WBEC Ridderkerk new results.

[Leo Dijksman]

Geschrieben von: / Posted by: Leo Dijksman at 10 January 2004 16:36:26:

New results Premier division:

Round 13:
Rebel 12 ½1 1.5/2 = 2 games to go!
Gandalf 4.32h ½0 0.5/2
------------------------------------
Round 14:
Ruffian 2.0.0 ½11½ 3.0/4 = No "insufficient material" (both!)?
Pharaon 2.62 ½00½ 1.0/4
------------------------------------
Gandalf 4.32h 101 2.0/3 = 1 game to go!
GreenLightChess 3.00 010 1.0/3
------------------------------------

Next matches:
Round 14:
LittleGoliath 3.9po - Rebel 12
The King 3.23 - Aristarch 4.21
Crafty 19.06 - SmarThink 0.17a
WBNimzo 2000b - WARP 0.58
Zarkov 4.5e - Yace Paderborn
Nejmet 3.07 - Gromit 3.82

New results 2nd Division:

Round 14:
Butcher 1.42c ½1½0 2.0/4
Resp 0.19 ½0½1 2.0/4
------------------------------------
King Of Kings 2.52 1110 3.0/4
Amateur 2.70 0001 1.0/4
------------------------------------
Horizon 4.1 ½110 2.5/4
Knightx 1.80a ½001 1.5/4
------------------------------------
XiniX 2.9.11 - Esc 1.16 0-1 = 3 games to go!
------------------------------------

Next matches:
Round 15:
The Baron 1.1 b3 - Horizon 4.1
Chezzz 1.0.3 - Amy 0.8.3po
Movei 0.08.131 - Hagrid 0.7.56
SlowChess 2.89b - The CrazyBishop 0.04.5
List 4.61 - Ktulu 4.1
Queen 2.42 - Resp 0.19
King Of Kings 2.52 - Butcher 1.42c
Esc 1.16 - Knightx 1.80a
Amateur 2.70 - XiniX 2.9.11

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
10/01/2003:
Updated: SEE v0.6.7
Updated: Adamant v1.2b
Leo.


WBEC Ridderkerk homepage.

[Roger Brown]

Geschrieben von: / Posted by: Roger Brown at 10 January 2004 21:19:06:
Als Antwort auf: / In reply to: WBEC Ridderkerk new results. geschrieben von: / posted by: Leo Dijksman at 10 January 2004 16:36:26:

The race for second place - yes it seems to be all over bar the shouting (or an inexplicable fall-off in Deep Sjeng's form) - is heating up nicely.

Leo, how do you do it? Even better question, how do you stand this excitement year after year?

:-0

Later.

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 10 January 2004 21:24:36:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Roger Brown at 10 January 2004 21:19:06:

The race for second place - yes it seems to be all over bar the shouting (or an inexplicable fall-off in Deep Sjeng's form) - is heating up nicely.

No
It is not over and the king still have chances to win.
Deep Sjeng is the favourite to win but nothing is finished.
The king3.23 has it's chances.
Uri

[Lance Perkins]

Geschrieben von: / Posted by: Lance Perkins at 10 January 2004 22:30:50:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Uri Blass at 10 January 2004 21:24:36:

I too have high hopes for The King 3.23, because Leo is using BookThinker with it. Thinker will never make it to the Premier Division, but at least, BookThinker made an apperance there in some form. Hahaha.
But its going to be tough. The remaining games for DSJ are now just against the bottom of the field.

The race for second place - yes it seems to be all over bar the shouting (or an inexplicable fall-off in Deep Sjeng's form) - is heating up nicely.

No
It is not over and the king still have chances to win.
Deep Sjeng is the favourite to win but nothing is finished.
The king3.23 has it's chances.
Uri

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 10 January 2004 22:55:14:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Lance Perkins at 10 January 2004 22:30:50:

I too have high hopes for The King 3.23, because Leo is using BookThinker with it. Thinker will never make it to the Premier Division, but at least, BookThinker made an apperance there in some form. Hahaha.

How do you know that thinker will never promote to the premier division?
If you improve it enough it may do it.
The gap between the premier division and the first division is not so big.
Uri

[Roger Brown]

Geschrieben von: / Posted by: Roger Brown at 11 January 2004 02:13:25:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Uri Blass at 10 January 2004 22:55:14:

The gap between the premier division and the first division is not so big.
Uri

Hello Uri,
I admit to being one of those persons who must be shown - at least on this point.
Have you really had a look at some of the names in the First Division? Have you had a look at the names in the cellar of the Premier Division?
Please, I implore you, elucidate your point above.
Later.

[Thomas Mayer]

Geschrieben von: / Posted by: Thomas Mayer at 11 January 2004 02:20:40:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Roger Brown at 11 January 2004 02:13:25:

Hi Roger,

The gap between the premier division and the first division is not so big.

I admit to being one of those persons who must be shown - at least on this
point. Have you really had a look at some of the names in the First
Division? Have you had a look at the names in the cellar of the Premier
Division?
Please, I implore you, elucidate your point above.

I definitely agree with you, Roger - Quark v2.05 might have won the 1st Division but there would be no hope to hold the premier division - there is a very huge difference in strength in my opinion. (But I think it starts already in the 2nd Division which is IMO clearly stronger then the 3rd. - but anyway, every Division is a big step - if you promote a big one forward - and when you relegate you haven't done enough between the Editions...
Greets, Thomas

[Sune Fischer]

Geschrieben von: / Posted by: Sune Fischer at 11 January 2004 02:37:24:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Thomas Mayer at 11 January 2004 02:20:40:

Hi Roger,

The gap between the premier division and the first division is not so big.

I admit to being one of those persons who must be shown - at least on this
point. Have you really had a look at some of the names in the First
Division? Have you had a look at the names in the cellar of the Premier
Division?
Please, I implore you, elucidate your point above.

I definitely agree with you, Roger - Quark v2.05 might have won the 1st Division but there would be no hope to hold the premier division - there is a very huge difference in strength in my opinion. (But I think it starts already in the 2nd Division which is IMO clearly stronger then the 3rd. - but anyway, every Division is a big step - if you promote a big one forward - and when you relegate you haven't done enough between the Editions...
Greets, Thomas

I think there is an overlap between the divisions, and the difference is not bigger than at any given sunday anything can happen.
For instance take a look at the just finished class C of RWBC
http://home.t-online.de/home/g.simon.rgbg/ccr3.htm
1 Gothmog_040, NO 19.5
2 Kaissa17, BY 19.5
3 Tytan_276, PL 18
4 SlowChess_289, US 17.5
5 Frenzee146, DK 17.5
6 Postmodernist_1007, UK 16
After 23 games they are pretty close and these engines have been scattered in WBEC over three divisions.
-S.

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 11 January 2004 07:03:37:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Thomas Mayer at 11 January 2004 02:20:40:

Hi Roger,

The gap between the premier division and the first division is not so big.

I admit to being one of those persons who must be shown - at least on this
point. Have you really had a look at some of the names in the First
Division? Have you had a look at the names in the cellar of the Premier
Division?
Please, I implore you, elucidate your point above.

I definitely agree with you, Roger - Quark v2.05 might have won the 1st Division but there would be no hope to hold the premier division - there is a very huge difference in strength in my opinion. (But I think it starts already in the 2nd Division which is IMO clearly stronger then the 3rd. - but anyway, every Division is a big step - if you promote a big one forward - and when you relegate you haven't done enough between the Editions...
Greets, Thomas

luck is a significant factor.
look at amyan's result.
in the last time it was only one place above the baron that went down and
this time it even got to place 4 and there was no improvement in amyan.
I think that Quark2.05 could survive or not survive in the premier based on luck.
There is a difference in strength but not a very huge difference.
Movei probably does not promote to the first division but the same program could beat Fritz8 and Yace with black and draw against Rebel and elchinito with white in thorsten's tournament when the only loss is against Hiarcs9(I consider 135 that is playing in most of thorsten's games to be the same as 131 because the only difference is being 2% faster).
I decided not to send an upgrade in the next few rounds because every result that an upgrade can achieve is going to look bad(I also do not have a big improvement and I am not sure that I have an improvement)
Note that for testing at long time control I prefer to use test suites(I have not enough computer time for significant number of games at long time control and 135 is best so far in the gcp test suite at 300 seconds per move)
latest versions got better results in fast time control against previous version but I am not happy if it does not also get better result in the gcp test suite at 300 seconds per move(I am also not sure if better result in games against itself means being better and I got worse results in a match of 143 against Junior5 (5 minutes against 15 minutes) but the difference is not significant(145 also got slightly more than 50% against 143 in the nunn match
but difference again was not significant)
difference between versions after 135 and 135 are only small difference in the search.
unfortunately I have not enough computer time for it and I used a lot of computer time to prove in 2 correspondence games that my opponents have no way to escape from a loss.
I agreed to play in the olympiad because I was asked again and again to participate in tournament after winning the israeli championship
but I said no to other tournaments in the near future.
I have 6/7 (5 wins and 2 draws) and
I will probably finish with at least 8.5/10
Uri

[Roger Brown]

Geschrieben von: / Posted by: Roger Brown at 11 January 2004 07:55:32:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Sune Fischer at 11 January 2004 02:37:24:

I think there is an overlap between the divisions, and the difference is not bigger than at any given sunday anything can happen.

Hello Sune,
All the best for the New Year!
Yes, I believe that luck plays a part in all contests - I cannot forget Robert Horry's shot that essentially kept the Lakers in the NBA Finals - but after 68 WBEC games (and counting) the thing tends to sort itself out.
Leo's top ten contains a number of the usual suspects including Green Light Chess, an engine worthy of much more buzz than it currently enjoys.
Incidentally, take a look at the engines Gunther has seeded in the top flight tournament on his pages. Those names seem familiar - certainly the top fifteen.
GLC is way too low but I have a feeling that situation will be rectified.
The Premier top fifteen will smash the cream (the top five) of the First Division. I will put a small sum on it.
Ah, Thomas, sorry about that!
All the very best to Quark in the Premier but it is a jungle.
Later.

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 11 January 2004 15:49:59:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Sune Fischer at 11 January 2004 02:37:24:

I think there is an overlap between the divisions, and the difference is not bigger than at any given sunday anything can happen.
For instance take a look at the just finished class C of RWBC
http://home.t-online.de/home/g.simon.rgbg/ccr3.htm
1 Gothmog_040, NO 19.5
2 Kaissa17, BY 19.5
3 Tytan_276, PL 18
4 SlowChess_289, US 17.5
5 Frenzee146, DK 17.5
6 Postmodernist_1007, UK 16
After 23 games they are pretty close and these engines have been scattered in WBEC over three divisions.

I agree that there is certainly an overlap between the divisions of WBEC, but I don't agree
with your choice of example above. Except for Frenzee, all the top 6 engines in class C of
RWBC have developed very quickly during last autumn. The versions which played in
RWBC could have been very different from those that play(ed) in WBEC. I know that this
is the case for Gothmog. The WBEC version is at least 100 points weaker than the
RWBC version, which is again at least 100 points weaker than the current version.
Tord

[Sune Fischer]

Geschrieben von: / Posted by: Sune Fischer at 11 January 2004 16:02:46:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Tord Romstad at 11 January 2004 15:49:59:

I think there is an overlap between the divisions, and the difference is not bigger than at any given sunday anything can happen.
For instance take a look at the just finished class C of RWBC
http://home.t-online.de/home/g.simon.rgbg/ccr3.htm
1 Gothmog_040, NO 19.5
2 Kaissa17, BY 19.5
3 Tytan_276, PL 18
4 SlowChess_289, US 17.5
5 Frenzee146, DK 17.5
6 Postmodernist_1007, UK 16
After 23 games they are pretty close and these engines have been scattered in WBEC over three divisions.

I agree that there is certainly an overlap between the divisions of WBEC, but I don't agree
with your choice of example above. Except for Frenzee, all the top 6 engines in class C of
RWBC have developed very quickly during last autumn. The versions which played in
RWBC could have been very different from those that play(ed) in WBEC. I know that this
is the case for Gothmog.
The WBEC version is at least 100 points weaker than the
RWBC version, which is again at least 100 points weaker than the current version.
Tord

That's the case for Gothmog and Tytan, the other 4 have idential version numbers (if slowchess289 and slowchess289b is the same)
As you can see PM1007 did quite well in div 1, certainly in no danger of demotion.
SlowChess did also well in div 2 finished midfield but frenzee had great difficulties promoting from div 3,
it was more luck than anything beating Terra I think.
All that could be one big statistical fluke, but if it isn't it does suggest that the divisions aren't too far apart.
Yeah man, isn't it about time for a long vacation?
This frantic pace of yours is killing us all!
-S.

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 11 January 2004 17:07:09:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Sune Fischer at 11 January 2004 16:02:46:

I think there is an overlap between the divisions, and the difference is not bigger than at any given sunday anything can happen.
For instance take a look at the just finished class C of RWBC
http://home.t-online.de/home/g.simon.rgbg/ccr3.htm
1 Gothmog_040, NO 19.5
2 Kaissa17, BY 19.5
3 Tytan_276, PL 18
4 SlowChess_289, US 17.5
5 Frenzee146, DK 17.5
6 Postmodernist_1007, UK 16
After 23 games they are pretty close and these engines have been scattered in WBEC over three divisions.

I agree that there is certainly an overlap between the divisions of WBEC, but I don't agree
with your choice of example above. Except for Frenzee, all the top 6 engines in class C of
RWBC have developed very quickly during last autumn. The versions which played in
RWBC could have been very different from those that play(ed) in WBEC. I know that this
is the case for Gothmog.
The WBEC version is at least 100 points weaker than the
RWBC version, which is again at least 100 points weaker than the current version.

That's the case for Gothmog and Tytan, the other 4 have idential version numbers (if slowchess289 and slowchess289b is the same)
As you can see PM1007 did quite well in div 1, certainly in no danger of demotion.
SlowChess did also well in div 2 finished midfield but frenzee had great difficulties promoting from div 3,
it was more luck than anything beating Terra I think.
All that could be one big statistical fluke, but if it isn't it does suggest that the divisions aren't too far apart.
Yeah man, isn't it about time for a long vacation?
This frantic pace of yours is killing us all!

OK. I should have checked before posting, instead of trusting my memory. Sorry.
No, it's definitely not time for a vacation. I should rather start working much harder.
Within a year or so, everybody will have a 64-bit computer. Frenzee and all other
bitboarders will suddenly become twice as fast, while I will remain as a proud member
of the sub-200kN/s club. If I want to keep any hope of competing with the rest of you,
I need to be far ahead before then.
Tord

[Sune Fischer]

Geschrieben von: / Posted by: Sune Fischer at 11 January 2004 17:49:10:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Tord Romstad at 11 January 2004 17:07:09:

Yeah man, isn't it about time for a long vacation?
This frantic pace of yours is killing us all!

No, it's definitely not time for a vacation. I should rather start working much harder.
Within a year or so, everybody will have a 64-bit computer. Frenzee and all other
bitboarders will suddenly become twice as fast, while I will remain as a proud member
of the sub-200kN/s club. If I want to keep any hope of competing with the rest of you,
I need to be far ahead before then.
Tord

I looking forward to getting a 64 bit chip as well, imagine a shift going from ~12 clocks to 1-2 clocks!
Finally it's going to be possible to work with bitboards as they always were intended.
But I'm sure Gothmog will get a good speedup too, so in the end I don't think the differences will be that big.
My main problem has always been to figure out if a change was an improvement or not, typicly I get conflicting results like 8% better against Crafty and 12% worse against the old version...
-S.

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 11 January 2004 17:53:08:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Sune Fischer at 11 January 2004 17:49:10:

Yeah man, isn't it about time for a long vacation?
This frantic pace of yours is killing us all!

No, it's definitely not time for a vacation. I should rather start working much harder.
Within a year or so, everybody will have a 64-bit computer. Frenzee and all other
bitboarders will suddenly become twice as fast, while I will remain as a proud member
of the sub-200kN/s club. If I want to keep any hope of competing with the rest of you,
I need to be far ahead before then.
Tord

I looking forward to getting a 64 bit chip as well, imagine a shift going from ~12 clocks to 1-2 clocks!
Finally it's going to be possible to work with bitboards as they always were intended.
But I'm sure Gothmog will get a good speedup too, so in the end I don't think the differences will be that big.
My main problem has always been to figure out if a change was an improvement or not, typicly I get conflicting results like 8% better against Crafty and 12% worse against the old version...
-S.

My conflicting results can be only the opposite because if it has not more than 50% against older version I do not continue testing against other programs.
Uri

[Anthony Cozzie]

Geschrieben von: / Posted by: Anthony Cozzie at 11 January 2004 21:00:07:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Tord Romstad at 11 January 2004 17:07:09:

Within a year or so, everybody will have a 64-bit computer. Frenzee and all other
bitboarders will suddenly become twice as fast, while I will remain as a proud member
of the sub-200kN/s club. If I want to keep any hope of competing with the rest of you,
I need to be far ahead before then.
Tord

Some of us have dual opterons already
If I can just get a 64 bit kernel running . . .
anthony
P.S. Zappa is 40% faster on the opteron in 32 bit mode clock/clock. Opteron is just a beast of a CPU. One of the reasons Amateur has been killing people of late is his Athlon FX.
P.P.S. With 32 bit mode Zappa gets 550 knps in middlegame. I'm hoping for 800+ with 64 bits . . .

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 11 January 2004 22:05:55:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Anthony Cozzie at 11 January 2004 21:00:07:

Within a year or so, everybody will have a 64-bit computer. Frenzee and all other
bitboarders will suddenly become twice as fast, while I will remain as a proud member
of the sub-200kN/s club. If I want to keep any hope of competing with the rest of you,
I need to be far ahead before then.
Tord

Some of us have dual opterons already
If I can just get a 64 bit kernel running . . .
anthony
P.S. Zappa is 40% faster on the opteron in 32 bit mode clock/clock. Opteron is just a beast of a CPU. One of the reasons Amateur has been killing people of late is his Athlon FX.
P.P.S. With 32 bit mode Zappa gets 550 knps in middlegame. I'm hoping for 800+ with 64 bits . . .

And I will probably have a dual PowerMac G5 within half a year. Unfortunately I don't have
any good idea about how to make use of the 64-bitness of the processor. I could of course
start to gradually add bitboards to my engine, but I'm afraid it would make it more difficult
to port Gothmog to PalmOS, which after all is my main goal at the moment. It is also a
problem that bitboards are not very useful for hexagonal chess. Whenever possible, I
prefer to keep my code as general as possible, and to avoid making assumptions about
the size or shape of the board or of the word size of the processor.
At least I easily avoid that problem by running Mac OS X.
If you have some time: Could you please try Gothmog 0.4.5 on your dual opteron? The
program includes a benchmark feature. You run it by typing "gothmog bench" from the
command line. Running the benchmark takes 5 minutes. On my PowerBook G4 550 MHz
the result is about 48 kN/s, while on the PIV 2.4 GHz in my office the result is 160 kN/s.
By the way, do you plan for Zappa, the best thing to ever happen to America, to become a
public engine at some stage? It definitely seems to be an interesting engine, and I would
love to try it out (especially if you will release a Linux or OS X version).
Tord

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 11 January 2004 22:13:31:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Sune Fischer at 11 January 2004 17:49:10:

I looking forward to getting a 64 bit chip as well, imagine a shift going from ~12 clocks to 1-2 clocks!
Finally it's going to be possible to work with bitboards as they always were intended.
But I'm sure Gothmog will get a good speedup too, so in the end I don't think the differences will be that big.
My main problem has always been to figure out if a change was an improvement or not, typicly I get conflicting results like 8% better against Crafty and 12% worse against the old version...

I'm not so sure. The current version of Gothmog wouldn't have any profit at all from running
on a 64-bit processor. The only 64-bit variables I have are my Zobrist keys and my node
counters. I would have to make a substantial rewrite in order to take advantage of 64-bit
processors. Perhaps I will give it a try when I get a G5, but I'm afraid it will be difficult
without sacrificing the possibility of porting Gothmog to PalmOS.
Yes, it is frustrating. I usually end up trusting my intuition, which is probably wrong at
least 50% of the time.
Tord

[Marcus Prewarski]

Geschrieben von: / Posted by: Marcus Prewarski at 12 January 2004 22:41:32:
Als Antwort auf: / In reply to: Re: WBEC Ridderkerk new results. geschrieben von: / posted by: Tord Romstad at 11 January 2004 22:13:31:

I looking forward to getting a 64 bit chip as well, imagine a shift going from ~12 clocks to 1-2 clocks!
Finally it's going to be possible to work with bitboards as they always were intended.
But I'm sure Gothmog will get a good speedup too, so in the end I don't think the differences will be that big.
My main problem has always been to figure out if a change was an improvement or not, typicly I get conflicting results like 8% better against Crafty and 12% worse against the old version...

I'm not so sure. The current version of Gothmog wouldn't have any profit at all from running
on a 64-bit processor. The only 64-bit variables I have are my Zobrist keys and my node
counters. I would have to make a substantial rewrite in order to take advantage of 64-bit
processors. Perhaps I will give it a try when I get a G5, but I'm afraid it will be difficult
without sacrificing the possibility of porting Gothmog to PalmOS.
Yes, it is frustrating. I usually end up trusting my intuition, which is probably wrong at
least 50% of the time.
Tord

I think every program will get some benefit. Things like instruction fetches will happen 64bits at a time instead of 32, as well as other data transfers.

-Marcus