Magic Chess almost ready for release...

[ZirconiumX]

...Need to clean up my code, etc.

I have to do several things, as bacteria plays better than it

info depth 1
info score -43 nodes 1 nps 90 seldepth 1 time 11 pv
info depth 2
info score 30 nodes 24 nps 510 seldepth 2 time 47 pv b8c6
info depth 3
info score -47 nodes 268 nps 2330 seldepth 3 time 115 pv b8c6 b1c3
info depth 4
info score 255 nodes 853 nps 4930 seldepth 4 time 173 pv b8c6 e4e5 c6e5
info depth 5
info score 192 nodes 3440 nps 12329 seldepth 5 time 279 pv b8c6 e4e5 c6e5 g1e2
info depth 6
info score 320 nodes 6946 nps 17235 seldepth 6 time 403 pv b8c6 e4e5 c6e5 a2a3 g8f6
info depth 7
info score 237 nodes 17511 nps 26371 seldepth 7 time 664 pv b8c6 e4e5 c6e5 a2a3 g8f6 g1f3
info depth 8
info score 395 nodes 22909 nps 27015 seldepth 8 time 848 pv b8a6 e4e5 a8b8 e5e6 d7e6 a2a3 d8d2
info depth 9
info score 395 nodes 43388 nps 31102 seldepth 9 time 1395 pv b8a6 e4e5 a8b8 e5e6 b8a8 e4e5 b7g2
info depth 10
info score 298 nodes 100646 nps 34163 seldepth 10 time 2946 pv b8a6 f1a6 a8b8 a6b5 g8f6 b5d7 f6d7 e4e5 d7e5
info depth 11
info score 261 nodes 201373 nps 32168 seldepth 11 time 6260 pv b8a6 e4e5 a8b8 e5e6 f7e6 a2a3 g8f6 a3a4 a6c5 b1c3
bestmove b8a6

See what I mean about playing strength...

Matthew:out

[UncombedCoconut]

Minor note: you want to say "score cp #" instead of "score #".

[Sven]

Another minor note: I am not fully convinced of your engine being ready for release Looking at the PVs tells me that your search has bugs. For instance in depth 4 it prints "b8c6 e4e5 c6e5". Assuming that the game started with 1.e4 prior to this search, the engine accepts to lose a pawn for nothing, so there must be something completely wrong with your search. Subsequent PVs are even worse, showing sacrifices like Qd8xd2 or Bb5xd7 in the early opening.

Sven

[BubbaTough]

Another minor note: I am not fully convinced of your engine being ready for release Looking at the PVs tells me that your search has bugs. For instance in depth 4 it prints "b8c6 e4e5 c6e5". Assuming that the game started with 1.e4 prior to this search, the engine accepts to lose a pawn for nothing, so there must be something completely wrong with your search. Subsequent PVs are even worse, showing sacrifices like Qd8xd2 or Bb5xd7 in the early opening.

Sven

I haven't looked to carefully, but it may work fine and just have no quiescent search.

-Sam

[Sven]

Another minor note: I am not fully convinced of your engine being ready for release Looking at the PVs tells me that your search has bugs. For instance in depth 4 it prints "b8c6 e4e5 c6e5". Assuming that the game started with 1.e4 prior to this search, the engine accepts to lose a pawn for nothing, so there must be something completely wrong with your search. Subsequent PVs are even worse, showing sacrifices like Qd8xd2 or Bb5xd7 in the early opening.

Sven

I haven't looked to carefully, but it may work fine and just have no quiescent search.

-Sam

Code: Select all

b8a6 e4e5 a8b8 e5e6 b8a8 e4e5 b7g2

has illegal moves if the analysis starts after moving 1.e4 from the initial chess board, which I assume based on previous posts of Matthew.

Code: Select all

b8a6 f1a6 a8b8 a6b5 g8f6 b5d7 f6d7 e4e5 d7e5

Here black does not recapture the Ba6, and White sacrifices the bishop on d7 and the pawn on e5. Does not make sense, even for an engine without qsearch.


Sven

[BubbaTough]

errrr...yes that does look bad, particularly the illegal move.

-Sam

[Sven]

errrr...yes that does look bad, particularly the illegal move.

It might be a PV display resp. PV calculation problem in this case (maybe a problem of not "null-terminating" the PV and therefore keeping old garbage from different subtrees).

But the wrong sacrifices remain. An engine without qsearch would favor lines ending with a capture even when that capture is losing. But Bx(knight)a6 - any - Bb5, or Bx(pawn)d7 followed by Nxd7, both in the middle of the PV, are results of a wrong search. And that should be fixed.

Sven