TCEC Super Final News...

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Steve Maughan
Posts: 1061
Joined: Wed Mar 08, 2006 7:28 pm
Location: Florida, USA
Contact:

TCEC Super Final News...

Post by Steve Maughan » Sun Oct 30, 2016 8:36 am

From the TCEC Facebook page...

After two preliminary rounds, qualifications, candidates stage and a rapid, it is time for the TCEC Season 9 Superfinal.

* The Superfinal is scheduled to start in the period 5th-7th November 2016. The delay is due to personal problems of the TD, no further such is expected
* The exact start date depends on when the server providers will be ready. The TCEC team is aiming for a 44 core server and we hope it will be ready on time.
* The news service on Chessdom about TCEC will also resume starting Nov 1st.
* Houdini and Stockfish are expected to submit their versions for the Superfinal no later than November 4th. If no new versions are submitted, the ones from Stage 3 will be used
* The Superfinal is the culmination of Season 9, but it will also be part of the preparation for Season 10, which we plan to start early next year.
http://www.chessprogramming.net - Maverick Chess Engine

User avatar
Eelco de Groot
Posts: 4162
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: TCEC Super Final News...

Post by Eelco de Groot » Sun Oct 30, 2016 9:05 am

Thanks Steven for head's up!
  • Good to hear from Martin Thoresen!
  • Time to make final changes is a bit short, but we've had enough time I suppose :)
  • 44 cores means that a NUMA-ready program will have either a significant advantage or it should be certain that Hyperthreading is switched off in BIOS (See Lukas Cimiotti's and Ronald's comments in Rybkaforum). Or is that only if you run more than 44 threads? Now I'm not sure anymore... Stockfish is not NUMA ready, asmFish is...
Last edited by Eelco de Groot on Sun Oct 30, 2016 9:11 am, edited 2 times in total.
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

JJJ
Posts: 1287
Joined: Sat Apr 19, 2014 11:47 am

Re: TCEC Super Final News...

Post by JJJ » Sun Oct 30, 2016 9:05 am

That is really great news :) People were worried.
Also it's great to know about season 10 !

Nay Lin Tun
Posts: 522
Joined: Mon Jan 16, 2012 5:34 am

Re: TCEC Super Final News...

Post by Nay Lin Tun » Sun Oct 30, 2016 9:35 am

It is sad to know that stockfish is much slower than ASM fish and NUMA is still not ready yet. Even if ASM fish is 100 elo stronger than stockfish, it is the stockfish that successfully reached to superfinal , and it is not eligible to replace with ASMfish. Hopefully ASMfish be accepted and replaced as the main stockfish project for season 10. And still hopefully stockfish be a good competitor until the last game against new Houdini 5 under such conditions.

Sean Evans
Posts: 1777
Joined: Thu Jun 05, 2008 8:58 pm
Location: Canada

Re: TCEC Super Final News...

Post by Sean Evans » Sun Oct 30, 2016 9:38 am

Steve Maughan wrote:From the TCEC Facebook page...

* The Superfinal is scheduled to start in the period 5th-7th November 2016. The delay is due to personal problems of the TD, no further such is expected.
Prison sentence?

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: TCEC Super Final News...

Post by Lyudmil Tsvetkov » Sun Oct 30, 2016 9:43 am

Nice to hear about the 44-core upgrade. With the VLTC this will boost strength somewhat.

But, I prefer to have a 100-elo stronger version and play on single core rather than having a 100-elo weaker engine playing on 44 cores.

I suspect SF 8 single core will win SF 7 16 cores and draw SF 7 44 cores in a 100-game match, or?

User avatar
Eelco de Groot
Posts: 4162
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: TCEC Super Final News...

Post by Eelco de Groot » Sun Oct 30, 2016 10:37 am

Nay Lin Tun wrote:It is sad to know that stockfish is much slower than ASM fish and NUMA is still not ready yet. Even if ASM fish is 100 elo stronger than stockfish, it is the stockfish that successfully reached to superfinal , and it is not eligible to replace with ASMfish. Hopefully ASMfish be accepted and replaced as the main stockfish project for season 10. And still hopefully stockfish be a good competitor until the last game against new Houdini 5 under such conditions.
No, of course asmFish can not replace Stockfish. For the new season 10, I don't see asmFish becoming the main branch, and two essentially identical programs in TCEC is not going to happen either.

This really is due only to a Windows limitation, not really anything related to NUMA. The problem would not exist on the same system running the tournament under Linux for instance.

It just will mean a foregone conclusion, that the program that would set affinity to the physical cores will win the final. No question about that, unless the TD would decide to turn hyperthreading off in the BIOS if both opponents would agree to that. It is possible that Houdart also is not ready, but we don't know much about Houdini 5 yet.
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

Rochester
Posts: 55
Joined: Sat Feb 20, 2016 5:11 am

Re: TCEC Super Final News...

Post by Rochester » Sun Oct 30, 2016 12:25 pm

The better is two cluster with the ponder. No share the system in the weird way like before. The real chess player also think all the time and use the own brain. You think they can do so now?

Lyudmil Tsvetkov
Posts: 6052
Joined: Tue Jun 12, 2012 10:41 am

Re: TCEC Super Final News...

Post by Lyudmil Tsvetkov » Sun Oct 30, 2016 1:30 pm

My prediction, based on gut feeling and some illogical maths: 20-30 wins in all, depending on the book choice, though not certain how they will be split (also depends very much on the book choice).

If 2-move short neutral book is used, I go for 30 wins, slightly favouring Houdini.

If a larger opening book is used with prevalence of open positions, or Sicilians, I go for 20 wins, with considerable SF advantage.

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: TCEC Super Final News...

Post by APassionForCriminalJustic » Sun Oct 30, 2016 2:12 pm

Nay Lin Tun wrote:It is sad to know that stockfish is much slower than ASM fish and NUMA is still not ready yet. Even if ASM fish is 100 elo stronger than stockfish, it is the stockfish that successfully reached to superfinal , and it is not eligible to replace with ASMfish. Hopefully ASMfish be accepted and replaced as the main stockfish project for season 10. And still hopefully stockfish be a good competitor until the last game against new Houdini 5 under such conditions.
Gentlemen calm down. Are we forgetting that Stockfish has 'Use per-thread counterMoveHistory' and also the 'Per-thread TB hit counters? Couple those two babies with Stockfish's LAZY-SMP and we have some serious speed. Houdini will absolutely have no advantage in this superfinal. There is no need to be concerned with an engine that was 250+ Elo behind Stockfish. It does not matter what opening you throw out there - the advantage for Stockfish will not change. With Komodo out of the way in this Season 9 Stockfish will mop the floor up in what I believe will be the most lopsided superfinal as of yet. Let the games do themselves justice now.

Hyperthreading will not be enabled.‎

You should spin your sentence around to say, "Will Houdini development be a good competitor"? And I do not say that jokingly.

- It is good to have ideal thoughts, beliefs, and virtues - but logic is truth. - APassionForCriminalJustice-

Post Reply