first post

Drofa 2.0.2 Normal

Wanted to take another look at Drofa...which on my initial test, locked up the Pi. Not sure what happened then, but Drofa calculated a full game this time around...albeit a weak game. Drofa was badly beaten by tabletop Tasc R30...Drofa 14. Qf2?? was Game Over:

[Event "30 sec/mv"]
[Black "Tasc R30 2.2 Aktiv"]
[Date "Dec 23, 2020"]
[Result "0-1"]
[Round "Drofa loses badly"]
[Site "performance book"]
[White "Drofa 2.0.2 Normal"]

1.d4 c6 2.c4 d5 3.Nc3 Nf6 4.Nf3 dxc4 5.a4 Bf5 6.e3 e6 7.Bxc4 Bb4 8.O-O O-O 9.Nh4 Nbd7 10.Nxf5 exf5 11.Qc2 g6 12.f3 Nb6 13.Bb3 Re8 14.Qf2 Nbd5 15.Qg3 f4 16.Qg5 fxe3 17.Nxd5 Nxd5 18.Bxe3 Rxe3 19.Qg4 Rxb3 20.Ra3 Bxa3 21.bxa3 Ne3 22.Qg3 Nxf1 23.Kxf1 Rb1+ 24.Kf2 Qxd4+ 25.Ke2 Re8+ 26.Qe5 Rxe5# 0-1

Found some info on the web...
From the author: "Drofa started as fork of the Shallow Blue chess engine."
Well, that explains the weak play...Shallowblue is quite weak.

so...Gave Drofa a chance against Excalibur Ivan...which is about 1800 (I also gave Ivan a time handicap of 10s/mv vs Drofa 30s/mv). Drofa had a winning game by move 22, but blundered with inexplicable 22. Qxd5???? Game Over. That's bad...the author needs to fix that.

[Event "Handicap 30 vs 10 sec/mv"]
[Black "Ivan 10 sec/mv"]
[Date "Dec 23, 2020"]
[Result "0-1"]
[Round "Drofa has win...loses"]
[Site "performance book"]
[White "Drofa 2.0.2 Normal"]

1.d4 d5 2.c4 e6 3.Nf3 dxc4 4.e3 b5 5.a4 c6 6.Nc3 b4 7.Ne4 f5 8.Ng3 c3 9.bxc3 bxc3 10.Bc4 Bd6 11.Qb3 Ne7 12.Bxe6 Bxe6 13.Qxe6 Qd7 14.Qa2 Nd5 15.O-O Nb4 16.Qb3 Nd5 17.e4 fxe4 18.Nxe4 Bb4 19.Ba3 a5 20.Nc5 Bxc5 21.Rfe1+ Be7 22.Qxd5 Qxd5 23.Rxe7+ Kd8 24.Ne5 Kc8 25.Nf7 Rd8 26.Nxd8 Kxd8 27.Rxg7 Qxd4 28.Rxh7 c2 29.Rf1 Qxa4 30.Bb2 Qe4 31.Bf6+ Kc8 32.Re7 Qf4 33.Re8+ Kb7 34.Bb2 Qb4 35.Bc1 Nd7 36.Re2 Qc4 37.Ree1 Rg8 38.Re3 Qd5 39.Rg3 Rxg3 40.hxg3 a4 41.g4 Qd3 42.Bh6 Qd4 43.Bc1 Qxg4 44.f3 Qd4+ 45.Kh2 Qd6+ 46.Kg1 a3 47.Kh1 Qd1 48.Rg1 a2 49.Bb2 a1=B 50.Bxa1 c1=Q 51.Rxd1 Qxd1+ 52.Kh2 Qxa1 53.Kg3 Qd4 54.Kh2 Qf2 55.f4 Qh4+ 56.Kg1 Nf6 57.Kf1 Qxf4+ 58.Ke2 Kb6 59.g3 Qxg3 60.Kd2 Ne4+ 61.Kc1 Qc3+ 62.Kb1 Nd2+ 63.Ka2 Qc2+ 64.Ka3 Qb3# 0-1

Looks like Drofa is very weak, but apparently FF on the Pi (the Pi lockup on the initial test game is a mystery).

Your compilation seems buggy, or you were doing something wrong.
Neither 2.02dev, nor official 2.00 plays those moves in both games (even compiled 2.02dev today, just to check your strange games).

Just one example, (Qf2 is not considered, but it is no error BTW), the real incredible error Qg3??? never is searched at all

Code:

118128 >first : usermove 118128 >first : f8e8
118148 <first :   1      60      0         59 c2d3
118148 <first :   2      47      0        302 c2d3 b4d6
118148 <first :   3      43      0       6017 c1d2 c6c5 a4a5
118148 <first :   4      43      0       6785 c1d2 c6c5 a4a5 c5d4
118148 <first :   5      55      1      10413 c1d2 f5f4 a1e1 d8d6 c3e4
118158 <first :   6      51      1      13443 c1d2 a8c8 a4a5 b6d5 c2d3 b4c3
118158 <first :   7      51      2      16825 c1d2 a8c8 a4a5 b6d5 c2d3 b4c3 b2c3
118198 <first :   8      48      6      61749 c2d3 d8e7 e3e4 b4c3 b2c3 f5e4 f3e4 a8d8
118268 <first :   9      47     12     122400 c2d3 d8d6 c1d2 b6d5 c3d5 f6d5 b3d5 c6d5 a1c1
118378 <first :  10      49     24     236173 c2d3 b4c3 d3c3 b6d5 c3d3 c6c5 c1d2 c5d4 d3d4 d8d6
118918 <first :  11      54     78     776770 c2d3 b4a5 c1d2 b6d5 a1e1 d5c3 b2c3 a5c7 e3e4 f5f4
119108 <first :  12      48     97     970300 c2d3 b4a5 c1d2 b6d5 a1e1 d5c3 b2c3 a5c7 e3e4 f5f4 e1b1 c7d6
119478 <first :  13      49    134    1337892 c2d3 b4a5 c1d2 b6d5 a1b1 d5c3 b2c3 a5b6 d2e1 b6c7 e3e4 c6c5
121278 <first :  14      41    314    3119725 c2d3 b4a5 c1d2 b6d5 a1b1 d5c3 b2c3 a5b6 e3e4 c6c5 d2e3 f5f4 e3f4 c5d4
123038 <first :  15      41    490    4858966 c2d3 b4a5 c1d2 b6d5 a1b1 d5c3 b2c3 a5b6 e3e4 c6c5 d2e3 f5f4 e3f4 c5d4 c3d4
144168 <first :  16      50   2603   25299035 c2d3 b4c3 b2c3 b6d7 d3e2 d8c7 a1b1 a8d8 e3e4 f5e4 f3e4 e8e4 b3f7 g8h8
153074 <first :  17      48   3494   33589859 c2d3 b4c3 b2c3 b6d7 d3e2
218402 <first :  18      48  10026   88599259 c2d3 b4c3

https://lichess.org/analysis/standard/r2qr1k1/pp3p1p/2p2np1/3n1p2/Pb1P4/1BN1PP2/1P3QPP/R1B2RK1_w_-_-_5_15

Code:

806491 >first : usermove 806491 >first : b6d5
806531 <first :   1      34      0         92 c3d5
806531 <first :   2      34      0        257 c3d5 f6d5
806531 <first :   3      42      0        449 c3d5 f6d5 e3e4
806531 <first :   4      42      0        943 c3d5 f6d5 e3e4 f5e4
806531 <first :   5      42      0       1380 c3d5 f6d5 e3e4 f5e4 f3e4
806531 <first :   6      33      0       5303 c1d2 d5c3 b2c3 b4d6 e3e4 f5e4
806541 <first :   7      33      0       6113 c1d2 d5c3 b2c3 b4d6 e3e4 f5e4 f3e4
806541 <first :   8      33      0       7225 c1d2 d5c3 b2c3 b4d6 e3e4 f5e4 f3e4 f6e4
806551 <first :   9      30      4      40562 c1d2 d5c3 b2c3 b4d6 a4a5 d8c7 f2h4 f6d5 e3e4
806551 <first :  10      34      5      47556 c1d2 d5c3 b2c3 b4d6 a4a5 d8c7 f2h4 f6d5 e3e4 f5e4
806661 <first :  11      31     16     152599 c1d2 d8b6 b3c4 d5c3 b2c3 b4d6 e3e4 f5e4 f3e4 f6e4 f2f7
806791 <first :  12      32     28     275937 c1d2 d8b6 b3c4 d5c3 b2c3 b4f8 a4a5 b6c7 a5a6 b7b5 f2h4 f8d6
807091 <first :  13      36     59     572879 c1d2 d8b6 b3c4 b6a5 f1c1 d5c3 d2c3 b4c3 c1c3 f6d5 c3a3 a8d8 a1c1
807721 <first :  14      34    121    1183091 c1d2 d8b6 b3c4 b6a5 f1c1 d5c3 d2c3 b4c3 c1c3 f6d5 c4d5 a5d5 c3c5 d5e6
809501 <first :  15      36    300    2877945 c1d2 d8b6 b3c4 b6a5 f1c1 d5c3 d2c3 b4c3 c1c3 f6d5 c4d5 a5d5 c3c5 d5b3 c5c3
814351 <first :  16      36    785    7458976 c1d2 d8b6 b3c4 b6a5 f1c1 a8d8 a1b1 a7a6 c4a2 b7b5 a2d5 f6d5 c3d5 d8d5 d2b4 a5b4
823537 <first :  17      34   1703   15598262 c1d2 d8b6 b3c4 b4d6 a4a5 b6c7

FYI Drofa 2.00 already was over 800! rating points above ShallowBlue...

Piman wrote:

Drofa 2.0.2 Normal

Wanted to take another look at Drofa...which on my initial test, locked up the Pi. Not sure what happened then, but Drofa calculated a full game this time around...albeit a weak game. Drofa was badly beaten by tabletop Tasc R30...Drofa 14. Qf2?? was Game Over:

[Event "30 sec/mv"]
[Black "Tasc R30 2.2 Aktiv"]
[Date "Dec 23, 2020"]
[Result "0-1"]
[Round "Drofa loses badly"]
[Site "performance book"]
[White "Drofa 2.0.2 Normal"]

1.d4 c6 2.c4 d5 3.Nc3 Nf6 4.Nf3 dxc4 5.a4 Bf5 6.e3 e6 7.Bxc4 Bb4 8.O-O O-O 9.Nh4 Nbd7 10.Nxf5 exf5 11.Qc2 g6 12.f3 Nb6 13.Bb3 Re8 14.Qf2 Nbd5 15.Qg3 f4 16.Qg5 fxe3 17.Nxd5 Nxd5 18.Bxe3 Rxe3 19.Qg4 Rxb3 20.Ra3 Bxa3 21.bxa3 Ne3 22.Qg3 Nxf1 23.Kxf1 Rb1+ 24.Kf2 Qxd4+ 25.Ke2 Re8+ 26.Qe5 Rxe5# 0-1

Found some info on the web...
From the author: "Drofa started as fork of the Shallow Blue chess engine."
Well, that explains the weak play...Shallowblue is quite weak.

so...Gave Drofa a chance against Excalibur Ivan...which is about 1800 (I also gave Ivan a time handicap of 10s/mv vs Drofa 30s/mv). Drofa had a winning game by move 22, but blundered with inexplicable 22. Qxd5???? Game Over. That's bad...the author needs to fix that.

[Event "Handicap 30 vs 10 sec/mv"]
[Black "Ivan 10 sec/mv"]
[Date "Dec 23, 2020"]
[Result "0-1"]
[Round "Drofa has win...loses"]
[Site "performance book"]
[White "Drofa 2.0.2 Normal"]

1.d4 d5 2.c4 e6 3.Nf3 dxc4 4.e3 b5 5.a4 c6 6.Nc3 b4 7.Ne4 f5 8.Ng3 c3 9.bxc3 bxc3 10.Bc4 Bd6 11.Qb3 Ne7 12.Bxe6 Bxe6 13.Qxe6 Qd7 14.Qa2 Nd5 15.O-O Nb4 16.Qb3 Nd5 17.e4 fxe4 18.Nxe4 Bb4 19.Ba3 a5 20.Nc5 Bxc5 21.Rfe1+ Be7 22.Qxd5 Qxd5 23.Rxe7+ Kd8 24.Ne5 Kc8 25.Nf7 Rd8 26.Nxd8 Kxd8 27.Rxg7 Qxd4 28.Rxh7 c2 29.Rf1 Qxa4 30.Bb2 Qe4 31.Bf6+ Kc8 32.Re7 Qf4 33.Re8+ Kb7 34.Bb2 Qb4 35.Bc1 Nd7 36.Re2 Qc4 37.Ree1 Rg8 38.Re3 Qd5 39.Rg3 Rxg3 40.hxg3 a4 41.g4 Qd3 42.Bh6 Qd4 43.Bc1 Qxg4 44.f3 Qd4+ 45.Kh2 Qd6+ 46.Kg1 a3 47.Kh1 Qd1 48.Rg1 a2 49.Bb2 a1=B 50.Bxa1 c1=Q 51.Rxd1 Qxd1+ 52.Kh2 Qxa1 53.Kg3 Qd4 54.Kh2 Qf2 55.f4 Qh4+ 56.Kg1 Nf6 57.Kf1 Qxf4+ 58.Ke2 Kb6 59.g3 Qxg3 60.Kd2 Ne4+ 61.Kc1 Qc3+ 62.Kb1 Nd2+ 63.Ka2 Qc2+ 64.Ka3 Qb3# 0-1

Looks like Drofa is very weak, but apparently FF on the Pi (the Pi lockup on the initial test game is a mystery).

temporarily wrote:

Your compilation seems buggy, or you were doing something wrong.

Thank you for the feedback.
Just an FYI: I DON'T compile...I just test the Pi3B compiled engines as they are presented to me.
You need to have a PM with Scally to figure out what's going on.
I test.
And clearly my tests have found that something in the compile is not correct....and that initial Pi freeze game (which now I am convinced was not a fluke) is another clue as to what hasn't ported correctly to the Pi.

I'd like to see your engine run, correctly...as it's supposed to ...on the Pi.

This is fixable and issues like this have been seen previously on other engines....and they were fixed.
800 points is an excellent improvement. Let's get Drofa working on the Pi. Your input to Scally will help him.

I just checked your comment:

Just one example, (Qf2 is not considered, but it is no error BTW), the real incredible error Qg3??? never is searched at all

You are correct, my bad: Qf2 is not the error move....Qg3?? is the loser. Apparently, I was having a small TIA at the time of that typing.