Some games: Gothmog 0.4.7

[Patrick Buchmann]

Geschrieben von:/Posted by: Patrick Buchmann at 19 March 2004 13:09:06:

Hi,
There are some games from Gothmog 0.4.7 on my homepage.
First Ruel : 2557.
Patrick


UCI Engines Ligue - Division Senior

[Tord Romstad]

Geschrieben von:/Posted by: Tord Romstad at 19 March 2004 15:26:00:
Als Antwort auf:/In reply to: Some games: Gothmog 0.4.7 geschrieben von:/posted by: Patrick Buchmann at 19 March 2004 13:09:06:

Hi,
There are some games from Gothmog 0.4.7 on my homepage.
First Ruel : 2557.

Interesting. The pattern repeats itself so often that I'm getting more and
more sure it cannot possibly be a coincidence: Gothmog always scores better
in the UEL than in my own testing. 2557 is very close to Yace (2564) and
Pepito (2551). I have played matches against both of these opponents the
last couple of days, and found that Gothmog was clearly weaker. It lost
40-60 against Yace Paderborn and 42.5-57.5 against Pepito 1.59.
On the other hand, the improvement shown compared to the last few versions of
Gothmog are consistent with the results in my own tests.
One possible reason for our very different results could be that you test on
much slower hardware (PIII 800 MHz vs PIV 2.4 GHz). Perhaps Gothmog plays
relatively better on slow machines? Another possibility is that it has
something to do with the time control. Your games are played without
increment, while I always use a small increment in my tests. It is possible
that Gothmog's time management is more optimal in games without increment.
As soon as I have some CPU time, I will run some matches without increment
and check whether the results are better.
Tord

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 19 March 2004 15:36:39:
Als Antwort auf:/In reply to: Re: Some games: Gothmog 0.4.7 geschrieben von:/posted by: Tord Romstad at 19 March 2004 15:26:00:

Hi,
There are some games from Gothmog 0.4.7 on my homepage.
First Ruel : 2557.

Interesting. The pattern repeats itself so often that I'm getting more and
more sure it cannot possibly be a coincidence: Gothmog always scores better
in the UEL than in my own testing. 2557 is very close to Yace (2564) and
Pepito (2551). I have played matches against both of these opponents the
last couple of days, and found that Gothmog was clearly weaker. It lost
40-60 against Yace Paderborn and 42.5-57.5 against Pepito 1.59.
On the other hand, the improvement shown compared to the last few versions of
Gothmog are consistent with the results in my own tests.
One possible reason for our very different results could be that you test on
much slower hardware (PIII 800 MHz vs PIV 2.4 GHz). Perhaps Gothmog plays
relatively better on slow machines? Another possibility is that it has
something to do with the time control. Your games are played without
increment, while I always use a small increment in my tests. It is possible
that Gothmog's time management is more optimal in games without increment.
As soon as I have some CPU time, I will run some matches without increment
and check whether the results are better.
Tord

Another possibility that I can think about is learning if you do not use something like the nunn test.
I do not play long matchs because I am sure that with my very small book movei may lose the same game again and again.
Uri

[Patrick Buchmann]

Geschrieben von:/Posted by: Patrick Buchmann at 20 March 2004 09:00:14:
Als Antwort auf:/In reply to: Re: Some games: Gothmog 0.4.7 geschrieben von:/posted by: Tord Romstad at 19 March 2004 15:26:00:

Hi,
There are some games from Gothmog 0.4.7 on my homepage.
First Ruel : 2557.

Interesting. The pattern repeats itself so often that I'm getting more and
more sure it cannot possibly be a coincidence: Gothmog always scores better
in the UEL than in my own testing. 2557 is very close to Yace (2564) and
Pepito (2551). I have played matches against both of these opponents the
last couple of days, and found that Gothmog was clearly weaker. It lost
40-60 against Yace Paderborn and 42.5-57.5 against Pepito 1.59.
On the other hand, the improvement shown compared to the last few versions of
Gothmog are consistent with the results in my own tests.
One possible reason for our very different results could be that you test on
much slower hardware (PIII 800 MHz vs PIV 2.4 GHz). Perhaps Gothmog plays
relatively better on slow machines? Another possibility is that it has
something to do with the time control. Your games are played without
increment, while I always use a small increment in my tests. It is possible
that Gothmog's time management is more optimal in games without increment.
As soon as I have some CPU time, I will run some matches without increment
and check whether the results are better.
Tord

Hi Tord,
A new tournament from David Rabel is on my homepage.
5'+2", Athlon XP 1700+, 640Mo, Win2000 SP4, Fritz 8 GUI
Hash 32Mo, Biblio Nunn openings,Tablebases 5-pièces
Score List Ktul Tao Litt Goth Phar Frui
---------------------------------------------------------------------------
1: List 5.12 19.0 / 24 XXXX 1110 111= 111= 1=== 1011 11=1
2: Ktulu 4.2 15.0 / 24 0001 XXXX =10= =111 1101 =110 1==1
3: Tao 5.6 13.5 / 24 000= =01= XXXX 01=0 0110 11=1 1111
4: Little Goliath Nemesis 12.5 / 24 000= =000 10=1 XXXX 11=1 =1=1 =110
5: Gothmog 0.4.7 10.0 / 24 0=== 0010 1001 00=0 XXXX 1110 0110
6: Pharaon 2.62 9.0 / 24 0100 =001 00=0 =0=0 0001 XXXX 1111
7: Fruit 1.0 5.0 / 24 00=0 0==0 0000 =001 1001 0000 XXXX
---------------------------------------------------------------------------
84 games: +33 =18 -33
PGN on my homepage.
5'+2": increment time, Athlon XP 1700+.
Regards,
Patrick


UCI Engines Ligue - Division Senior - 19.03.04

[Tord Romstad]

Geschrieben von:/Posted by: Tord Romstad at 20 March 2004 12:45:19:
Als Antwort auf:/In reply to: More games: Gothmog 0.4.7 geschrieben von:/posted by: Patrick Buchmann at 20 March 2004 09:00:14:

Hi Tord,
A new tournament from David Rabel is on my homepage.
5'+2", Athlon XP 1700+, 640Mo, Win2000 SP4, Fritz 8 GUI

Thanks, Patrick!
An interesting thing to observe here is that this tournament is played with increment
and on faster hardware compared to your gauntlet. Gothmog's results are clearly worse
in David's tournament. This confirms my suspicion that Gothmog plays better on slow
hardware, or that its time management could be improved in games with increment.
Tord