nTCEC Season 1 Stage 3 - 16 cores of chess fun

[Graham Banks]

I'd thought it was well known not to set up an egbb path in Chiron when using ChessGUI.
It's okay to put a path for both Nalimov and GaviotaTablebases because only one of them will be used.

Although I've had the occasional time loss with Chiron on 4 cores and 8 cores, they've not really been that frequent (perhaps once in every 30 games or so at worst).

Yes agreed, Martin must have been very unlucky with it.

Hard to know the reason.
I guess that if Martin has provided the Chiron author with the debug files that ChessGUI automatically creates for each game, the reason or reasons could be pinpointed and addressed,
Perhaps something related to large pages or hash size used or time control? No way to know for sure without the debug files being studied I guess.

[Martin Thoresen]

When a human player withdraws from a tournament, recent (and fair) practice has been to erase his/her score and have all his games thus treated as a bye for the opponent. I hope the same thing is done here, as Graham has suggested.

Regards,
CL

Hi Carl,

Yes, all results involving Chiron will be changed to won for the opponent if they weren't already. Thanks for the input.

Best,
Martin

[Graham Banks]

When a human player withdraws from a tournament, recent (and fair) practice has been to erase his/her score and have all his games thus treated as a bye for the opponent. I hope the same thing is done here, as Graham has suggested.

Regards,
CL

Hi Carl,

Yes, all results involving Chiron will be changed to won for the opponent if they weren't already. Thanks for the input.

Best,
Martin

Well done Martin.
You're doing a great job overall.

[Martin Thoresen]

Hard to know the reason.
I guess that if Martin has provided the Chiron author with the debug files that ChessGUI automatically creates for each game, the reason or reasons could be pinpointed and addressed,
Perhaps something related to large pages or hash size used or time control? No way to know for sure without the debug files being studied I guess.

Well, Andrea told me he is unsure of the reason at this moment. But he has got all debug files but couldn't find anything useful in them afaik.

He has tested with the hash sizes used in nTCEC himself, without problems.
Large pages are not used in nTCEC.

He already discovered an endgame tablebase bug due to these losses on time so we decided to disable them for the rest of the tournament, but then the very next game it crashed again so it was probably only part of the problem.

The TC could be a thing to look at too. In my blitz testing with 4' + 2" it also lost a few games on time, even with 5 sec margin in ChessGUI.

My guess would be something related to NUMA... not sure.

In any case, I really hope Andrea gets to the root of the problem because this is very unfortunate for him.

Best,
Martin

[Robert Flesher]

Vitruvius 1.19 has now played 26 games without loss! Quite impressive, when Houdini has lost 3! What are statistics experts saying - new star or not?

I'm not an expert in statistics, but what matters is the full score, not how many games you did not lose.

Still, Vitruvius is doing impressively well and proving itself strong in this tournament.

I said many many, months ago when beta testing Vitruvius that it is a genius at getting passed pawns. I hope perhaps some of my input that something to do with this Komodo is also doing very well, good luck!

[S.Taylor]

I find hard to understand how Chiron can crash so often that it had to be withdrawn. None of the testers, as far as I know, have reported any such problems. Chiron 1.5 running on my system also Carl

It does crash on just 4 cores, but not very often. I guess 16 cores makes the problem worse. I didn't know Martin had withdrawn it though.

But it's not playing with 16 cores - just 4 cores.

OK, well it crashes on 4 too. But I'm surprised it had to be withdrawn, the crashes don't occur that often. An over-reaction I think.

If you would have been watching game 25 chiron-quazar, i expect you too would understand any over-reaction.
It was a game which may deserve a brilliancy prize if there was one for this tournament. Not for losing, but for winning. brilliantly.

It is shocking to me that nothing could be done to reset the game and clocks from that point and to play the remainder of the game.... probably only a few moves.

So i can well understand the author wanting to withdraw it now, since no mercy was given for this brilliantly winning game.

I also know that i was waiting over an hour for the next move and knew nothing about why chiron was taking so long.

[Laskos]

Vitruvius 1.19 has now played 26 games without loss! Quite impressive, when Houdini has lost 3! What are statistics experts saying - new star or not?

I'm not an expert in statistics, but what matters is the full score, not how many games you did not lose.

Still, Vitruvius is doing impressively well and proving itself strong in this tournament.

I said many many, months ago when beta testing Vitruvius that it is a genius at getting passed pawns. I hope perhaps some of my input that something to do with this Komodo is also doing very well, good luck!

Underperforming Houdini 3 has some 30% chances to not qualify for the next round.

[bupalo]

SO vitruvius in out and the strongest programs are going to make a very interesting stage 4. Very surprice ofr stockfish we are going to see a new spring for this program

[Martin Thoresen]

.
Final standings after 14 of 14 rounds:

Code: Select all

N Engine            Pts    SB Ho St Ry Ko Vi Hi Qu Ch 

1 Houdini 3        10.0 56.50    == 01 == =1 1= 11 11 
2 Stockfish 250313 10.0 54.75 ==    0= == =1 11 11 11 
3 Rybka 4.1         9.5 55.50 10 1=    == == == 11 11 
4 Komodo 4534       9.5 54.50 == == ==    01 11 1= 11 
5 Vitruvius 1.19    7.0 38.00 =0 =0 == 10    0= 1= 11 
6 Hiarcs 14         6.0 29.00 0= 00 == 00 1=    == 11 
7 Quazar 0.4        4.0 14.25 00 00 00 0= 0= ==    11 
8 Chiron 1.5        0.0  0.00 00 00 00 00 00 00 00    

Finished games: http://www.tcec-chess.net/stage_3.php

Congratulations to Houdini, Stockfish, Rybka and Komodo for qualifying for Stage 4 that will start in a few days.
.

[gladius]

.
Final standings after 14 of 14 rounds:

Code: Select all

N Engine            Pts    SB Ho St Ry Ko Vi Hi Qu Ch 

1 Houdini 3        10.0 56.50    == 01 == =1 1= 11 11 
2 Stockfish 250313 10.0 54.75 ==    0= == =1 11 11 11 
3 Rybka 4.1         9.5 55.50 10 1=    == == == 11 11 
4 Komodo 4534       9.5 54.50 == == ==    01 11 1= 11 
5 Vitruvius 1.19    7.0 38.00 =0 =0 == 10    0= 1= 11 
6 Hiarcs 14         6.0 29.00 0= 00 == 00 1=    == 11 
7 Quazar 0.4        4.0 14.25 00 00 00 0= 0= ==    11 
8 Chiron 1.5        0.0  0.00 00 00 00 00 00 00 00    

Finished games: http://www.tcec-chess.net/stage_3.php

Congratulations to Houdini, Stockfish, Rybka and Komodo for qualifying for Stage 4 that will start in a few days.
.

Great tournament Martin!! Looking forward to stage 4 .