Stockfish vs. Lc0: IMHO disappointing result for Lc0

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

User avatar
Ovyron
Posts: 4556
Joined: Tue Jul 03, 2007 4:30 am

Re: Stockfish vs. Lc0: IMHO disappointing result for Lc0

Post by Ovyron »

Yeah, at this point you'd rather use regular Leela with one of the stronger nets.

No doubt Leela has caught up to Stockfish, and depending on hardware, since kiudee settings were found some people have been reporting that Stockfish had been getting destroyed in games.

By the way things are progressing, it's possible CPU Leela is going to eventually catch up to Stockfish :shock:
User avatar
mvanthoor
Posts: 1784
Joined: Wed Jul 03, 2019 4:42 pm
Location: Netherlands
Full name: Marcel Vanthoor

Re: Stockfish vs. Lc0: IMHO disappointing result for Lc0

Post by mvanthoor »

Guenther wrote: Sat Feb 15, 2020 9:11 am 5. You ran only 20 games in your first test, which was a lottery for the +1 =15 -4 result... (and you did it again in your new test - 20 games is nothing)
Yes, I know. I wanted to just get a general idea of Fat Fritz's strength as compared to Stockfish.

For that, 20 games ending in 11-9 for Stockfish is enough. The conclusion is that "Stockfish on a 6820HQ CPU is probably somewhat stronger than Fat Fritz on an M2000M" (which is a weak graphics card compared to current day's top cards). I understand that in another 20 games, it might have turned out to be 20-20 for both engines. Fat Fritz, even on such a weak graphics card, is in the playground of the stronger engines. That's all I wanted to know.
Author of Rustic, an engine written in Rust.
Releases | Code | Docs | Progress | CCRL
User avatar
Guenther
Posts: 4606
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: Stockfish vs. Lc0: IMHO disappointing result for Lc0

Post by Guenther »

mvanthoor wrote: Sat Feb 15, 2020 2:43 pm
Guenther wrote: Sat Feb 15, 2020 9:11 am 5. You ran only 20 games in your first test, which was a lottery for the +1 =15 -4 result... (and you did it again in your new test - 20 games is nothing)
Yes, I know. I wanted to just get a general idea of Fat Fritz's strength as compared to Stockfish.

For that, 20 games ending in 11-9 for Stockfish is enough. The conclusion is that "Stockfish on a 6820HQ CPU is probably somewhat stronger than Fat Fritz on an M2000M" (which is a weak graphics card compared to current day's top cards). I understand that in another 20 games, it might have turned out to be 20-20 for both engines. Fat Fritz, even on such a weak graphics card, is in the playground of the stronger engines. That's all I wanted to know.
You also simply omitted point 4 of my answer too, why, didn't you like it too?
https://rwbc-chess.de

trollwatch:
Chessqueen + chessica + AlexChess + Eduard + Sylwy
User avatar
mvanthoor
Posts: 1784
Joined: Wed Jul 03, 2019 4:42 pm
Location: Netherlands
Full name: Marcel Vanthoor

Re: Stockfish vs. Lc0: IMHO disappointing result for Lc0

Post by mvanthoor »

Guenther wrote: Sat Feb 15, 2020 3:34 pm You also simply omitted point 4 of my answer too, why, didn't you like it too?
Liking or not liking point 4 didn't have anything to do with it. You said the current Lc0 nets are stronger than the ones of about half a year ago. I saw no reason to comment on that, as it's undoubtedly true. I'd be amazed if the networks had become weaker :P
Author of Rustic, an engine written in Rust.
Releases | Code | Docs | Progress | CCRL