Chess.com 2018 computer chess championship

[Guenther]

After a little bit ( 5 Min) of probing their HTML code, I found a temp solution to the problem.

Grab it from here: https://newman.chess.com/games.pgn

PS:

Pro Tip: If you use the non secure version of the link http://newman.chess.com/games.pgn it will also work in Hiarcs Chess Explorer and you can then use your favorite engine to analyze the current game while it is in progress.

Thanks!

From the games it looks like Ivanhoe crashed two times (vs. LC0 always in lost pos.) and Laser one time,
but still the games were simply repeated until there was another result than 'unterminated'?

Can someone verify this? I have games here and stripped off the pv with a macro.

[Guenther]

From the games it looks like Ivanhoe crashed two times (vs. LC0 always in lost pos.) and Laser one time,
but still the games were simply repeated until there was another result than 'unterminated'?

Can someone verify this? I have games here and stripped off the pv with a macro.

I think it is a bit unfortunate to decide about the unterminated games after the tournament.
This means the current standing is quite unreliable.

• Engine crash:
  In the event that one of the participating engines crashes, stalls, or disconnects during a game, our tournament software will automatically replay the game.
  The replayed game will be for exhibition only and only the original game will be scored.
  At the end of the tournament, the game in question will be examined by the tournament director and the Chess.com team to determine proper adjudication, as follows:
  If the engine that crashed was in a winning position, the game will be adjudicated as a draw.
  If the position was a clear draw, the game will be adjudicated as a win for the crashing engine's opponent.
  Otherwise, the game will be adjudicated as a loss for the crashing engine.
  If a neutral event occurs due to a technical error by the CCCC system or anything outside the two engines playing, the game will be restarted and replayed in it's entirety.
  The original game will be adjudicated and will count if the result of the restarted game does not match the result of the original game.

[jp]

The standings now say lc0 2.5/3
but it has three draws against Fizbo, Gull & Ivanhoe.

Can they not add up numbers or are they giving handicap points?

[AdminX]

The standings now say lc0 2.5/3
but it has three draws against Fizbo, Gull & Ivanhoe.

Can they not add up numbers or are they giving handicap points?

See: http://blog.lczero.org/2018/09/cccc-rou ... and-3.html

[jp]

Thanks. Wow.

[elcabesa]

I'm trying to understand vajolet crashes at tcec, cccc conditions seems very similar to tcec eccept ram which is 8 GB instead of 16 GB.
There is a list of crashed engines at cccc tournament?

[Gary Internet]

They are only using Fizbo 1.9 instead of 2.0. I was told in the chat that they're doing this because Fizbo 2.0 only runs on Linux and that it's not stable.

[DrCliche]

Code: Select all

Ethereal 10.86 (bmi2)

setoption name Threads value 46
info string set Threads to 46
go movetime 10000
info depth 25 seldepth 32 score cp 24 time 3125 nodes 206966264 nps 66208000 tbhits 0 hashfull 996 pv d2d4 g8f6 c2c4 e7e6 g1f3 d7d5 b1c3 f8b4 c4d5 e6d5 c1g5 e8g8 e2e3 b8d7 f1d3 c7c5 e1g1 b4c3 b2c3 c5c4 d3c2 d8a5 g5f4 f6e4 c2e4
bestmove d2d4 ponder g8f6

setoption name Threads value 92
info string set Threads to 92
go movetime 10000
info depth 22 seldepth 31 score cp 26 time 7984 nodes 533074064 nps 66759000 tbhits 0 hashfull 1000 pv d2d4 d7d5 g1f3 e7e6 c2c4 g8f6 b1c3 c7c5 c1g5 c5d4 f3d4 d5c4 e2e3 b8d7 f1c4 f8e7 e1g1 e8g8 c4e2 a7a6 d4f3 h7h6 g5f4 g7g5
bestmove d2d4 ponder d7d5

nps +1%

Code: Select all

Stockfish 130818 (bmi2)

setoption name Threads value 46
go movetime 10000
info depth 27 seldepth 26 multipv 1 score cp 46 upperbound nodes 720732280 nps 72066021 hashfull 999 tbhits 0 time 10001 pv d2d4 g8f6
bestmove d2d4 ponder g8f6

setoption name Threads value 92
go movetime 10000
info depth 26 seldepth 32 multipv 1 score cp 45 nodes 916696131 nps 91660447 hashfull 999 tbhits 0 time 10001 pv d2d4 g8f6
bestmove d2d4 ponder g8f6

nps +27%

There's something hinky with those numbers. How can Ethereal have more than double the nodes with almost the exact same nodes per second? Even more concerning is that the numbers being reported in the actual tournament are inconsistent with the numbers reported here. In your test, Stockfish and Ethereal got 72.07 Mnps and 66.21Mnps, respectively, while contemplating the starting position on 46 threads. In the actual game between those two engines, they got 54.85Mnps and 46.27Mnps at the start. That's only 76.1% and 69.9% of the claimed performance.

Regardless, in the interest of generating "the best possible chess", it seems a peculiar decision to hamstring the top engines for the failings of their juniors. (And of course, Leela, with a $40,000+ machine all to her lonesome, must deal with none of this nonsense.)

Wow I wasn't aware those Graphic cards were so expensive.
Yes that's quite a disproportionate bang for the buck.

Pardon the uncivil comparison here, but honestly, this is Donald Trump levels of absurdity and denial of reality. People have been telling you this and more, plainly and in no uncertain terms, for weeks, on this forum and elsewhere, ever since information about the hardware was made public.

Yes JP, we're all aware that running an engine on a high-end GPU system is much more expensive than on a high-end CPU system.
But does that mean we shouldn't do it?

But you don't have a high-end CPU system. You've failed to outperform a $720 CPU.

I haven't heard one engine author complain yet about the cost ratio...on the contrary, they're very excited to participate.

If you haven't heard any engine authors laugh or complain about various absurdities at CCCC, including but not limited to the cost disparity, the most charitable thing I can say is that you are willfully ignorant. A less charitable interpretation might be that in many instances you are simply not telling the truth.

[AndrewGrant]

There's something hinky with those numbers. How can Ethereal have more than double the nodes with almost the exact same nodes per second? Even more concerning is that the numbers being reported in the actual tournament are inconsistent with the numbers reported here. In your test, Stockfish and Ethereal got 72.07 Mnps and 66.21Mnps, respectively, while contemplating the starting position on 46 threads. In the actual game between those two engines, they got 54.85Mnps and 46.27Mnps at the start. That's only 76.1% and 69.9% of the claimed performance.

The numbers are fine. UCI outputs are not at the end of 10 seconds, but ad the last completed search before 10 seconds. A better test would have been to issue go depth <x>, but its no issue.

As the for Mnps difference, well this is all due to pondering I would assume. ~70% would actually impress me for hyperthreads, especially since I often rag on Intel's HT (more reasons than just the speed).

[jp]

You know, there are are many types of people...ones that jump in, work hard, and make something happen...
ones that graciously accept what's been given for free, and make the best of it and enjoy it...
and ones that stand on the sidelines, waving their arms, complaining about everything.

Please tell us which type of person you are. Do you mean you are jumping in, working hard & giving stuff for free?
chess.com is a commercial business & is doing this as part of their business. It's totally OK, but it's not a charity.
Are you being paid by chess.com or doing TD for them for free? No problem either way, but "graciously accept what's been given for free"??