mmt Vs. Ovyron (G4 D5 BG2)

[Ovyron]

Oh, in that case Komodo's eval is pure nonsense.

[zullil]

I'd prefer to see what Komodo says at depth 60, but for that I'd need to run it on Louis's machine.

[d]r2qkb1r/pQ1nnppp/2p5/3p4/6b1/2N5/PP1PPPBP/R1B1K1NR w KQkq - 0 8
Komodo-13.2.5 says -1.32 at depth 32. I'm using MultiPV = 3, so depth 60 probably won't happen.

[jdart]

IMO game play (especially with learning of past results) is a more reliable testing method that trying to do search from the root position, even with a very strong engine or at long time control. I think there are many cases where it is difficult to get the "truth" of an opening from such a search. The engines miss things. A-B engines are throwing out huge hunks of the search tree, and NN searchers are just not visiting that much of the tree at all.

[Alayan]

Playing games also throws out huge chunks of the search tree.

[jdart]

Playing games also throws out huge chunks of the search tree.

Yes, it does, but with each move the horizon is pushed out farther.

[Alayan]

Playing games is a really inefficient method to root out blunders in the lines that are played. It's decent to quickly assess a position (for a tuning set, that's all you need), but to get the truth out of a position, I wouldn't rely on it.

Now, pure search from the root is also rather inefficient, that's for sure.

[jp]

For whatever it is worth, ChessDBCN considers Ne7 cxd5 exd5 d4 Rb8 Qa6 Rb6 Qd3 Ng6 as a forced line.

What is this ChessDBCN? Can you tell me a bit about it?

[Ovyron]

What is this ChessDBCN? Can you tell me a bit about it?

It's a live online database:

https://www.chessdb.cn/queryc_en/

You can explore the chess lines that it has and request analysis for the moves that it doesn't; it will expand the moves you insert automatically and backsolve their scores to propagate them backwards so previous positions have them ranked correctly in real time.

[Ovyron]

Time for a conditional:

IF

8.d4

THEN

8...Rb8

[koedem]

Which network will Lc0 be using? Default or specialized settings? Also, what endgame tables will be available to your engines?

Sergio 384x30-t40-1808 with specialized settings: in-between default and "Kiudee." If you have any suggestions about settings, please let me know.

I have 6-men tablebases but I will download 7-men and put the ones that are needed on an NVMe SSD (if it gets to this point).

If you're looking to run very long searches, the standard Leela binary will not be very useful since AFAIK they haven't changed the data type of their Q. (at least last time I checked Q was a float which means that for long searches (> 100MN) the eval and with it the search will become stuck due to rounding errors essentially)
For such long searches I still use a relatively old forked version that changes Q to be a double.