Main
Date: 26 Aug 2007 16:17:51
From: M Winther
Subject: Blitz rating, Fritz10 and Genius
I tested Fritz10 against Genius5 i a five game 5-min blitz match. I
set the hash tables to equal sizes. Genius won with 4-1. The result is
curious while Fritz10 represents the latest in engine technology while
Genius5 is quite old. Fritz10 did not have much of a chance. Genius
always seems able to strengthen his positions in positional games.
This strengthens my suspicion that Genius is stronger than almost all
other programs on blitz levels. What is measured today is strength in
standard time (40/2h in the SSDF list), and I suspect that the
programs are optimized for this setting. However, most people use the
programs for blitz and analysis, I believe, and they seldom allow the
program two minutes for a move. So there really ought to be a blitz
rating list, too. Then I suspect Genius would be among the foremost. I
wouldn't be surprised if the freeware Genius3 is stronger in blitz
than most other modern engines, too, despite the fact that it allows
only a small hash table.

On handheld devices I suspect that we would get a similar
result. Genius is probably among the strongest programs in blitz.
While handhelds are never used for longer time-levels, I wonder if
Genius isn't the most appropriate program.

Mats




 
Date: 27 Aug 2007 21:47:28
From: M Winther
Subject: Re: Blitz rating, Fritz10 and Genius
Den 2007-08-26 16:17:51 skrev M Winther <[email protected] >:

> I tested Fritz10 against Genius5 i a five game 5-min blitz match. I
> set the hash tables to equal sizes. Genius won with 4-1. The result is
> curious while Fritz10 represents the latest in engine technology while
> Genius5 is quite old. Fritz10 did not have much of a chance. Genius
> always seems able to strengthen his positions in positional games.
> This strengthens my suspicion that Genius is stronger than almost all
> other programs on blitz levels. What is measured today is strength in
> standard time (40/2h in the SSDF list), and I suspect that the
> programs are optimized for this setting. However, most people use the
> programs for blitz and analysis, I believe, and they seldom allow the
> program two minutes for a move. So there really ought to be a blitz
> rating list, too. Then I suspect Genius would be among the foremost. I
> wouldn't be surprised if the freeware Genius3 is stronger in blitz
> than most other modern engines, too, despite the fact that it allows
> only a small hash table.
>
> On handheld devices I suspect that we would get a similar
> result. Genius is probably among the strongest programs in blitz.
> While handhelds are never used for longer time-levels, I wonder if
> Genius isn't the most appropriate program.
>
> Mats
>



The strange result was due to the fact that I had "permanent brain" notched
in both programs. Evidently Fritz didn't like this and played badly. When
I turned it off Fritz10 began to play chess and won a game. So the above
result was an illusion, then.

However, it would still be worthwhile with a blitz rating list, while chess players
use the programs only for rapid games and analysis.

Mats



  
Date: 27 Aug 2007 16:28:32
From: SAT W-7
Subject: Re: Blitz rating, Fritz10 and Genius
That was interesting ....

Now you should try and play ten min games and see how they do ....



 
Date: 26 Aug 2007 19:09:58
From: M Winther
Subject: Re: Blitz rating, Fritz10 and Genius
Den 2007-08-26 16:17:51 skrev M Winther <[email protected] >:

> I tested Fritz10 against Genius5 i a five game 5-min blitz match. I
> set the hash tables to equal sizes. Genius won with 4-1. The result is
> curious while Fritz10 represents the latest in engine technology while
> Genius5 is quite old. Fritz10 did not have much of a chance. Genius
> always seems able to strengthen his positions in positional games.
> This strengthens my suspicion that Genius is stronger than almost all
> other programs on blitz levels. What is measured today is strength in
> standard time (40/2h in the SSDF list), and I suspect that the
> programs are optimized for this setting. However, most people use the
> programs for blitz and analysis, I believe, and they seldom allow the
> program two minutes for a move. So there really ought to be a blitz
> rating list, too. Then I suspect Genius would be among the foremost. I
> wouldn't be surprised if the freeware Genius3 is stronger in blitz
> than most other modern engines, too, despite the fact that it allows
> only a small hash table.
>
> On handheld devices I suspect that we would get a similar
> result. Genius is probably among the strongest programs in blitz.
> While handhelds are never used for longer time-levels, I wonder if
> Genius isn't the most appropriate program.
>
> Mats
>


I played one more blitz game Fritz10-Genius5 and Genius won
this, too. He wins on time in a winning position. PGN is below.
So now it's 5-1 to Genius. Isn't this rekable? Genius5 is from
1996!

I tested the blitz capability of Fritz10 by a four game tournament
against ProDeo, which is an improved Rebel12, a very strong
program. Fritz10 won by 31/2 - 1/2. Fritz10 is very strong in blitz,
but it has no chance against Genius5.

/Mats


[Event "Blitz:5'"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "Fritz 10"]
[Black "Genius 5"]
[Result "*"]
[PlyCount "81"]

1. Nf3 Nf6 2. c4 c6 3. g3 d5 4. Bg2 dxc4 5. Qc2 Qd5 6. O-O Bf5 7. Qc3 e6 8. Ng5
Qc5 9. e4 Bg6 10. d4 cxd3 11. Qxd3 h6 12. Nh3 Qb4 13. Nc3 Nbd7 14. a3 Ne5 15.
Qc2 Qc4 16. Bf4 Neg4 17. Rad1 e5 18. Bc1 Bc5 19. b3 Qe6 20. Na4 Bd4 21. Nf4
Bxf2+ 22. Rxf2 exf4 23. Nc5 Qe7 24. Rxf4 O-O 25. h3 Ne5 26. Qc3 Rfe8 27. Bb2 b6
28. Nd3 Rad8 29. Qxe5 Qxe5 30. Bxe5 Rxe5 31. Nxe5 Rxd1+ 32. Kf2 Nh5 33. Nxg6
Rd2+ 34. Kg1 Nxf4 35. Nxf4 Rb2 36. b4 Rb3 37. Kf2 Rxa3 38. e5 Rb3 39. Bxc6 Rxb4
40. Ke3 Rb3+ {time} 0-1

Isn'