Stockfish 4 v Houdini 3 at long time control

Graham Banks · Post by **Graham Banks** » Mon Sep 30, 2013 7:40 am

PawnStormZ wrote:
Hi All.

Below is a link to the 100 games just played between Houdini 3 and StockFish 4...

http://www.datafilehost.com/d/b280b844

Mammoth effort Pal. Well done!

Modern Times · Post by **Modern Times** » Mon Sep 30, 2013 8:17 am

carldaman wrote:
PawnStormZ wrote:
Hi All.

Below is a link to the 100 games just played between Houdini 3 and StockFish 4...

http://www.datafilehost.com/d/b280b844

Pal, just so you know, the download has a .exe file extension tacked on at the end, which is quite odd.

CL

Untick the tick-box on the screen and you get a pgn file

A big effort this, it too one month from start to finish. You need to have patience for sure with these long time control matches.

carldaman · Post by **carldaman** » Mon Sep 30, 2013 7:25 pm

Modern Times wrote:
carldaman wrote:
PawnStormZ wrote:
Hi All.

Below is a link to the 100 games just played between Houdini 3 and StockFish 4...

http://www.datafilehost.com/d/b280b844

Pal, just so you know, the download has a .exe file extension tacked on at the end, which is quite odd.

CL
Untick the tick-box on the screen and you get a pgn file

A big effort this, it too one month from start to finish. You need to have patience for sure with these long time control matches.

Hey Ray

Thanks for your kind suggestion, it worked!

A big tip of the hat to Pal for running such a long match!

Thanks,
CL

Ponti · Post by **Ponti** » Tue Oct 01, 2013 7:22 pm

Each engine uses its default settings; no large ages; no end-game tablebases; 512 hash each; 4 threads each; ponder off.

Sorry, but... what does "no large ages" mean ?

A. Ponti

PawnStormZ · Post by **PawnStormZ** » Wed Oct 02, 2013 6:36 am

Thanks Graham, and All, for the kind words. I am glad that the match was interesting, and that the games were enjoyed.

PawnStormZ · Post by **PawnStormZ** » Wed Oct 02, 2013 6:40 am

Carl, that is a new annoying thing that the file-share site does, to try and "encourage" you to take their download manager. I should have mentioned what Ray said about the checkbox to avoid this. It is easy to overlook it (I wonder why).

PawnStormZ · Post by **PawnStormZ** » Wed Oct 02, 2013 6:54 am

Hi Aloisio. I am sure that others will correct me if needed but in simple terms, I think it is a way that windows (or other OS also?) can keep a large block of memory "all together" so that the program (like a chess engine) can access that memory faster. It helps the program read and write data more quickly out of and into memory.

I think that it has its issues because after programs open and close a bit (which happens often when running a tourney) then there is not a large enough continuous "page" to "hold" and give out. I did not want to bother with something like this which seems more trouble than it is worth, though obviously it has its advantages, or no one would be using it.

jd1 · Post by **jd1** » Wed Oct 02, 2013 9:08 am

Ponti wrote:Each engine uses its default settings; no large ages; no end-game tablebases; 512 hash each; 4 threads each; ponder off.

Sorry, but... what does "no large ages" mean ?

A. Ponti

I'm pretty sure he means "no large pages" ...

beram · Post by **beram** » Wed Oct 02, 2013 7:55 pm

Hi Pal,

I have been playing through some games and I have questions about two of them

1)
In this drawn position in game 34, Houdini plays the directly losing move 97. a7 ?? at depth 1

Code: Select all

97.a7 { ev=2,d=1,mt = 00:29 }

Why could that be for ?

[d] 8/1K6/P1n5/3k1p2/5P2/5P2/8/8 w - - 0 97

2)
Game 67 won by Stockfish, the moves end at move 7, not only in the zip file but also in your original posting

[Event ""]
[Site "ChessPalace"]
[Date "2013.09.16"]
[Round "34.1"]
[White "Stockfish 4 64 SSE4.2"]
[Black "Houdini 3 x64"]
[Result "1-0"]
[Time "4:44:46 AM"]
[TimeControl "40/7200:20/3600:1800+10"]
[PlyCount "13"]
[Number "67"]
[Termination "time"]
[BlackType "program"]
[WhiteType "program"]
[Variant "normal"]

1.e4 { Eval = 0, Depth = 1, mt = 00:00 } c5 { ev=0,d=1,mt = 00:00 }

2.c3 { ev=0,d=1,mt = 00:00 } Nf6 { ev=0,d=1,mt = 00:00 }

3.e5 { ev=0,d=1,mt = 00:00 } Nd5 { ev=0,d=1,mt = 00:00 }

4.d4 { ev=0,d=1,mt = 00:00 } cxd4 { ev=0,d=1,mt = 00:00 }

5.Nf3 { ev=0,d=1,mt = 00:00 } Nc6 { ev=0,d=1,mt = 00:00 }

6.cxd4 { ev=0,d=1,mt = 00:00 } d6 { ev=0,d=1,mt = 00:00 }

7.Bc4 { ev=0,d=1,mt = 00:00 }

1-0

PawnStormZ · Post by **PawnStormZ** » Thu Oct 03, 2013 4:34 am

Hi Bram. I have no idea why Houdini moved on depth 1 in game 34 at move 97. I do not know how much time it had left at that point, but it looks like it took just about the full "increment" before moving. Something must have caused it to "get stuck" at depth 1 for the whole 30 seconds. It was taking about the same time for a few moves before but getting much deeper.

With game 67 there must have been some problem with the GUI sending the moves of the game (after the start moves) to the pgn file. My original file only has the moves through the end of the start position after White's 7th. Sorry, but there is nothing that I can do at this point to recover the moves. Another possibility is that Houdini did not respond after the book moves, and the GUI considered it a loss on time; that is supported by the fact that the comment for "termination" says "time", and the time between the start of this game and the next was 2 hours; Houdini must have "hung" on its first move out of book.

If the match had been close, I would re-play at least game 67, maybe 34 too, but here it is probably not worth it. I am considering re-running the whole match, with the same start positions, but with Houdini set to contempt 0. That should allow it to play at full strength against an opponent that is at least its equal.