 30.00
26.Qxg3 22.80 26. ... Qxg3+ 30.00
27.Nxg3 30.00 27. ... Nf4+ 30.00
28.Kh1 0.00 28. ... Rxh3 30.00
29.exd5 0.00 29. ... Nd3 30.00
PERFORMANCE EVALUATION
474.00 408.60
68.70% #REF! 59.22% #REF!
2336 0 2013 0
COMBINED RATING 2175

And here is a game where it thrashes the the Millennium Genius at 40/120 tournament time control: white RS Chess Master Black MCG

1. d4 c6 2.e4 d5 3.e5 Bf5 4. Ne2 e6 5.Ng3 Bg6 6.h4 h6 7.h5 Bh7 8.Bd3 Bxd3 9.Qxd3 Qb6 10.0-0 c5 11.c3 Nc6 12.Rd1 Be7 13.b3 cxd4N 14.cxd4 Bh4 15.Ne2 Rd8 16.Nbc3 a6 17.Qf3 Be7 18.Qg3 Kf8 19.Bb2 Rc8 20.Rac1 Bb4 21.Na4 Qa5 22.a3 Qb5? 23. Nf4 Be7 24.Nc5 Bxc5 25.Rxc5 Qb6 26.Qf3 Ke8 27.Qg4 KF8?? 28.Rxd5 Qxb3 29.Rd2 Nge7 30.Rd6 Na5 31.Rd7 Nac6?? 32.d5 exd5 33.R7xd5 Ra8 34.e6!! Rh7 35.exf7! Nd4 36.R5xd4 Kxf7 37.Rd7 Ke8 38.Ng6 Qf7 39.Qe4 Qxf2+ 40.Rxf2 Kxd7 41.Qxe7+ Kc6 42.Rxc2+ Kb5 43.Qb4 MATE

The MCG seems to have a problem in Caro-Kann lines the Sapphire also won its only victory (out of 20 games) against it in a Caro and in the recent 30 game tournament series the Star Diamond I believe also beat it in a Caro Kann. Perhaps the book needs attention here?

[spacious_mind]

And so to game 5 for the Radio Shack Chess Master. In this game it scored 2175 to give an overall combined rating of 2208 for the five games. This compares to 2233 for the Cougar and the GK2000 clones of RS Mega 2050 and the Barracuda scored in the 2040,s. So it seems likely that it is a GK2100 clone of some description. Sorted?
Well perhaps not. A friend who also owns an RS Master also ran the tests at the same times of 30s av and his scored somewhat worse with an overall rating of 2115 being particularly worse in the 5th game?? So rather like the Obsidian the question arises are there different versions with perhaps some running at 10mhz and others 16mhz or program tweaks given that Radio Shack had some input with Saitek? The enigma continues.
However given that it cost me 25 euro from a thrift shop I am reasonabley
happy with the purchase!

You cannot complain about 25 Euro Morsch's computers have a lot of built-in randomness therefore it is very possible that the tests get completed with some score differences. The difference between 2208 and 2115 might be just that rather than different versions. Even with ponder off you will see random with it. Also there are so many settings it would be good to confirm that you have all the parameters the same between your two tests.

ps.. can you post Test Game 1 so that I have all 5 for the tables.

Best regards

[lexman]

As requested game 1 for the RS Chess Master
radio shack
chess master yr 1998
frans morsch
30s av
HARDWARE DESCRIPTION
WHITE BLACK
MOVE PLAYED PT MOVE PLAYED PT
START TEST 4. ... Nf6 11.40
5.Nf3 30.00 5. ... Bf5 0.00
6.Bxf7+ 8.40 6. ... Bh5 30.00
7.Qxf3 30.00 7. ... Nc6 6.60
8.Qxb7 30.00 8. ... Nbd7 30.00
9.Nb5 30.00 9. ... Bd6 30.00
10.Nxa7 30.00 10. ... Rb8 30.00
11.Bb5+ 30.00 11. ... Nxc4 30.00
12.a4 30.00 12. ... Nd6 9.00
13.Qc6+ 30.00 13. ... Ke7 5.40
14.Qxb5+ 30.00 14. ... Nd7 12.60
15.a4 30.00 15. ... exd5 16.20
16.Qxd5 0.00 16. ... c5 0.00
17.Qxd5 30.00 17. ... 0-0 30.00
18.c3 16.20 18. ... c6 30.00
19.Qa4 0.00 19. ... Qe4+ 30.00
20.Qxd7+ 30.00 20. ... Kf8 30.00
21.Qd8+ 30.00 CHECKMATE
414.60 331.20
81.29% 0.00% 64.94% 0.00%
2764 0 2208 0 combined 2486

rating calculation

RADIO SHACK CHESS MASTER 30sec AV 2486 1983 2412 2115 2175 2208

regards

[paulwise3]

Here the results of the 5 testgames for the Saitek Executive and the Novag Emerald.

Code: Select all

           Tot    W1   B1   A1     W2   B2   A2     W3   B3   A3     W4   B4   A4     W5   B5   A5
Executive 2070 | 2564 2208 2386 | 2085 1817 1951 | 2713 1897 2305 | 1892 2157 2024 | 2007 1357 1682
Emerald   2224 | 3032 2312 2672 | 2377 1778 2077 | 2827 2373 2600 | 1671 1764 1718 | 2135 1972 2053

Like lexman mentioned, the Executive also scored bad in game 5, especially for black: 1357 !?
And as I posted before, the Emerald uses more than double the thinking time. I started test 4 for both at the same time. When the Executive finished it, the Emerald was not even halfway. And I completed test 5 for the Executive before test 4 was finished for the Emerald, may have taken more then 40 minutes before it completed test 4 then, and still scoring poor on that test...
On total average the Emerald scores 154 elo points better then the Executive, which is remarkable if the Executive really is a GK2000 clone. I guess I will let them play some testgames against each other
And I will send Nick the details by mail.

Regards, Paul

[BillT]

Hi Nick
Ive been running some models through the tests, but can you confirm how you've set up the boards you've done so far with regards the time control (All at 30 sec average ?)
For example,most of the tests I've done so far, I have been setting up for longer times than 30sec average .....say 1 minute and then interupting on 30-35 secs. The thought behind this is to stop moves being made too quickly,and equal times per move on each board
Also as you know some of these boards when set on average of 30 secs can take several minutes per move in certain situations. Surely its important that everyone is using the same time per move? Otherwise surely we'll see inacurate comparisons.or maybe I've missed something.?

Great tests regards

Bill