go perft command #2091
go perft command #2091
Conversation
borg323
added
the
not for merge
| Perft(ChessBoard(current_position_.fen), *params.perft, true); | ||
| const auto time = std::chrono::duration_cast<std::chrono::milliseconds>( | ||
| std::chrono::steady_clock::now() - start); | ||
| CERR; |
There was a problem hiding this comment.
The standalone 'CERR;' statement appears redundant and may be unintentional; consider removing it or clarifying its purpose with a comment.
| CERR; |
Copilot uses AI. Check for mistakes.
| auto new_board = board; | ||
| new_board.ApplyMove(move); | ||
|
|
||
| new_board.Mirror(); |
There was a problem hiding this comment.
Review the call to 'new_board.Mirror()' to ensure it aligns with the intended perft calculation, as altering board orientation may affect the expected move count.
| new_board.Mirror(); |
Copilot uses AI. Check for mistakes.
| @@ -47,6 +47,7 @@ struct GoParams { | |||
| std::optional<int> depth; | |||
| std::optional<int> mate; | |||
| std::optional<int> nodes; | |||
| std::optional<int> perft; | |||
There was a problem hiding this comment.
Would it make more sense to have just perft command rather than go perft, and then route to a separate function rather than to EngineClassic?
Especially with many search algorithms, bringing perft closer to search complicates more than help.
Even if we'd keep go perft instead of perft, I'd prefer Uci layer to call completely different function rather than something in the Engine class.
There was a problem hiding this comment.
It is go perft because it is modeled after the stockfish version. Regarding the rest, I wasn't planning on getting this merged. If you think it is worth having then I'll do my best.
There was a problem hiding this comment.
I believe it makes sense to have it, but probably not through uci but from the command line ./lc0 perft similarly to bench and backendbench.
borg323
removed
the
not for merge
No description provided.