Author: Dann Corbit
Date: 12:53:38 08/19/99
Go up one level in this thread
On August 19, 1999 at 09:27:07, Vincent Lejeune wrote:
>On August 18, 1999 at 14:33:50, Dann Corbit wrote:
>
>>On August 18, 1999 at 13:01:52, Dann Corbit wrote:
>>
>>>On August 18, 1999 at 12:33:59, Vincent Lejeune wrote:
>>>
>>>>Position n°100 of the Aemis 1 test is reported as unsolved at Shep's Site.
>>>>
>>>>I've test it with Rebel10c (stronger setting), it find the keymove (Ng4!) after
>>>>21 sec on my Celeron@450 Mhz with 100 Mbyte hashtable.
>>>OK, dumb question, is Ng4 actually better than Qh4 which has a very good eval?
>>Here are some C.A.P. records. They favor Qh4. If that is wrong, I suspect one
>>of two things:
>>0. Zugzwang
>>1. Positional
>>
>>5bk1/1r4r1/1pRp1n2/p2Pp2p/P3Pp2/1QB2PqN/1P2BNp1/6K1 b - - acd 13; ce 144; pv Qh4
>>Bxe5 dxe5 d6+ Kh7 Qe6 Rbf7 Rc8 Nd7 Bb5 Nc5 Qxe5 Rf6 Rc6; pm Qh4; bm Ng4; id
>>"Aemis Test Suite No.100";
>>
>>5bk1/1r4r1/1pRp1n2/p2Pp2p/P3Pp2/1QB2PqN/1P2BNp1/6K1 b - - acd 15; acn 622115281;
>>acs 5001; ce 57; pv Qh4 Bxe5 dxe5 d6+ Kh7 Qe6 Rbf7 Qxe5 Nd7 Qd5 Kh8 Bd3 Nf6 Qd4
>>Qg3 Bc4;
>
>
>Rebel 10c analysis:
>
>After Qh4, 6 min, 50 Mnodes, best move Rc6xd6, score : -1.90
>After Ng4, 5 min, 35 Mnodes, best move f3xg4 , score : -3.49
>
>The theme of the position is : strong attack (queen+2 passed pawn+rook) on white
>king.
>
>It's strange that crafty don't see that on depth 15
>
>What's the crafty's evaluation when you play 1..Ng4 on board ?
First, the bare records:
5bk1/1r4r1/1pRp4/p2Pp2p/P3Ppn1/1QB2PqN/1P2BNp1/6K1 w - - acd 13; acn 675423462;
acs 3601; c0 "After Ng4"; ce -8; pv fxg4 hxg4 Bxg4 Rxg4 Nxg4 Qxh3 Nf2 Qe3 Qd1 f3
Bd2 Qd4 Qxf3 Qxd2 Qxg2+;
5bk1/1r4r1/1pRp1n2/p2Pp2p/P3Pp1q/1QB2P1N/1P2BNp1/6K1 w - - acd 15; acn
290699249; acs 3601; c0 "After Qh4"; ce -182; pv Bxe5 dxe5 d6+ Kh7 Qe6 Rbf7 Qf5+
Kh8 Qxe5 Nd7 Qd4 Qf6 Qd5;
Partial list of crafty's evaluation after Ng4 shows that disaster does not show
up till ply 13 (which would be 14 from previous position) and how bad it really
is does not get completed until over one hour! On ply 13, after 1:48, crafty
saw some kind of disaster, but did not "fully comprehend" how bad it was even
after one full hour. I am not sure why crafty 'hit the wall' so hard here. I
plan to send an email to the crafty list server to discuss this particular
position.
s depth time score variation (1)
8 1.80 0.39 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxh3
4. Nf2 Qf3 5. Rxb6
8-> 2.12 0.39 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxh3
4. Nf2 Qf3 5. Rxb6
9 2.51 -- 1. fxg4
9 41.34 -0.27 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxg4
4. Nf2 Qe2 5. Qc4 Qf3
9-> 42.33 -0.27 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxg4
4. Nf2 Qe2 5. Qc4 Qf3
10 46.75 0.06 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxg4
4. Nf2 Qe2 5. Qd1 Qxd1+ 6. Nxd1 f3
7. Nf2
10-> 48.11 0.06 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxg4
4. Nf2 Qe2 5. Qd1 Qxd1+ 6. Nxd1 f3
7. Nf2
11 55.19 -0.33 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxh3
4. Nf2 Qe3 5. Qc2 f3 6. Qd2 Qxd2 7.
Bxd2
11-> 59.62 -0.33 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxh3
4. Nf2 Qe3 5. Qc2 f3 6. Qd2 Qxd2 7.
Bxd2
12 1:19 -0.08 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxh3
4. Nf2 Qe3 5. Qd1 f3 6. Bd2 Qd4 7.
Qxf3 Qxd2 8. Qxg2+
12-> 1:26 -0.08 1. fxg4 hxg4 2. Bxg4 Rxg4 3. Nxg4 Qxh3
4. Nf2 Qe3 5. Qd1 f3 6. Bd2 Qd4 7.
Qxf3 Qxd2 8. Qxg2+
13 1:48 -- 1. fxg4
time=60:00 cpu=100% mat=-1 n=675423462 fh=6% nps=187612
ext-> checks=103468015 recaps=339783 pawns=1473048 1rep=13505712 thrt:20709
predicted=0 nodes=675423462 evals=1904181
endgame tablebase-> probes done=0 successful=0
Now, for the other position, you can ignore the time spent on ply 13 (I was
compiling several hundred megabytes of C source code at the time which is why
the NPS is so low for this record):
s depth time score variation (1)
7 1.19 -0.73 1. Rxb6 Ra7 2. Rc6 Rg3 3. Qb6 Rf7 4.
Qd8 Rxh3 5. Nxh3 Qxh3 6. Bxa5
7-> 1.29 -0.73 1. Rxb6 Ra7 2. Rc6 Rg3 3. Qb6 Rf7 4.
Qd8 Rxh3 5. Nxh3 Qxh3 6. Bxa5
8 2.26 -0.65 1. Rxb6 Ra7 2. Qc2 Rg3 3. Ra6 Rag7
4. b3 Rc7 5. Rxa5 Rxh3 6. Kxg2
8-> 2.51 -0.65 1. Rxb6 Ra7 2. Qc2 Rg3 3. Ra6 Rag7
4. b3 Rc7 5. Rxa5 Rxh3 6. Kxg2
9 4.74 -- 1. Rxb6
9 5.30 -1.38 1. Rxb6 Rxb6 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Kxg2 Nd7 5. Qa7 Rg3+ 6. Kf1 Nc5
9 8.08 -0.74 1. Ba6 Rbf7 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Nxh3 Qxh3 5. Qf2
9-> 8.27 -0.74 1. Ba6 Rbf7 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Nxh3 Qxh3 5. Qf2
10 13.70 -- 1. Ba6
10 23.20 -3.47 1. Ba6 Rg3 2. Be1 Rxh3 3. Kxg2 Qg3+
4. Kf1 Rg7 5. Ke2 Qg2 6. Bc8 Nd7 7.
Bxd7 Rxd7 8. Rxb6
10 24.70 -1.47 1. Rxb6 Rxb6 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Kxg2 Nd7 5. Qa7 Rh2+ 6. Kf1 Nc5
10-> 31.94 -1.47 1. Rxb6 Rxb6 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Kxg2 Nd7 5. Qa7 Rh2+ 6. Kf1 Nc5
11 34.56 -1.48 1. Rxb6 Rxb6 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Kxg2 Nd7 5. Qa7 Rh2+ 6. Kf1 Nc5
7. Be1
11-> 49.53 -1.48 1. Rxb6 Rxb6 2. Qxb6 Rg3 3. Bxa5 Rxh3
4. Kxg2 Nd7 5. Qa7 Rh2+ 6. Kf1 Nc5
7. Be1
12 1:09 -- 1. Rxb6
12 3:08 -6.53 1. Rxb6 Rg3 2. Bf1 gxf1=Q+ 3. Kxf1
Rxf3 4. Ke2 Rxh3 5. Nxh3 f3+ 6. Kd1
Qxh3 7. Bxa5 Rxb6 8. Qxb6 Nxe4
12 5:05 -5.77 1. Ba6 Rg3 2. Qc2 Rxh3 3. Nxh3 Qxh3
4. Qxg2+ Rg7 5. Qxg7+ Kxg7 6. Be2 Nd7
7. b4 Qg3+ 8. Kf1
12 33:19 -1.64 1. Bxe5 dxe5 2. d6+ Kh7 3. Qe6 Rbf7
4. Qxe5 Nd7 5. Qe8 Kh8 6. Rc7 Qf6 7.
Bb5
12-> 33:49 -1.64 1. Bxe5 dxe5 2. d6+ Kh7 3. Qe6 Rbf7
4. Qxe5 Nd7 5. Qe8 Kh8 6. Rc7 Qf6 7.
Bb5
13 35:20 -1.75 1. Bxe5 dxe5 2. d6+ Kh7 3. Qe6 Rbf7
4. Qf5+ Kh8 5. Qxe5 Nd7 6. Qe8 Kh7
7. Bc4 Rf6 8. Be6
13-> 37:19 -1.75 1. Bxe5 dxe5 2. d6+ Kh7 3. Qe6 Rbf7
4. Qf5+ Kh8 5. Qxe5 Nd7 6. Qe8 Kh7
7. Bc4 Rf6 8. Be6
14 45:52 -1.82 1. Bxe5 dxe5 2. d6+ Kh7 3. Qe6 Rbf7
4. Qf5+ Kh8 5. Qxe5 Nd7 6. Qd4 Qf6
7. Qd5 <HT>
14-> 51:01 -1.82 1. Bxe5 dxe5 2. d6+ Kh7 3. Qe6 Rbf7
4. Qf5+ Kh8 5. Qxe5 Nd7 6. Qd4 Qf6
7. Qd5 <HT>
time=60:00 cpu=99% mat=-1 n=290699249 fh=91% nps=80737
ext-> checks=17794262 recaps=987059 pawns=783124 1rep=2216259 thrt
predicted=0 nodes=290699249 evals=54694382
endgame tablebase-> probes done=0 successful=0
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.