Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: avoid move

Author: A. Cozzie

Date: 19:18:21 08/11/05

Go up one level in this thread


On August 11, 2005 at 22:00:27, A. Cozzie wrote:

>On August 11, 2005 at 15:15:02, A. Cozzie wrote:
>
>>On August 11, 2005 at 15:08:39, John Merlino wrote:
>>
>>>On August 11, 2005 at 11:52:20, Will Singleton wrote:
>>>
>>>>[d]2rq3k/Qp1r2pp/1P6/2p1pP2/1b6/3PBB1b/4PP1P/R1R3K1 b - - am Bxf5
>>>
>>>CM9_R1 on an AMD 2500 does not see anything better than Bxf5 in under five
>>>minutes:
>>>
>>>Time	Depth	Score	Positions	Moves
>>>0:00	1/3	0.65	1779		1...Qe7 2.Be4 Rf8
>>>0:00	1/4	0.14	4229		1...Qe7 2.Be4 Rf8 3.Qa2 Bxf5 4.Bxf5
>>>					Rxf5
>>>0:00	1/4	-0.15	8696		1...Bxf5 2.Bg2 Qe7 3.Rc4 Be6
>>>0:00	1/5	-0.15	15259		1...Bxf5 2.Bg2 Qe7 3.Rc4 Be6
>>>0:00	1/6	-0.08	40337		1...Bxf5 2.Bxb7 Rb8 3.Bg2 Rxa7
>>>					4.bxa7 Ra8 5.Bxa8 Qxa8 6.Bxc5
>>>0:00	1/7	-0.08	172253		1...Bxf5 2.Bxb7 Rb8 3.Bg2 Rxa7
>>>					4.bxa7 Ra8 5.Bxa8 Qxa8 6.Bxc5
>>>0:01	1/8	-0.06	288147		1...Bxf5 2.Bxb7 Rb8 3.Bc6 Rxa7
>>>					4.bxa7 Ra8 5.f3 Be6 6.Bxa8 Qxa8
>>>					7.Bxc5
>>>0:03	1/9	0.03	790884		1...Bxf5 2.Bxb7 Rb8 3.Bc6 Rxa7
>>>					4.bxa7 Ra8 5.Bxa8 Qxa8 6.Bxc5 Bd2
>>>					7.Rc4
>>>0:07	1/10	0.00	1792747		1...Bxf5 2.Bxb7 Rb8 3.Bg2 Rxa7
>>>					4.bxa7 Ra8 5.Bxc5 Bd2 6.Rc2 Bf4
>>>					7.Bxa8 Qxa8
>>>0:16	1/11	0.20	4334263		1...Bxf5 2.Bxb7 Rb8 3.Bg2 Rxa7
>>>					4.bxa7 Ra8 5.Bxa8 Qxa8 6.Bxc5 Bd2
>>>					7.Rc2 Bf4 8.e4
>>>0:42	1/12	0.20	11644968	1...Bxf5 2.Bxb7 Rb8 3.Bg2 Rxa7
>>>					4.bxa7 Ra8 5.Bxa8 Qxa8 6.Bxc5 Bd2
>>>					7.Rc2 Bf4 8.e4
>>>2:08	1/13	0.54	35947673	1...Bxf5 2.Bxb7 Rb8 3.Bg2 Rxb6
>>>					4.Qa8 Rb8 5.Qxb8 Qxb8 6.Ra8 Qxa8
>>>					7.Bxa8 Ba3 8.Rc2 Rc7 9.Bd5 Bb4
>>>
>>>jm
>>
>>This is a tough position.  Zappa (4 CPUs) needs 17 ply / 3 minutes (or 14 ply /
>>1 minute with SE) to see -2.5 for Black after Bxf5.
>>
>>However the alternative (Qe7) is also lost, so I guess the correct goal here
>>should be 'get +1 for white'.
>>
>>anthony
>
>Since Bob posted some logs I guess I'll post mine as well :)
>
>1P:
>
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Bc6xa8 Qd8xa8
>6. Be3xc5 Bf5-h3 7. f2-f3 Bb4-d2 8. Rc1-b1
> = (-0.26)	Depth: 14/32	00:00:34.45	29108kN (845 KN/s, 0 splits, 0 aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Bc6xa8 Qd8xa8
>6. Be3xc5 Bf5-h3 7. f2-f3 Bb4-d2 8. Rc1-b1
> = (-0.26)	Depth: 14/32	00:00:48.91	41047kN (839 KN/s, 0 splits, 0 aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
>6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8 Qg5xc5 10. Rb1-b8
>Bh3-c8
> = (-0.78)	Depth: 15/33	00:01:07.96	57612kN (848 KN/s, 0 splits, 0 aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
>6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8 Qg5xc5
> = (-0.78)	Depth: 15/34	00:04:03.25	197999kN (814 KN/s, 0 splits, 0 aborts)
>
>4P:
>
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Bc6xa8 Qd8xa8
>6. Be3xc5 Bf5-h3 7. f2-f3 Bb4-d2 8. Rc1-b1
> = (-0.26)	Depth: 14/32	00:00:15.62	47502kN (3040 KN/s, 3056 splits, 186 aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
>6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8 Qg5xc5 10. Rb1-b8
>Bh3-c8
> = (-0.78)	Depth: 15/36	00:00:21.64	67188kN (3104 KN/s, 3445 splits, 236 aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
>6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8
> = (-0.78)	Depth: 15/36	00:01:24.71	261796kN (3090 KN/s, 6479 splits, 529
>aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
>6. Bc6xa8 Qd8xa8 7. f2-f3 Bb4-d2 8. Rc1-b1 Bh3-c8 9. Ra1-a6
> = (-0.62)	Depth: 16/36	00:01:33.42	290983kN (3115 KN/s, 6909 splits, 554
>aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
>6. Bc6xa8 Qd8xa8 7. f2-f3 Bb4-d2 8. Rc1-b1 Bh3-c8 9. Ra1-a6
> = (-0.62)	Depth: 16/36	00:02:13.38	410600kN (3078 KN/s, 8294 splits, 639
>aborts)
>1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Bc6xa8 Qd8xa8
>6. Be3xc5 Bb4xc5 7. Rc1xc5 Bf5-h3 8. f2-f3 Kh8-g8 9. Ra1-a5 Kg8-f7
> = (-2.18)	Depth: 17/37	00:02:49.30	525774kN (3105 KN/s, 8852 splits, 694
>aborts)
>1 ...Qd8-e7!!
> = (-2.17)	Depth: 17/37	00:03:14.29	607841kN (3128 KN/s, 9206 splits, 748
>aborts)
>1... Qd8-e7 2. Qa7-a4 Rc8-d8 3. Be3xc5 Bb4xc5 4. Rc1xc5 Rd7-d4 5. Rc5-c4 Bh3xf5
>6. Rc4xd4 e5xd4 7. Qa4-a5 Bf5-h3 8. Ra1-c1 Qe7-h4 9. Rc1-c7 Qh4-f4
> = (-1.01)	Depth: 17/38	00:07:37.85	1432121kN (3128 KN/s, 10235 splits, 872
>aborts)
>1... Qd8-e7 2. Qa7-a4 Rc8-d8 3. Be3xc5 Bb4xc5 4. Rc1xc5 Rd7-d4 5. Rc5-c4 Bh3xf5
>6. Rc4xd4 e5xd4 7. Qa4-a5 Bf5-h3 8. Ra1-c1 Qe7-h4 9. Rc1-c7
> = (-1.01)	Depth: 17/39	00:14:25.88	2688520kN (3105 KN/s, 14093 splits, 1179
>aborts)
>
>
>with SE:
>
>1... Bh3xf5 2. Bf3xb7 Rc8-b8[S] 3. Bb7-c6 Rd7xa7[S] 4. b6xa7 Rb8-a8 5. Be3xc5
>Bf5-h3 6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8 Qg5xc5 10.
>Rb1-b8
> = (-0.78)	Depth: 13/34	00:00:26.98	81118kN (3006 KN/s, 6469 splits, 683 aborts)
>1... Bh3xf5 2. Bf3xb7[S] Rc8-b8[S] 3. Bb7-c6 Rd7xa7[S] 4. b6xa7 Rb8-a8 5. Bc6xa8
>Qd8xa8 6. Be3xc5 Bb4xc5 7. Rc1xc5 Bf5-h3 8. f2-f3 Kh8-g8 9. Ra1-a5 Kg8-f7
> = (-2.18)	Depth: 14/39	00:01:00.20	186791kN (3102 KN/s, 9725 splits, 1039
>aborts)
>1 ...Qd8-e7!!
> = (-2.17)	Depth: 14/39	00:01:06.57	207894kN (3123 KN/s, 10153 splits, 1067
>aborts)
>1... Qd8-e7 2. Qa7-a4 Rc8-d8 3. Qa4-c2 Qe7-e8 4. Bf3-e4 Qe8-h5[S] 5. Kg1-h1
>Bh3xf5 6. Be4xf5 Qh5xf5[S] 7. Be3xc5 Bb4xc5 8. Qc2xc5 Rd8-f8 9. f2-f3
> = (-1.06)	Depth: 14/39	00:01:50.89	346824kN (3127 KN/s, 11339 splits, 1198
>aborts)
>1... Qd8-e7 2. Qa7-a4 Rc8-d8 3. Qa4-c2 Qe7-e8 4. Bf3-e4 Qe8-h5[S] 5. Kg1-h1
>Bh3xf5 6. Be4xf5 Qh5xf5[S] 7. Be3xc5 Bb4xc5 8. Qc2xc5 Rd8-f8 9. f2-f3
> = (-1.06)	Depth: 14/39	00:02:42.32	500853kN (3086 KN/s, 18166 splits, 2098
>aborts)
>
>So SE pretty much kick ass here.  The parallel speedup is only 2.9 / 4 (scaling
>3.8 / 4) though :(
>
>anthony

I wondered at why my speedup was so bad, and somehow I had lost a piece of code
telling it when to split :)  So when I put it back in, I get:

1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8 Qg5xc5 10. Rb1-b8
Bh3-c8
 = (-0.78)	Depth: 15/34	00:00:17.54	56682kN (3231 KN/s, 3344 splits, 220 aborts)
1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
6. Rc1-b1 Qd8-g5 7. Kg1-h1 e5-e4 8. Bc6xe4 Bb4-c3 9. Be4xa8 Qg5xc5
 = (-0.78)	Depth: 15/38	00:01:08.36	216183kN (3162 KN/s, 6174 splits, 480
aborts)
1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
6. Bc6xa8 Qd8xa8 7. f2-f3 Bb4-d2 8. Rc1-b1 Bh3-c8 9. Ra1-a6
 = (-0.62)	Depth: 16/38	00:01:17.66	247085kN (3181 KN/s, 6706 splits, 524
aborts)
1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Be3xc5 Bf5-h3
6. Bc6xa8 Qd8xa8 7. f2-f3 Bb4-d2 8. Rc1-b1 Bh3-c8 9. Ra1-a6
 = (-0.62)	Depth: 16/38	00:01:53.33	357477kN (3154 KN/s, 8127 splits, 609
aborts)
1... Bh3xf5 2. Bf3xb7 Rc8-b8 3. Bb7-c6 Rd7xa7 4. b6xa7 Rb8-a8 5. Bc6xa8 Qd8xa8
6. Be3xc5 Bb4xc5 7. Rc1xc5 Bf5-h3 8. f2-f3 Kh8-g8 9. Ra1-a5 Kg8-f7
 = (-2.18)	Depth: 17/38	00:02:28.55	472224kN (3179 KN/s, 8806 splits, 660
aborts)
1 ...Qd8-e7!!
 = (-2.17)	Depth: 17/38	00:02:53.08	552953kN (3195 KN/s, 9215 splits, 715
aborts)
1... Qd8-e7 2. Qa7-a4 Rc8-d8 3. Be3xc5 Bb4xc5 4. Rc1xc5 Rd7-d4 5. Rc5-c4 Bh3xf5
6. Rc4xd4 e5xd4 7. Qa4-a5 Bf5-h3 8. Ra1-c1 Qe7-f7 9. Rc1-c7 Qf7-f4
 = (-1.01)	Depth: 17/38	00:06:43.47	1281458kN (3176 KN/s, 10446 splits, 825
aborts)

So now the speedup is 3.56, and the scaling is still 3.8/4.0 or something.

Actually, since this is a fail-low position, its pretty tough to get a good
speedup :(

anthony



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.