Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is Lokasoft's claim correct?

Author: Sune Fischer

Date: 02:33:36 02/25/04

Go up one level in this thread


On February 24, 2004 at 17:38:49, Robert Hyatt wrote:

>On February 24, 2004 at 16:31:38, Dan Andersson wrote:
>
>> That's not what they wrote though. The quote should be:
>> 'Deep Sjeng is the only program we know of that can find this move instantly.'
>>And it isn't untrue AFAIK. But here is the output from one engine that does find
>>it in 42 seconds.
>>Engine  UCI 76 MB 933MHz CuMine:
>> 5   00:00  0,32   c5xd4 Nf3xd4 Rc8xc3 Bd3c4 Qd8b8 Nd2f3
>> 6   00:01  0,41   c5xd4 Nf3xd4 Rc8xc3 Bd3c4 Qd8b8 Rh2h1 f5f4
>> 7   00:04  0,32   c5xd4 c3xd4 Nd5c3+ b2xc3 Ne7d5 Rd1c1 Nd5xc3+ Rc1xc3 Rc8xc3
>>g4xf5 Rc3xa3 f5xe6 d7xe6
>> 8   00:08  0,32   c5xd4 c3xd4 Nd5c3+ b2xc3 Ne7d5 Rd1c1 Nd5xc3+ Rc1xc3 Rc8xc3
>>g4xf5 Rc3xa3 f5xe6 d7xe6
>> 9-  00:25  0,03   c5xd4 c3xd4 Nd5c3+ b2xc3 Ne7d5 Rd1c1 Nd5xc3+ Rc1xc3 Rc8xc3
>>g4xf5 Rc3xa3 f5xe6 d7xe6
>> 9   00:27  0,09   c5xd4 Nf3xd4 Nd5xc3+ b2xc3 Qd8c7 f2f4 Ne7d5 Nd4b5 Nd5xc3+
>>Nb5xc3 Qc7xc3 Nd2c4 b6b5
>> 9+  00:42  0,09   Nd5xc3+ b2xc3
>> 9   00:43  0,35   Nd5xc3+ b2xc3 Ne7d5 Rd1c1 c5xd4 c3xd4 Nd5c3+ Rc1xc3 Rc8xc3
>>g4xf5 e6xf5 Bd3c4+ d7d5 Bc4a6 Bb7xa6 Qe2xa6
>>10   00:47  0,57   Nd5xc3+ b2xc3 Ne7d5 Rd1c1 c5xd4 c3xd4 Nd5c3+ Rc1xc3 Rc8xc3
>>g4xf5 Qd8c7 Rh2h1 Rc3xa3 Rh1g1
>>11   01:02  0,60   Nd5xc3+ b2xc3 Ne7d5 Rd1c1 c5xd4 c3xd4 Nd5c3+ Rc1xc3 Rc8xc3
>>Kb1b2 Qd8c7 g4xf5 e6xf5 Bd3c4+ Rc3xc4 Qe2xc4+ Qc7xc4 Nd2xc4 Bb7xf3 Nc4d6
>>12   01:30  0,76   Nd5xc3+ b2xc3 Ne7d5 Rd1c1 c5xd4 c3xd4 Nd5c3+ Rc1xc3 Rc8xc3
>>Kb1b2 Qd8c7 Nf3e5 b6b5 g4xf5 d7d6 Ne5d7
>>13   02:39  0,54   Nd5xc3+ b2xc3 Ne7d5 Rd1c1 c5xd4 Nf3xd4 Nd5xc3+ Rc1xc3 Rc8xc3
>>g4xf5 Rc3xa3 f5xe6 Qd8e7 Nd2b3 d7xe6 Qe2h5 h7h6
>>
>>MvH Dan Andersson
>
>Here is one that finds it a _lot_ quicker.  :)
>
>                7     0.37   0.05   1. ... cxd4 2. Nxd4 Qc7 3. f4 Bc6 4.
>                                    Nxc6 Qxc6 5. gxf5 Nxf5
>                7     0.45  -0.03   1. ... Nxc3+ 2. bxc3 Nd5 3. Rc1 cxd4
>                                    4. cxd4 Nc3+ 5. Rxc3 Rxc3 6. gxf5 exf5
>                7->   0.51  -0.03   1. ... Nxc3+ 2. bxc3 Nd5 3. Rc1 cxd4
>                                    4. cxd4 Nc3+ 5. Rxc3 Rxc3 6. gxf5 exf5
>                8     0.59  -0.10   1. ... Nxc3+ 2. bxc3 Nd5 3. Rc1 cxd4
>                                    4. cxd4 Nc3+ 5. Rxc3 Rxc3 6. gxf5 exf5
>                                    7. Ka2
>                8->   0.98  -0.10   1. ... Nxc3+ 2. bxc3 Nd5 3. Rc1 cxd4
>                                    4. cxd4 Nc3+ 5. Rxc3 Rxc3 6. gxf5 exf5
>                                    7. Ka2
>                9     1.15  -0.25   1. ... Nxc3+ 2. bxc3 Nd5 3. Rc1 cxd4
>                                    4. cxd4 Nc3+ 5. Rxc3 Rxc3 6. gxf5 exf5
>                                    7. Bc4+ d5 8. Bb3
>                9->   2.02  -0.25   1. ... Nxc3+ 2. bxc3 Nd5 3. Rc1 cxd4
>                                    4. cxd4 Nc3+ 5. Rxc3 Rxc3 6. gxf5 exf5
>                                    7. Bc4+ d5 8. Bb3
>
>
>That is using one cpu on my dual xeon (other processor is running some tests
>right now).  .45 seconds would drop to somewhere between .2 and .3 using both
>cpus with SMT on.  Quad opteron would drop that to under .1 seconds. :)


I believe that this is not really a tactical shoot, this is an evaluation based
sacrifice. Some engines will find this very fast, others won't find it at all.

Mine still likes white better after Nxc3 bxc3 and a few minutes of search.
Of course it has no king safety at the moment, but aside from that it looks like
white is not completely dead :)

9	71	97	     634373	2...Nd5 4.Kb2 cxd4 5.cxd4 Nc3 6.Qf1 Nxd1+ 7.Qxd1 Qf6 8.Qb1
10	64	207	    1450310	2...Nd5 4.Rc1 cxd4 5.cxd4 Nc3+ 6.Rxc3 Rxc3 7.Kb2 Qc7
8.gxf5 exf5
11	59	447	    3347308	2...Nd5 4.Rc1 cxd4 5.cxd4 Nc3+ 6.Rxc3 Rxc3 7.Kb2 Qc7
8.gxf5 exf5 9.Rg2 d5
12	78	913	    7020065	2...Nd5 4.Rc1 cxd4 5.cxd4 Nc3+ 6.Rxc3 Rxc3 7.Kb2 Qc7 8.e4
fxe4 9.Nxe4 Bxe4 10.Qxe4
13	76	7849	   60175453	2...Nd5 4.Rc1 cxd4 5.cxd4 Nc3+ 6.Rxc3 Rxc3 7.Kb2 Qc7
8.Ne5 d5


-S.



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.