Author: Dann Corbit
Date: 13:09:39 01/25/01
Go up one level in this thread
On January 25, 2001 at 15:38:54, Georg v. Zimmermann wrote: >Sorry about the multiple posting. >http://www.icdchess.com/forums/1/message.shtml?151959 >http://www.icdchess.com/forums/1/message.shtml?151960 >http://www.icdchess.com/forums/1/message.shtml?151972 >I hope a moderator will delete them. CCC is very slow for me today , I do not >know what caused this. > >Thanks for your help. I run the WAC at st 10, on a k6400 with standart hash and >no egtb, crafty compiled with MSVC6, using the winnt makefile´with adjusted /Zm. >Unfortunately I entered something into the dos window by accident while running >the test which stoped crafty (?!) and it only continued after I hit enter. Does >this render the result useless ? You cannot produce ANY keyboard input while the tests are running. If you accidentally bump the keyboard, backspace over whatever you wrote and press the enter key and it will start again. >I also got "EG fault: a problem occurred during epdpfga processing" after the >last position every time. What does this mean ? A blank line or end of file marker or anything else which the SAN kit is crabby about. The thing that really bugs me is that it stops whenever it finds an error. Hence, if you want to process 100 rows and something is wrong with row #2, it will stop at row 2 and not process any of the subsequent rows. I would rather see it: 1. Diagnose the error exactly, putting the bad record and a DESCRIPTION of what is wrong into the error log. 2. Continue on its merry way with the next record. >I got the following results : > >A.) > >total: 300 >total acd: 2714 mean total acd: 9.05 >total acn: 411804960 mean total acn: 1372683.20 >total acs: 2617 mean total acs: 8.72 >total mean node frequency: 157357.65 Hz >solve: 290 solve percent: 96.67 >solve acd: 2624 mean solve acd: 9.05 >solve acn: 397008916 mean solve acn: 1368996.26 >solve acs: 2514 mean solve acs: 8.67 >solve mean node frequency: 157919.22 Hz >unsol: 10 unsol percent: 3.33 Only 10 were unsolved. >unsol acd: 90 mean unsol acd: 9.00 >unsol acn: 14796044 mean unsol acn: 1479604.40 >unsol acs: 103 mean unsol acs: 10.30 >unsol mean node frequency: 143650.91 Hz > >B.) > >total: 300 >total acd: 2743 mean total acd: 9.14 >total acn: 409701793 mean total acn: 1365672.64 >total acs: 2632 mean total acs: 8.77 >total mean node frequency: 155661.78 Hz >solve: 288 solve percent: 96.00 >solve acd: 2632 mean solve acd: 9.14 >solve acn: 391506684 mean solve acn: 1359398.21 >solve acs: 2509 mean solve acs: 8.71 >solve mean node frequency: 156040.93 Hz >unsol: 12 unsol percent: 4.00 12 were unsolved >unsol acd: 111 mean unsol acd: 9.25 >unsol acn: 18195109 mean unsol acn: 1516259.08 >unsol acs: 123 mean unsol acs: 10.25 >unsol mean node frequency: 147927.72 Hz > >Which is better and by how much and what do all those numbers tell me ? >total: 300 300 EPD records in total were processed. >total acd: 2714 mean total acd: 9.05 The average depth in plies for the whole set was just a bit over 9. >total acn: 411804960 mean total acn: 1372683.20 The average nodes per record examined was 1.4 million. >total acs: 2617 mean total acs: 8.72 It took just under 9 seconds per record, on average. >total mean node frequency: 157357.65 Hz 160K NPS was the average speed. >solve: 290 solve percent: 96.67 290/300 solved is 96.67% correct. >solve acd: 2624 mean solve acd: 9.05 For the records that were solved, the average was a hair over 9 plies of depth. >solve acn: 397008916 mean solve acn: 1368996.26 For the records that were solved, about 1.4 million nodes were examined >solve acs: 2514 mean solve acs: 8.67 It took just under 9 seconds per record that was solved >solve mean node frequency: 157919.22 Hz The NPS speed for the records that were solved was 158K NPS >unsol: 10 unsol percent: 3.33 Ten records were not solved. This is just over 3% >unsol acd: 90 mean unsol acd: 9.00 The average depth of the unsolved problems was exactly 9 plies on average >unsol acn: 14796044 mean unsol acn: 1479604.40 The average nodes examined on the unsolved records was 1.48 million >unsol acs: 103 mean unsol acs: 10.30 The average time attempted on unsolved records was a fraction over ten seconds >unsol mean node frequency: 143650.91 Hz The NPS for unsolved records was about 144K NPS HTH Crafty has an egtb special help command.
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.