What happend to TCEC?

[Henk]

High quality chess means that almost all games end in a draw. Also no end games are played for they consult table bases all the time.

Use engines for analysis only that's all there is left.

Playing games is only for developers to test their engines.

[hgm]

For games on the 'ultimate hardware' you have to watch Johnny play at WCCC. TCEC is a far cry from 2048 cores.

That's just BS. Jonny on 2048 cores is weaker than SF on any i7 laptop. So what were you trying to tell us?
2048 cores is anyway totally pointless, almost zero difference in strength (certainly less than 5 Elo) between 2048 of 256 cores.

You seem to have some problems understanding plain English. I said Johnny's hardware at WCCC was much better than what is used for TCEC. I made no claim whatsoever on the resulting level of play. The point is that describing TCEC as 'ultimate hardware' is just laughable.

As to your claims about how Johnny on 2048 cores would fair against Stockfish on an i7: you obviously have no clue. You didn't try it, you have no idea how Johnny uses the cluster and how it scales. You are just making things up.

[Modern Times]

High quality chess means that almost all games end in a draw. Also no end games are played for they consult table bases all the time.
.

With the top engines yes. I used to sometimes run 8CPU long time control tournaments with Komodo and Stockfish but I got bored with the 90% draw rate. I've no desire to run any again unless it is something special, like chess960 and at a shorter time control.

[Dann Corbit]

High quality chess means that almost all games end in a draw. Also no end games are played for they consult table bases all the time.

Use engines for analysis only that's all there is left.

Playing games is only for developers to test their engines.

Draw percentage for TCEC is about 47%, though the finals will obviously be much higher.

I believe in the exciting draw, though I do admit that beauty is in the mind of the beholder.

I also highly appreciate the use of engines for analysis. However, I dispute that this is their only use.

[Modern Times]

The point is that describing TCEC as 'ultimate hardware' is just laughable.
.

For the average person who at most has an 8 core single socket CPU and will never run a dual socket workstation class machine with 20+ cores, it probably does qualify as "ultimate" to them even though that is not the dictionary definition. "Very high end" is a better description of course.

Will 20+ desktop cores be standard for all of us in a few years ? Who knows,

[Dann Corbit]

The issue is that WCCC use uneven hardware, whereas TCEC does not.

I don't really have an issue with that. That is pretty much the tradition of the WCCC.

That way a Mac and a Windows program and a program on Solaris can all compete.

Ernst Heinz ran on DEC hardware, for instance.

In the case of WCCC, the goal is the strongest possible combination of book + engine + hardware. That is a laudable goal, just a very different goal from TCEC.

Additionally, if you really want to know which engine is stronger then it is not TCEC to examine but CCRL and CEGT. TCEC, like WCCC, are used to determine a champion, and that is all that they accomplish.

Good, clean fun if you ask me.

[Dann Corbit]

The point is that describing TCEC as 'ultimate hardware' is just laughable.
.

For the average person who at most has an 8 core single socket CPU and will never run a dual socket workstation class machine with 20+ cores, it probably does qualify as "ultimate" to them even though that is not the dictionary definition. "Very high end" is a better description of course.

Will 20+ desktop cores be standard for all of us in a few years ? Who knows,

This is close enough to ultimate hardware:
CPUs: 44 Cores -> 2 x Intel Xeon E5 2699 v4 @ 2.8 GHz
Look to see what it costs to build a machine configured like TCEC.

The Jonny system with the zillion cores system it used for WCCC would be literally pulverized by:

Stockfish
or
Houdini
or
Komodo

on TCEC hardware.

I guess in a thousand games Jonny would be lucky to win one.

Jonny is a good engine.
But not in the same class as Stockfish or Houdini or Komodo.

And when you have giant piles of cores (for the most part) they just make the tree a little bushier. They won't give you much greater depth. At some point, they will probably even subtract Elo.

It would be interesting if Jonny was public to see how it scales for cores beyond 4 (e.g. the 12 core tests of CEGT). I guess that almost all the cores past 64 are just adding heat to the room. But it is also possible that Jonny has some revolutionary SMP algorithm. Since it is private (past version 4), there is no way to know.

[hgm]

For the average person who at most has an 8 core single socket CPU and will never run a dual socket workstation class machine with 20+ cores, it probably does qualify as "ultimate" to them even though that is not the dictionary definition. "Very high end" is a better description of course.

Computer-Chess enthousiasts tend to have above-average powerful hardware themselves. 20 cores is just 2.5 times as much as 8 cores. A doubling of the TC would already be enough to catch up in terms of play level.

I already see a rift appearing: Dann likes the extremely high level of play caused by the very slow TC and powerful hardware. Many others, however, say here they would like faster games, and lower level of play to reduce draw rate. (I would side with the latter; even if it concerned an interesting variant rather than orthodox Chess, I would never be interested to watch an event similar to TCEC.)

Actually the way used in TCEC to obtain the games of that level is an enormous waste of resources (and thus money). Speed is hardly of any importance for producing Chess games between engines. It is not like weather prediction or ballistic-missile defense. Using highly overclocked high-end CPUs gives you much fewer instructions per kilowatt-hour than using a somewhat slower CPU within specs. Who cares if it takes 8 hours or 8 days to produce a game? No new versions of the engines will have appeared within that time either way. Once you have the game, you can replay it on your web server in accelerated way, compressing 8 days of play into a one- or two-hour broadcast. People would never know the difference.

From an engineering perspective this is a much more sensible way to accomplish the task at hand. Several average (= cheap) computers can be used to produce in parallel. There is no shortage of such computers, and people who own them could team up to provide the games.

[hgm]

The Jonny system with the zillion cores system it used for WCCC would be literally pulverized by:

Stockfish
or
Houdini
or
Komodo

on TCEC hardware.

I guess in a thousand games Jonny would be lucky to win one.

I wonder how you can be so confident on this. You certainly did not try it. (That is a friendly way of saying: you are just dreaming.) In reality Komodo had great trouble beating Johnny at WCCC, even though it was using a 60 core system. All the regular games ended draw. It took 3 playoffs at successively shorter TC for Komodo to finally win one.

That hardly sounds like 'being pulverized'.

And this is fact, rather than purely wishful thinking...

[mjlef]

For games on the 'ultimate hardware' you have to watch Johnny play at WCCC. TCEC is a far cry from 2048 cores.

That's just BS. Jonny on 2048 cores is weaker than SF on any i7 laptop. So what were you trying to tell us?
2048 cores is anyway totally pointless, almost zero difference in strength (certainly less than 5 Elo) between 2048 of 256 cores.

How did you come to those conclusions? With what evidence?

I have spoken a bit with Johannes Zwanzger, The Jonny author, so I know a few of the things it does. One is multiple pondering. Having so much hardware available lets you ponder on pretty much any move your opponent might make. This would effectively double the amount of time a program would have to come up with a move. And a doubling of time, even at this number of cores, is certainly worth more than 5 elo. And he does a lot of other very clever things (but I will leave it up to him to mention them). On this hardware, Jonny is very strong, and I would consider it to be our strongest opponent this year (since I do not know much about this year's Shredder which also has probably gained a lot of elo).

Mark