What happend to TCEC?

[hgm]

50% of your time (assuming both engines take about the same time in the game) the engines are not pondering. So the ponder split would have 0% benefit during the normal search phase.

During ponder, you only benefit if the ponder guess was wrong, so 30% of the time. This also ignores that the extremely similar ponder positions will have lots of similar hash entries, so shifting to a new node is not a 100% cost. But we ignore the transposition table for now.

I see the best possible improvement as 0.5 * 0.3 = 0.15 = 15%.

What is a 15% speedup worth in Elo?
Certainly it is nothing to sneeze at, but it is not something I would quake in fear at.

You have no clue how Johnny uses the cores, and yet you make statements on what it could achieve by it. One of the things Johnny does is speculative deep pondering. Rather than thinking on the 3rd, 4th, ... move for ply 1, it thinks on the most likely moves for ply 3, 5, 7... as well. As a result the move it plays has usualy been though on for 5-10 times as long as the opponen's average think, with still more cores than the opponent has in total. Because the thinking already started 7 plies ago. And with help from hash-table input provided by other sub-clusters of cores.

[Guenther]

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

Where do the numbers come from and how old is Jonny 8.00?
I see no reference and no details for your numbers.

Guenther

[Evert]

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).

I think this is quite interesting.
Clearly algorithms most of us are familiar with will not scale to the insane number of cores available to Jonny (#cores >> #moves). Conversely, Jonny's algorithms probably do not scale to the number of cores most programs run on (#cores <= #moves), so in that sense it cannot be considered independently from its hardware.
I wonder: is the raw computing power sufficient that you can afford to be less selective in the same amount of time (increasing the Elo-per-ply)? I guess that depends on the speed of each core.

I see people argue (essentially) that Jonny's Elo/kWh is lower than that of conventional engines like Stockfish or Komodo. While that's a valid concern in the current climate, it seems to rather miss the point of something like this project.

[hgm]

Where do the numbers come from and how old is Jonny 8.00?
I see no reference and no details for your numbers.

It is obviously that some people here are completely clueless as to what they are talking about, and just fabricate the wildest stuff to provide 'evidence' for their pre-conceptions. As Uri already remarked, the Johnny playing at WCCC need not be Johnny 8.00, and in fact is quite unlikely to be that. So any argument based on the rating of Johnny 8.00 is null and void. Any argument based on scaling of SMP is void, as Johnny doesn't use the cores in the way they assume.

They know nothing, and blabber just nonsense.

Fact is that they never played Johnny-WCCC on 2048 cores against Stockfish. Fact is that Johnny did play 60-core Komodo on WCCC, and that the latter scored only one win and 7 draws against it.

And all this is off-topic as far as TCEC is concerned. Hardware in TCEC-like events is limited, and 2048-core clusters to conduct them a completely unrealistic goal at the current stage of technology.

[Milos]

Where do the numbers come from and how old is Jonny 8.00?
I see no reference and no details for your numbers.

And how old is SF8?
You believe that in the same time time frame a single core version of Johnny gained more Elo than SF. Yea sure. That must be the reason why in years the gap between SF and Johnny is increasing.
All wishful thinking. Johnny in WCCC is somehow magically 300 Elo stronger than Johnny 8. It's just speculative blabbering without substance offering zero arguments.

[hgm]

And the only one that even mentions it is you...

And thinking that the Johnny version used at WCCC is 300 Elo stronger is already close to delusional denial of facts. If it were 300 Elo stronger, in addition to having a time-odds advantage of a factor 10, it should have crushed Komodo 7-3 or so. But that did not happen.

[Adam Hair]

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

Where do the numbers come from and how old is Jonny 8.00?
I see no reference and no details for your numbers.

Guenther

That looks like the CEGT 40/20 MP list.

[Guenther]

Where do the numbers come from and how old is Jonny 8.00?
I see no reference and no details for your numbers.

And how old is SF8?
You believe that in the same time time frame a single core version of Johnny gained more Elo than SF. Yea sure. That must be the reason why in years the gap between SF and Johnny is increasing. :lol:
All wishful thinking. Johnny in WCCC is somehow magically 300 Elo stronger than Johnny 8. It's just speculative blabbering without substance offering zero arguments.

What a stupid question... I just asked for facts and not for your speculations.
Actually I have no idea, if Jonny 8.00 is one, two or three years old
and I won't speculate about its smp capabilities, if I cannot test or watch it.
(The last free version is 4.00 over 7 years old...)
OTH Mark Lefler already wrote more than one year ago
about Jonnys progress in smp for the WCCC 2016.
http://talkchess.com/forum/viewtopic.php?t=60620

Also there is simply no reference and nothing more than plain rating numbers thrown in by Dann.

Moreover I have not stated any opinion about a possible margin of the strength gap at all, because I am only interested in facts. All your interpretation is imagined.

[BrendanJNorman]

Whenever you see a 9 page topic on TalkChess - NO! It's not a popular topic...

Grown men are insulting each other.

[hgm]

And then there's Milos too...