What happend to TCEC?

[mjlef]

I don't think it would scale to 2400 core At least for now.
And I d be happy to see Komodo with 2400 core.

Most testing (such as from fastgm.de) show a big diminishing return on more cores after something like 16 core. I would expect it to reach elo at a high enough core count. Komodo would need to be heavily modified to take advantage of a huge machine like that. Something I am willing to do if I can get access to one for a few weeks.

[mjlef]

AKA Cato the Younger here.

...

My biggest problem right now is that I forgot my ChatWing password to my TCEC VIP nick. Yikes.

There is this guy in Moscow by the name of, I think, V. Putin. He might be able to help!

Maybe Anton should setup a gofundme for TCEC. It could be a measure of how much entertainment people would get out of a new Season.

this one has sufficient problems with Trump right now.

but maybe you know something the general public does not know.

what precisely happened?

It was just a (lame) joke. Both countries have a lot to worry about now.

[Henrik Dinesen]

AKA Cato the Younger here.

...

My biggest problem right now is that I forgot my ChatWing password to my TCEC VIP nick. Yikes.

There is this guy in Moscow by the name of, I think, V. Putin. He might be able to help!

Maybe Anton should setup a gofundme for TCEC. It could be a measure of how much entertainment people would get out of a new Season.

this one has sufficient problems with Trump right now.

but maybe you know something the general public does not know.

what precisely happened?

It was just a (lame) joke. Both countries have a lot to worry about now.

Make it several countries, make it many; and we have something close to facts. But never joke with chess!

Seriously: TCEC was there for a shorter while. Now it seems people miss it - why was it special (except from strong harware) ?!

[zenpawn]

Seriously: TCEC was there for a shorter while. Now it seems people miss it - why was it special (except from strong harware) ?!

I think a lot of it is due to the active, web-based chat, the presentation of various statistics (time usage, scores, PVs, etc.), and the ability to view the same for prior games.

[Dann Corbit]

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...

On your part.
You know for certain that Jonny is far weaker and you also realize that SMP loss makes the additional cores useless.

Jonny 8 is 300 Elo below the top engines:
http://www.cegt.net/40_40%20Rating%20Li ... liste.html

And thousands of cores are fairly irrelevant as all they are doing is adding heat, unless Amdahl's law has been voided:
https://en.wikipedia.org/wiki/Amdahl%27s_law

[Dann Corbit]

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

I guess that none of these showed up then:

1 Stockfish 8.0 x64 4CPU 3414
{ N/A 2 Komodo 11.01 x64 4CPU 3402}
4 Houdini 5.0 x64 4CPU 3380
6 Deep Shredder 13 x64 4CPU 3225
7 Fizbo 1.9 x64 4CPU 3207
8 Gull 3.0 x64 4CPU 3186
9 Ginkgo 2.0 x64 4CPU 3174
10 Andscacs 0.90 x64 4CPU 3174

Since Jonny is below them.
11 Jonny 8.00 x64 4CPU 3157

[Milos]

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).

Lol, how someone programming a strong engine can be so clueless???
Ever heard of Amdahl's law???
SF efficiency is 95.5% for up to 16 cores or on a single socket. I strongly doubt that Komodo or Johny have higher numbers. Once NUMA comes into play it drops down almost as low as 90%.
And what kind of argument is that with pondering, are you in kindergarten to compare apples and oranges???
If Jonny uses ponder (which btw doesn't double but can only increase to 70% effective time usage) other engine would be too so the comparison effectively becomes between 1024 and 128 cores at 1.7x original TC. So what? Again difference between 1024 cores and 128 cores is still less than 5 Elo.

[Milos]

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.

Stop BS-ing and stick to physics obviously you have no clue about parallel programming.
You didn't hear about Amdahl's law either and obviously never ever had any formal programming theory course in your life, you are just physicist wanna be programmer otherwise you'd know there is no magic trick that can make serial algorithm such a alpha-beta magically have parallel efficiency higher than 90 and few %.
So I don't need to know how Johny uses its cluster coz I know what is possible and what is not and I know that what you write is plain BS.
SF is 300 Elo stronger than Johnny and even infinite number of cores wouldn't help Johny get more than 20x effective speed up factor which can never account for 300 Elo and 2 doublings (4 cores i7).

[Milos]

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

Nope, you are the one having problems expressing yourself in English as already noted by quite a few native English speakers in CTF.
You said "you have to watch Johnny play at WCCC" directly implicating that Johnny's play on WCCC on 2048 cores is of some extremely high quality compared to TCEC. I simply indicated that quality of Johnny's play on WCCC is not better than quality of SF play on i7 laptop so why would anyone need to watch Johnny's play at WCCC???
The truth is that quality of Johnny's play at WCCC is laughable compared to quality of play of Komodo and SF at TCEC.

[Dann Corbit]

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

How can splitting the cores to do pondering double the speedup? That makes zero sense to me.

First, you must divide the cores for each new node pondered.
This could certainly be useful when you have lots of cores doing nothing, or next to nothing, as when you have thousands of them. So let's assume the division of the cores has zero cost.

IOW, 2400 cores divided by 24 ponder nodes gives 100 cores, which will have about the same speed as 2400 cores would have. So for the massive core count, this could give real tangible benefit, assuming that
we have lots of ponder misses.

Second (and far more important), the ponder node is right most of the time for all strong engines.
E.g., from http://inwoba.de/
we have this ponder hit table:

Code: Select all

B. Ponder hit and miss table:

 nr player                     :     hit     miss   hit%   games   pts%

  1 Stockfish 8                :     254      103   71.1       7   50.0
  2 Komodo 10.2                :     214       87   71.1       6   33.3
  3 Houdini 5.01               :     159       84   65.4       5   70.0

Average hit %  : 69.6
File           : 12LONG2L.pgn
Date           : 2016-12-05 08:25:43
Elapsed (h:m:s): 00:00:09

69.9%. With a slightly higher ponder hit (72.6) rate it seems possible that one gets better games
with 4 cores Ponder ON than with 8 Cores Ponder off!