Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fairly good examples of eval inconsistencies between Rybka versions...

Author: Vasik Rajlich

Date: 03:16:29 01/03/06

Go up one level in this thread


On January 02, 2006 at 19:09:12, enrico carrisco wrote:

>On January 02, 2006 at 04:26:55, Vasik Rajlich wrote:
>
>>On January 01, 2006 at 17:06:43, enrico carrisco wrote:
>>
>>>I stated here in CCC days after Rybka was released that the nps was manipulated
>>>but lacked any real convincing examples of this.  I even ran the idea by Chrilly
>>>when he was disassembing Rybka, but he had no comment on it.
>>>
>>>In the following position, we see two different evaluations for Rybka 1.0 Beta
>>>64-bit and Rybka 1.01 Preview 2 64-bit.  Of course, without knowing what
>>>undocumented changes (if any) were made from 1.0 Beta to Preview 2, it's hard to
>>>deduce anything concrete.  However, Vasik has posted here a few times that
>>>nothing was changed with the strength or eval (I believe...  Not trying to quote
>>>anything...)
>>>
>>>Specifically, the time-to-ply figures are odd.  Preview 2 appears to slow down
>>>in time to ply while Rybka 1.0 Beta seems to 'speed up' as the depth increases
>>>(from nps figures.)
>>>
>>>The second position, below, shows the differences in nps better -- where the
>>>qsearch demands a little recognition...  :)  Time-to-ply is very close (unlike
>>>the first position) but the nps differs substantially...
>>>
>>
>>Enrico,
>>
>>I already explained this here - did you really miss it?
>
>Hello Vas.  Yes, I am afraid I did miss it.  Do you have a link to the message?
>I notice from Uri's reply to this thread it seems you posted something about
>changing Rybka's node count..  Is this correct?  I would be interested to read
>your posting on this.
>
>>
>>If you don't mind, I have two nosy questions for you:
>>
>>1) What is the Hiarcs definition of node?
>
>I'm not clear on what you're asking here with your inclusion of HIARCS.  Our
>definition does not differ from the general definition of a node (any move made
>by the program in its "thinking" - i.e. updating a data structure.)
>
>Perhaps I am missing some sarcasm?
>
>>2) Do you have access to the Hiarcs sources?
>
>Again -- I'm not sure how this relates in the matter of my original question.
>It is well known that Mark is the programmer of HIARCS.  If my activities
>included programming, I would be listed as a co-author.
>
>If the inference is that without a programming contribution to HIARCS my
>knowledge or input on such a matter is discounted -- please just say so.
>
>However, I believe there are many knowledgeable programmers in this forum that
>do not have a top 10 (or even 20) engine.
>
>If I've struck a nerve with you, it was not my intention.  I just wanted to know
>why the positions I posted gave varied eval by Rybka versions.  I would have
>assumed other posts in this thread which make very direct inferences to Rybka's
>roots would have been more worthy of any sarcastic retort than my inquiry.
>
>Regards,
>
>-elc.
>

Hi Enrico,

actually, I wasn't being sarcastic. It's just that I don't want to explain what
exactly I do which doesn't leave much else to say :)

The reason for my questions is that I suspect that Hiarcs is doing something
similar to Rybka. The per-node tactical strength of Hiarcs is phenomenal. It
would be interesting to compare notes although in practice this is unlikely to
happen. (It's already unlikely that I find out who actually looks at the Hiarcs
sources :))

Re. Rybka Beta vs Rybka Preview 1.01, I changed Rybka's node-counting to give
partial credit for "extra work done inside a node". This was done to stabilize
the nps figure as in Rybka Beta it can fluctuate wildly. Uri was quite
displeased with this change :)

Vas

>>
>>Thanks,
>>Vas
>>
>>>Position 1 [time-to-ply oddness]:
>>>[D]4k3/1b4R1/q3p2p/2B2p2/8/1R6/1K6/8 w - - 0 88
>>>
>>>
>>>Rybka 1.0 beta 64 bit 512mb hash A64-2.7 (CP GUI):
>>>
>>>00:00:00.0 2.40 3 65 Rbxb7
>>>00:00:00.0 2.41 4 435 Rbxb7
>>>00:00:00.0 2.65 5 1210 Rbxb7 Qe2+ Kb3
>>>00:00:00.1 5.63 6 2476 Rbxb7 Qe2+ Kb3 Qd3+ Kb4
>>>00:00:00.1 5.63 7 4189 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5
>>>00:00:00.1 5.63 8 8222 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5
>>>00:00:00.1 6.50 9 16000 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5 Qd2+
>>>00:00:00.3 6.56 10 41344 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+
>>>00:00:00.5 7.43 11 118209 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+
>>>00:00:01.0 7.43 12 200997 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+
>>>00:00:02.0 7.43 13 477000 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+
>>>00:00:04.0 7.49 14 987862 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+
>>>00:00:08.2 7.49 15 2073381 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+
>>>00:00:17.7 7.49 16 4778724 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+
>>>00:00:49.2 8.35 17 19664761 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qb1+
>>>00:01:24.2 8.35 18 31087304 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qb1+
>>>00:02:51.4 8.41 19 66442682 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qb1+
>>>
>>>
>>>Rybka 1.01 Preview 2 64 bit 512mb hash A64-2.7 (CP GUI):
>>>
>>>00:00:00.0 2.40 3 946 Rbxb7
>>>00:00:00.0 2.41 4 2031 Rbxb7
>>>00:00:00.0 2.65 5 5626 Rbxb7 Qe2+ Kb3
>>>00:00:00.0 5.63 6 10221 Rbxb7 Qe2+ Kb3 Qd3+ Kb4
>>>00:00:00.0 5.63 7 15896 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5
>>>00:00:00.1 5.63 8 30368 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5
>>>00:00:00.1 6.50 9 53606 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Ka5 Qd2+ Kb6 Qb2+
>>>Kc6
>>>00:00:00.2 6.56 10 111628 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ Kc6
>>>Qxg7 Rxg7
>>>00:00:00.6 7.43 11 238776 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ Kc6
>>>Qxg7 Rxg7 f4
>>>00:00:01.0 7.43 12 407147 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qe2+ Kb6 Qb2+ Kc6
>>>Qxg7 Rxg7 f4
>>>00:00:02.6 7.43 13 1136628 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6
>>>Qa8+ Kb5 Qxb7+
>>>00:00:04.7 7.49 14 2097110 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6
>>>Qa8+ Kb5 Qxb7+
>>>00:00:08.3 7.49 15 3812588 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6
>>>Qa8+ Kb5 Qxb7+
>>>00:00:18.0 7.49 16 8337775 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qd2+ Kb5 Qd3+ Kb6 Qd8+ Ka6
>>>Qa8+ Kb5 Qxb7+
>>>00:00:59.2 7.49 17 29186799 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qb1+ Ka5 Qxb7 Rxb7 h5 Re7+
>>>Kf8 Rxe6+ Kf7
>>>00:01:50.9 8.35 18 55194222 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qb1+ Ka5 Qxb7 Rxb7 f4 Re7+
>>>Kf8 Rxe6+ Kg7
>>>00:03:30.0 8.41 19 104388523 Rbxb7 Qe2+ Kb3 Qd3+ Kb4 Qb1+ Ka5 Qxb7 Rxb7 f4 Kb5
>>>f3 Kc6 f2
>>>
>>>
>>>
>>>Position 2 [nps oddness]:
>>>[D]r2qkb1r/1p1b1ppp/p1nppn2/6B1/3NP3/2N5/PPPQ1PPP/2KR1B1R w kq - 0 1
>>>
>>>
>>>Rybka 1.0 beta 64 bit 512mb hash A64-2.7 (CB GUI):
>>>
>>>9.Nxc6
>>>  ²  (0.37)   Depth: 3   00:00:00
>>>9.Kb1
>>>  ²  (0.39)   Depth: 3   00:00:00
>>>9.Kb1
>>>  ²  (0.32)   Depth: 4   00:00:00
>>>9.Kb1 Be7
>>>  ²  (0.27)   Depth: 5   00:00:00
>>>9.Kb1 h6 10.Be3
>>>  =  (0.22)   Depth: 6   00:00:00  24kN
>>>9.Be2 Be7 10.Nxc6
>>>  =  (0.24)   Depth: 6   00:00:00  28kN
>>>9.Be2 Be7 10.Kb1 0-0
>>>  =  (0.25)   Depth: 7   00:00:00  49kN
>>>9.Be2 Be7 10.Qe3 0-0 11.Qg3
>>>  =  (0.23)   Depth: 8   00:00:00  93kN
>>>9.Be2 h6 10.Be3 Be7 11.b3 0-0
>>>  =  (0.16)   Depth: 9   00:00:01  215kN
>>>9.Nxc6 Bxc6 10.f3 Qa5 11.Bf4 0-0-0
>>>  ²  (0.28)   Depth: 9   00:00:01  259kN
>>>9.Nxc6 Bxc6 10.f3 Qa5 11.Kb1 Be7 12.Be2
>>>  ²  (0.28)   Depth: 10   00:00:02  389kN
>>>9.Nxc6 Bxc6 10.f3 Qc7 11.Kb1 Be7 12.Bf4 Rd8
>>>  =  (0.21)   Depth: 11   00:00:04  674kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0
>>>  =  (0.23)   Depth: 11   00:00:11  1658kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0 13.Nf3
>>>  =  (0.24)   Depth: 12   00:00:16  2429kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8
>>>  =  (0.20)   Depth: 13   00:00:27  4147kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8
>>>  =  (0.19)   Depth: 14   00:01:02  9484kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 0-0-0 12.Nxc6 Bxc6 13.Qe3 Kb8
>>>  =  (0.20)   Depth: 15   00:02:28  23322kN
>>>
>>>
>>>Rybka 1.01 Preview 2 64 bit 512mb hash A64-2.7 (CB GUI):
>>>
>>>9.Nxc6
>>>  ²  (0.37)   Depth: 3   00:00:00
>>>9.Kb1
>>>  ²  (0.39)   Depth: 3   00:00:00
>>>9.Kb1
>>>  ²  (0.32)   Depth: 4   00:00:00
>>>9.Kb1 Be7
>>>  ²  (0.27)   Depth: 5   00:00:00
>>>9.Kb1 h6 10.Be3
>>>  =  (0.22)   Depth: 6   00:00:00  50kN
>>>9.Be2 Be7 10.Nxc6
>>>  =  (0.24)   Depth: 6   00:00:00  61kN
>>>9.Be2 Be7 10.Kb1 0-0
>>>  =  (0.25)   Depth: 7   00:00:00  103kN
>>>9.Be2 Be7 10.Qe3 0-0 11.Qg3
>>>  =  (0.23)   Depth: 8   00:00:00  203kN
>>>9.Be2 h6 10.Be3 Be7 11.b3 0-0
>>>  =  (0.16)   Depth: 9   00:00:01  442kN
>>>9.Nxc6 Bxc6 10.f3 Qa5 11.Bf4 0-0-0
>>>  ²  (0.28)   Depth: 9   00:00:01  519kN
>>>9.Nxc6 Bxc6 10.f3 Qa5 11.Kb1 Be7 12.Be2
>>>  ²  (0.28)   Depth: 10   00:00:02  752kN
>>>9.Nxc6 Bxc6 10.f3 Qc7 11.Kb1 Be7 12.Bf4 Rd8
>>>  =  (0.21)   Depth: 11   00:00:04  1250kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0
>>>  =  (0.23)   Depth: 11   00:00:11  3124kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 0-0 13.Nf3
>>>  =  (0.24)   Depth: 12   00:00:16  4578kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8
>>>  =  (0.20)   Depth: 13   00:00:28  7698kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 h6 12.Bh4 Rd8 13.Nb3 Bc8 14.Qe3
>>>  =  (0.19)   Depth: 14   00:01:02  17348kN
>>>9.f4 Qc7 10.Be2 Be7 11.Kb1 0-0-0 12.Nxc6 Bxc6 13.Qe3 Kb8 14.e5 dxe5 15.fxe5 Nd5
>>>  =  (0.20)   Depth: 15   00:02:29  41641kN



This page took 0.02 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.