Author: John Merlino
Date: 22:19:40 09/06/02
Go up one level in this thread
On September 06, 2002 at 23:23:56, David Dahlem wrote: >On September 06, 2002 at 22:13:38, David Dahlem wrote: > >>On September 06, 2002 at 22:08:10, John Merlino wrote: >> >>>On September 06, 2002 at 21:53:58, John Merlino wrote: >>> >>>>On September 06, 2002 at 21:39:16, David Dahlem wrote: >>>> >>>>>On September 06, 2002 at 21:25:56, John Merlino wrote: >>>>> >>>>>>On September 06, 2002 at 21:20:33, David Dahlem wrote: >>>>>> >>>>>>>On September 06, 2002 at 21:16:35, John Merlino wrote: >>>>>>> >>>>>>>>On September 06, 2002 at 21:14:50, David Dahlem wrote: >>>>>>>> >>>>>>>>>Hi >>>>>>>>>I believe John Merlino said that in CM 9000 it would be possible to know the >>>>>>>>>solution times of the "Solve for Mate" option. I can't figure out how to know >>>>>>>>>the solution times. I tried the Mentor Lines window, but it appears to give >>>>>>>>>basically the same info as the regular Thinking Lines window. John, are you >>>>>>>>>there? >>>>>>>>>Regards >>>>>>>>>Dave >>>>>>>> >>>>>>>>The solution time is displayed in the Mentor Lines window, along with the rest >>>>>>>>of the thinking output while the Solve for Mate feature (or any other mentor >>>>>>>>feature, such as Advice, Game Analysis, etc.) is running. >>>>>>>> >>>>>>>>jm >>>>>>> >>>>>>>Well, i tried that, John. But the Mentor Lines window shows the mates found long >>>>>>>before the "Solve for Mate" actually solves the mate. Maybe i am missing >>>>>>>something somewhere? >>>>>>>Regards >>>>>>>Dave >>>>>> >>>>>>Send me the position you are using, so I can see what you are talking about. >>>>>>Also send me the exact steps you are doing in the interface. >>>>>> >>>>>>jm >>>>> >>>>>Hi John >>>>>Here is one of the positions i tried, a mate in 6. >>>>>6k1/R2b1p2/3N1npp/3p4/3P4/4PN2/2Q2PPP/6K1 w - - 0 1 >>>>>With both players human, i went to edit/paste/Forsythe Board Position and pasted >>>>>the position. The Mentor Lines window has to be opened first, since there is no >>>>>way to open it after "Solve for Mate" has started. I get the following in the >>>>>Mentor Lines window, and Solve for Mate still hadn't solved the mate after two >>>>>minutes, so i stopped it. >>>>> >>>>>Time Depth Score Positions Moves >>>>>0:00 1/8 Mate07 100162 1.Nxf7 Kf8 2.Qxg6 Ke7 3.Nd6 Ke6 >>>>> 4.Qf7+ Kxd6 5.Qxf6+ Be6 6.Qe7+ >>>>> Kc6 7.Qc5# >>>>>0:01 2/9 Mate07 172016 1.Nxf7 Kf8 2.Qxg6 Ke7 3.Nd6 Ke6 >>>>> 4.Qf7+ Kxd6 5.Qxf6+ Be6 6.Qe7+ >>>>> Kc6 7.Qc5# >>>>>0:05 3/10 Mate06 1036080 1.Nxf7 Kf8 2.N7e5 Ke7 3.Qc8 Ke6 >>>>> 4.Rxd7 Nxd7 5.Qxd7+ Kf6 6.Qf7# >>>>>0:12 4/11 Mate06 2731461 1.Nxf7 Kf8 2.N7e5 Ke7 3.Qc8 Ke6 >>>>> 4.Rxd7 Nxd7 5.Qxd7+ Kf6 6.Qf7# >>>>> >>>>>Regards >>>>>Dave >>>> >>>>Well, it appears we have our first bug. Congratulations! Apparently the Solve >>>>for Mate feature has some issues. >>>> >>>>This will be addressed in the next patch. Until then, you CAN view the output of >>>>this feature in the Mentor Lines window. You will notice that, when this feature >>>>is running, the X and Y values in the "Depth" column are always the same number, >>>>such as 1/1, 3/3, 5/5, etc. When I run this feature on the above position (on my >>>>PIII-600), it reports Mate in 6 in 17 seconds. When I use the default engine to >>>>analyse this position, the Mentor Lines window reports mate in 6 after 8 >>>>seconds. >>>> >>>>So, although the Solve for Mate feature itself may never actually announce that >>>>it has found a mate, you can always see its output in the Mentor Lines window. >>>> >>>>We apologize for that.... >>>> >>>>jm >>> >>>Well, maybe it isn't a bug after all. But I would definitely file it under >>>"potential design flaw". What happens with the Solve for Mate feature is that if >>>the engine finds a mate at a certain depth, it CONTINUES searching all other >>>moves at that depth. So, the unfortunate result of this feature's behavior is >>>that, even though the engine finds a mate in 6 after 18 seconds, it doesn't >>>FINISH with that depth until 4:18!! At THAT point the Solve for Mate dialog will >>>announce the mate. >>> >>>I am going to discuss this with Johan, and I suspect that the result will be >>>that the first mate found will trigger the announcement, rather than reaching >>>the end of that entire depth's search. >>> >>>jm >> >>I am wondering more about this. I get the same solution time in the Mentor Lines >>window without starting the Solve for Mate. But also notice it first says mate >>in 7, in both cases. Isn't this the same thing that the regular Thinking Lines >>window does. As we discussed in a previous thread, isn't this the same as >>regular analyze mode, not necessarily showing the shortest mate. I will test >>this on some known longer mates. >>Regards >>Dave > >I did some testing with a few longer known mates, and i think the reason what i >copied to the clipboard from the previous position first shows a mate in 7, is >because it's not possible to copy the contents of the Mentor Lines window to the >clipboard, or do anything else at all, while Solve for Mate is running. Stopping >Solve for Mate causes the Mentor Lines window to revert back to regular >analysis. Also once Solve for Mate announces a solution, the Mentor Lines window >reverts back to regular analysis. So copying the contents of the Mentor Lines >window only shows regular analysis, not Solve for Mate analysis. It is possible >to constantly watch the Mentor Lines window and visually see the solution time >for Solve for Mate. But if it solves the mate while i have gone for coffee, the >solution time is lost. :( >This is less than ideal, if a user wants to know the solution time. >Also, what is the default hash size for the Mentor engine, and is possible to >change it? >Regards >Dave Assuming this is addressed in the patch, when a mate is announced, we will also make sure that the time and depth are displayed as well. Also, you can change the hash size of the mentor engine by going to "Preferences -> Mentor". jm
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.