Author: Les Fernandez
Date: 21:01:13 12/10/04
Go up one level in this thread
On December 10, 2004 at 22:15:09, Uri Blass wrote: >On December 10, 2004 at 21:47:46, Uri Blass wrote: > >>On December 10, 2004 at 21:06:42, Uri Blass wrote: >> >>>On December 10, 2004 at 02:17:59, Les Fernandez wrote: >>> >>>>On December 10, 2004 at 00:09:02, Uri Blass wrote: >>>> >>>>>On December 09, 2004 at 23:57:26, Les Fernandez wrote: >>>>> >>>>>>On December 09, 2004 at 00:41:18, Uri Blass wrote: >>>>>> >>>>>>>Here are results for default movei for the first 11 positions in the seond set >>>>>>> >>>>>>>Note that movei was not designed to solve mates and I am sure it could do >>>>>>>clearly better than it in case that it was designed for that purpose. >>>>>>> >>>>>>>It gets a mate score in less than 62 seconds for everyone of them. >>>>>>>The big number is the number of nodes and I also gave translation of the score >>>>>>>to distance to mate. >>>>>>> >>>>>>>11 9987 371 2214115 mate in 7 d8g8 >>>>>>>14 9985 5462 28819188 mate in 8 b5a4 >>>>>>>14 9985 6195 32582992 mate in 8 b4a4 >>>>>>>14 9985 5337 27793757 mate in 8 c4a4 >>>>>>>14 9985 5728 29850523 mate in 8 d4a4 >>>>>>>14 9985 6157 32404314 mate in 8 b3a4 >>>>>>>14 9985 5748 30085900 mate in 8 c2a4 >>>>>>>11 9987 548 2972896 mate in 7 e8e5 >>>>>>>9 9987 195 978781 mate in 7 e7e1 >>>>>>>6 9993 6 22984 mate in 4 e5a5 >>>>>>>9 9987 160 836546 mate in 7 e4e1 >>>>>>> >>>>>>>Uri >>>>>> >>>>>>Hi Uri, >>>>>> >>>>>>First thanks for giving it a go. What hardware are you running them on and do >>>>>>you plan on running all of them? It is of interest to me to know the total time >>>>>>it took to solve the set, but understand that is is rather time consuming. >>>>>> >>>>>>Talk to you soon, >>>>>> >>>>>>Les >>>>> >>>>>I used A3000 >>>>>I will certainly not run all the positions manually. >>>>> >>>>>I may do it automatically not based on fixed time but based on stopping when it >>>>>finds mate but I need first to finish some tool to analyze epd file. >>>>> >>>>>Uri >>>> >>>>Hi Uri, >>>> >>>>I certainly can appreciate not doing it manually. Listen perhaps when you have >>>>time you can just look at the first 10 positions of the second set of positions >>>>since they were much more difficult. If you have trouble finding them I can >>>>always repost them. Thanks for your time. >>>> >>>>Les >>> >>>my results are for them and not for the easy positions. >>> >>>Uri >> >>I can add that I now test the easy positions automatically and it seems that >>part of them are not so easy. >> >>I expected based on Antony that movei on A1000 will solve all of them in a short >>time but it does not do it. >> >>I implemented function to analyze epd and not leave a position before mate >>unless very long time was used and I test it with this function. >> >>The fact that there are a lot of easy problems does not mean that all of them >>are easy and solving all the positions on p233 in less than a minute(not one of >>them) is clearly a good result. > >correction >I remembered wrong: > >Dann Corbit said: >"There is a utility that computes all the answers in under a minute on a Pentium >II, 166 MHz." > >There are not small number of cases when >Movei on A1000 needs some minutes to solve positions in the set that was defined >as the easy test. > >I see on the screen the following output that means that movei needed almost >1000 seconds to solve one of the position and more than 700 seconds to solve >another position(1000 seconds is the time limit that I gave movei to solve the >positions and with more than 1000 seconds it should not search even if it finds >no mate. > > 12 9983 98807 210114214 g4d4 a3e3 b8g3 f2g1 d4e3 g1h1 h3g2 h1g2 e3f2 g2h1 f8f5 >e2f1 f5h5 f1h3 h5h3 f3h2 h3h2 > > 12 9983 73247 155531978 h4d4 a3e3 b8g3 f2g1 d4e3 g1h1 h3g2 h1g2 e3f2 g2h1 f8f5 >e2f1 f5h5 f1h3 h5h3 f3h2 h3h2 > > 12 9983 14458 26428369 c3d4 a3e3 b8g3 f2g1 d4e3 g1h1 h3g2 h1g2 e3f2 g2h1 f8f5 >2f1 f5h5 f1h3 h5h3 f3h2 h3h2 > > 12 9983 14318 26932012 d3d4 a3e3 b8g3 f2g1 d4e3 g1h1 h3g2 h1g2 e3f2 g2h1 f8f5 e >2f1 f5h5 f1h3 h5h3 f3h2 h3h2 > >It seems that the positions are almost the same based on the pv and I suspect >that the relevant tool that Dan Corbit used simply does not clear it's hash >between positions. > >Uri Hi Uri, I must say first thank you for trying to solve these positions. When you are done with however many you can I would also be interested in knowing how long each ONE took to solve if you keep the output of the analyses. Good work Uri. Les
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.