Honey-XPro Release

[Graham Banks]

Mike - the link appears not to be working now?

[ChiefPushesWood]

Mike - the link appears not to be working now?

Here ya go, Graham.

CPW

[Graham Banks]

Mike - the link appears not to be working now?

Here ya go, Graham.

CPW

Thanks.

[Ovyron]

An undocumented feature that you can also combine tactical and multipv - say you want a wider search tactical to 3 ( search as if multipv was set to 8) and you set multipv to see just the best 4 best moves , although it is doing wider searches if was multipv was searched to 8, it will just display the 4 best moves.

Yes, I noticed it was missing when trying to use it with just MultiPV. But isn't this a problem with the MultiPV protocol itself? One should be free to go to the opening position, switch to MultiPV=4, use searchmoves to exclude e4, and then be able to see the top 4 moves of the engine besides e4 (faster, because e4 wasn't earched at all). Instead, when using searchmoves the engine switches back to MultiPV=1, but this is unwanted behavior by UCI.

Anyway, I'm ceasing to use Honey immediately.

Because one thing that McCain has, is what I call "going berserk". What happens is that McCain will love a move so much, that it'll inflate its score out of this world, showing +2.40 scores, very exciting. The point is not the score itself, the point is the distance from second best, which would be around +0.90.

The 2.40 score isn't real, of course, it's more like 1.20, but I've been playing those moves and beating people at corr chess, so going berserk has been critical and the main reason for using McCain over all other engines derivatives.

Honey doesn't do that, it'd show 1.30 for one move, and 1.20 for the other, and then I have to analyze to check if the 1.20 one can be bumped up or the 1.30 can be bumped down, as I'm unsure if the second move is best. With McCain there's no doubt, at least, I haven't seen it go berserk for a move that wasn't best, which is impressive. (so no point trying to refute the 2.40 to below 0.90 or improve the 0.90 to over 2.40 -- just trust McCain and play the move)

So thanks for all your job on Honey, but I'll stick to McCain and will be looking forward to next release.

[Joerg Oster]

An undocumented feature that you can also combine tactical and multipv - say you want a wider search tactical to 3 ( search as if multipv was set to 8) and you set multipv to see just the best 4 best moves , although it is doing wider searches if was multipv was searched to 8, it will just display the 4 best moves.

Yes, I noticed it was missing when trying to use it with just MultiPV. But isn't this a problem with the MultiPV protocol itself? One should be free to go to the opening position, switch to MultiPV=4, use searchmoves to exclude e4, and then be able to see the top 4 moves of the engine besides e4 (faster, because e4 wasn't earched at all). Instead, when using searchmoves the engine switches back to MultiPV=1, but this is unwanted behavior by UCI.

Working as expected here.

Code: Select all

uciok
setoption name Hash value 512
setoption name MultiPV value 3
ucinewgame
isready
readyok
go depth 18 searchmoves a2a3 a2a4 b2b3 b2b4 c2c3 c2c4 d2d3 d2d4 e2e3 f2f3 f2f4 g2g3 g2g4 h2h3 h2h4 b1a3 b1c3 g1f3 g1h3
info depth 18 seldepth 27 multipv 1 score cp 68 nodes 1835086 nps 1162182 hashfull 27 tbhits 0 time 1579 pv c2c4 g8f6 d2d4 e7e6 g1f3 c7c5 b1c3 c5d4 f3d4 f8b4 d4c2 b4c3 b2c3 e8g8 c1a3 f8e8 a3d6 b8c6 e2e3 g8h8 f1d3
info depth 18 seldepth 23 multipv 2 score cp 63 nodes 1835086 nps 1162182 hashfull 27 tbhits 0 time 1579 pv g1f3 d7d5 d2d4 e7e6 c2c4 g8f6 b1c3 f8b4 e2e3 e8g8 f1d3 c7c5 e1g1 d5c4 d3c4 c5d4 e3d4 b8c6 a2a3 b4d6
info depth 18 seldepth 25 multipv 3 score cp 57 nodes 1835086 nps 1162182 hashfull 27 tbhits 0 time 1579 pv d2d4 d7d5 g1f3 e7e6 c2c4 c7c5 e2e3 g8f6 c4d5 e6d5 b1c3 b8c6 f1e2 f8e7 e1g1 e8g8 d4c5 e7c5 d1b3 d5d4
bestmove c2c4 ponder g8f6

[MikeB]

An undocumented feature that you can also combine tactical and multipv - say you want a wider search tactical to 3 ( search as if multipv was set to 8) and you set multipv to see just the best 4 best moves , although it is doing wider searches if was multipv was searched to 8, it will just display the 4 best moves.

Yes, I noticed it was missing when trying to use it with just MultiPV. But isn't this a problem with the MultiPV protocol itself? One should be free to go to the opening position, switch to MultiPV=4, use searchmoves to exclude e4, and then be able to see the top 4 moves of the engine besides e4 (faster, because e4 wasn't earched at all). Instead, when using searchmoves the engine switches back to MultiPV=1, but this is unwanted behavior by UCI.

Anyway, I'm ceasing to use Honey immediately.

Because one thing that McCain has, is what I call "going berserk". What happens is that McCain will love a move so much, that it'll inflate its score out of this world, showing +2.40 scores, very exciting. The point is not the score itself, the point is the distance from second best, which would be around +0.90.

The 2.40 score isn't real, of course, it's more like 1.20, but I've been playing those moves and beating people at corr chess, so going berserk has been critical and the main reason for using McCain over all other engines derivatives.

Honey doesn't do that, it'd show 1.30 for one move, and 1.20 for the other, and then I have to analyze to check if the 1.20 one can be bumped up or the 1.30 can be bumped down, as I'm unsure if the second move is best. With McCain there's no doubt, at least, I haven't seen it go berserk for a move that wasn't best, which is impressive. (so no point trying to refute the 2.40 to below 0.90 or improve the 0.90 to over 2.40 -- just trust McCain and play the move)

So thanks for all your job on Honey, but I'll stick to McCain and will be looking forward to next release.

Can you a post a position or PM me with a position where you see this "going berserk" behavior in McCain - it was certainly unintended, but I would like to investigate further.

[MikeB]

I have it running fine on F15 , but curious about Book #1 . There are boxes to enter pathway for Books #2 and #3 , but not for #1 ?

Thanks for all your efforts .....

Tom G

Hi Tom, This will be available in the next release of McCain ( and the next release of Honey as well, but that would be later.). In addition, I'm adding a 4th book capability.

[MikeB]

Seemed to work when I tried it in Arena:

Hey Dann.
Yeah it's working for me in other GUI's as well. Just not Chessbase which is my main software. Curious is all.

Chief

The bottom line - Chessbase is definitely quirky and less forgiving than xBoard and Arena. To make it harder, I found no debug mode Chessbase * at least one I could find quickly, , but working with Arena in debug mode, I was able to pinpoint the drivers behind some of the quirkiness. To my surprise, may of the "features" were getting lost in Chessbase ( as well as in Arena, but I was able to fix most of them - while running Chessbase on Parallels Desktop VM on macOS, installing MinGW, git etc , although my windows exe's in a VM environment are so slooooow. Here are some snapshots from the forthcoming McBrain X4. I could not get the explorer window to work Chessbase to select the 4 opening books - so one must type or paste the path in Chessbase, but the icon is there for Arena and XBoard ( and probably Winboard too , although not tested). I also made the Play by Engine Elo or play by engine level a little simpler and it now works for the first time ever in the Chessbase Fritz as well as Arena. Soon -to be released, not sure when yet.

[MikeB]

Seemed to work when I tried it in Arena:

Hey Dann.
Yeah it's working for me in other GUI's as well. Just not Chessbase which is my main software. Curious is all.

Chief

The bottom line - Chessbase is definitely quirky and less forgiving than xBoard and Arena. To make it harder, I found no debug mode Chessbase * at least one I could find quickly, , but working with Arena in debug mode, I was able to pinpoint the drivers behind some of the quirkiness. To my surprise, may of the "features" were getting lost in Chessbase ( as well as in Arena, but I was able to fix most of them - while running Chessbase on Parallels Desktop VM on macOS, installing MinGW, git etc , although my windows exe's in a VM environment are so slooooow. Here are some snapshots from the forthcoming McBrain X4. I could not get the explorer window to work Chessbase to select the 4 opening books - so one must type or paste the path in Chessbase, but the icon is there for Arena and XBoard ( and probably Winboard too , although not tested). I also made the Play by Engine Elo or play by engine level a little simpler and it now works for the first time ever in the Chessbase Fritz as well as Arena. Soon -to be released, not sure when yet.

[MikeB]

Seemed to work when I tried it in Arena:

Hey Dann.
Yeah it's working for me in other GUI's as well. Just not Chessbase which is my main software. Curious is all.

Chief

The bottom line - Chessbase is definitely quirky and less forgiving than xBoard and Arena. To make it harder, I found no debug mode Chessbase * at least one I could find quickly, , but working with Arena in debug mode, I was able to pinpoint the drivers behind some of the quirkiness. To my surprise, may of the "features" were getting lost in Chessbase ( as well as in Arena, but I was able to fix most of them - while running Chessbase on Parallels Desktop VM on macOS, installing MinGW, git etc , although my windows exe's in a VM environment are so slooooow. Here are some snapshots from the forthcoming McBrain X4. I could not get the explorer window to work Chessbase to select the 4 opening books - so one must type or paste the path in Chessbase, but the icon is there for Arena and XBoard ( and probably Winboard too , although not tested). I also made the Play by Engine Elo or play by engine level a little simpler and it now works for the first time ever in the Chessbase Fritz as well as Arena. Soon -to be released, not sure when yet.