Author: Robert Hyatt
Date: 07:41:34 05/04/01
Go up one level in this thread
On May 04, 2001 at 09:46:12, Uri Blass wrote: >On May 04, 2001 at 09:25:21, Robert Hyatt wrote: > >>On May 04, 2001 at 06:55:36, Uri Blass wrote: >> >>>On May 04, 2001 at 06:22:48, Uri Blass wrote: >>> >>>Here is Deep Fritz's analysis after 33...Qb5 34.Qf1(I gave it only to finish >>>depth 15 but Deep Fritz "feels" bad and the score dropped by 0.72). >>> >>>Deeper blue could not see it because it was a root processor and only after >>>trading queens it could understand that it was bad. >>> >> >> >>I don't know where that comes from but DB was _not_ a "root processor" by >>any definition of the word. The evaluation was 100% tip evaluation... that >>was the purpose of the hardware in the first place.. > >I know that the evaluation was changed significantly after trading queens and >not because of deeper search. > > >Here is the relevant output from Deeper blue's logfiles > > > > 3(4) 30^ T=0 >Qb5f1q ra1f1Q Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 > 3(5) 46 T=0 >Qb5f1q ra1f1Q Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 > 4(5) 26 T=0 >Qb5f1q ra1f1Q Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 > 5(5)[Qxf1](14) 14 T=0 >Qb5f1q ra1f1Q Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 > 6(5)[Qxf1](-16)[Qxf1](-16) -16v T=0 >Qb5f1q ra1f1Q Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 > 6(5)[Qxf1](-27)[Qe2](-1)[b6](0)[h5](9) 9 T=1 >Ph6h5 bg2f3B Qb5f1q ra1f1Q Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 > 7(5) #[h5](-13)######################################[Qe2](23)# 23 T=3 >Qb5e2 qf1e2Q Bf3e2q bc3e5N Re8e5b bg2e4P Re5e4b > 8(6) #[Qe2](-7)[Qe2](-7) -7v T=3 >Qb5e2 qf1e2Q Bf3e2q ra1e1 Be2f3 re1f1 Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P >Kg8g7 kh1g2 Rd8d4 kg2g1 Pb7b5 kg1g2 Rd4e4 > 8(6) #[Qe2](-8)####################################### -8 T=7 >Qb5e2 qf1e2Q Bf3e2q ra1e1 Be2f3 re1f1 Ph6h5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P >Kg8g7 kh1g2 Rd8d4 kg2g1 Pb7b5 kg1g2 Rd4e4 > 9(6) #[Qe2](-21)#[Qxf1](3)###################################### 3 T=46 >Qb5f1q ra1f1Q Ph6h5 bc3e5N Re8e5b bg2f3B Pe4f3b rf1f3P Kg8g7 kh1g2 Rd8d4 kg2g1 >Pb7b5 kg1g2 Rd4e4 >10(6) #[Qxf1](2)####################################### 2 T=130 >Qb5f1q ra1f1Q Ph6h5 kh1g1 Pb7b5 bg2f3B Pe4f3b bc3e5N Re8e5b rf1f3P Kg8g7 kg1g2 >Pc6c5 pb4c5P Re5c5p >11(6)[TIMEOUT] 2 T=168 >Qb5f1q ra1f1Q Ph6h5 kh1g1 Pb7b5 kg1h1 Kg8f8 kh1h2 Ne5g4 ne3g4N Ph5g4n pf5f6 >Rd8d6 kh2g1 Re8e6 rf1e1 >--------------------------------------- >--> 34. .. Qxf1 <-- 6/41:43 >--------------------------------------- >hash guess ra1f1Q,Guessing Rxf1 > 7(4) #[h5](-30)[h5](-30) -30v T=0 >Ph6h5 kh1g1 Pb7b5 kg1h1 Kg8f8 kh1h2 Ne5g4 ne3g4N Ph5g4n pf5f6 Rd8d6 kh2g1 Re8e6 >rf1e1 > 7(6) #[h5](-66)############################## -66 T=1 >Ph6h5 kh1g1 Pb7b5 kg1h1 Kg8f8 kh1h2 Ne5g4 ne3g4N Ph5g4n pf5f6 Rd8d6 kh2g1 Re8e6 >rf1e1 > 8(6) #[h5](-50)############################## -50 T=5 >Ph6h5 kh1g1 Pb7b5 kg1h1 Kg8f8 kh1h2 Ne5g4 ne3g4N Ph5g4n pf5f6 Rd8d6 kh2g1 Re8e6 >rf1e1 > 9(6) #[h5](-50)###################<ch> 'rf1' >[151 sec (main.c:1847)][cont]########### -50 T=20 >Ph6h5 kh1g1 Pb7b5 kg1h1 Kg8f8 kh1h2 Ne5g4 ne3g4N Ph5g4n pf5f6 Rd8d6 kh2g1 Re8e6 >rf1e1 >10(6) #[h5](-50)############################## -50 T=49 >Ph6h5 kh1g1 Pb7b5 kg1h1 Kg8f8 kh1h2 Ne5g4 ne3g4N Ph5g4n pf5f6 Rd8d6 kh2g1 Re8e6 >rf1e1 >11(6) #[h5](-48)############[TIMEOUT] -48 T=169 >Ph6h5 kh1g1 Rd8d7 rf1f2 Pb7b5 bg2f3B Pe4f3b bc3e5N Re8e5b rf2f3P Kg8g7 pg3g4 >Ph5h4 kg1g2 > >You can see that the evaluation before Qxf1+ was 0.02 pawns for black at depth >17 and the evaluation after Qxf1+ Rxf1 was -0.50 pawns against black at depth >15,16 and -0.48 against black at depth 17. > >It is clear that the change in the evaluation was not because of deeper search. > >Maybe the word root processor is not correct(I remember that Ed said in the past >that it is preprocessoring and not root processoring) but it clearly did some >evaluation in the root. > >I know that there are programs that do no evaluation in the root(yace and >gandalf). > >There may be more programs that do no evaluation in the root but I know it only >about yace and gandalf from the programmers. > >I see that deep fritz often does not do evaluations in the root and in this case >watching the evaluation of Deep Fritz gives the impression that Deep Fritz does >not evaluates in a different way after trading queens. > >I can see simiar numbers before trading queens and after trading queens and I do >not see it in Deeper blue's logfiles. > >Uri There are other possible explanations here too. The most likely is that the first search you gave produced a score for depth 16. It started on depth=17 but failed to produce a score/move before it timed out. However, when it started pondering, the partial results of that search were available in the transposition table and could easily influence the score for the search iterations done during pondering. I can clip a couple of entries from a Crafty log file to show this, and crafty does _no_ root processing whatsoever today, except for checking to see (at the root) if it should enable the "trojan check" code because that particular position is on the board right now.
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.