Author: Uri Blass
Date: 17:31:24 10/28/00
Go up one level in this thread
On October 28, 2000 at 20:13:03, Joshua Lee wrote: >The only problem is that the new SSDF list has been changed by -100points the >results were kept the same for the games against humans however on Chris >Carson's webpage he has the results against humans as being 100points higher >than what is on the SSDF List Accident i guess. > >105. Mephisto Lyon 68020 12 MHz +67 (2117 12) 2050 19 -18 1442 59% 1986 >108. Mephisto Portorose 68020 12 MHz +54 (2081 10) 2027 26 -24 896 73% 1856 >117. Fidelity Mach IV 68020 20 MHz +103 (2077 19) 1974 18 -18 1495 48% 1988 >115. Mephisto MM 4 Turbo Kit 6502 MHz +148 (2022 9) 1874 19 -19 1374 47% 1897 >136. Fidelity Mach III 68000 16 MHz +74 (1967 25) 1893 14 -14 2410 52% 1880 >146. Mephisto Academy 6502 5 MHz +118 (1957 21) 1839 18 -18 1555 44% 1884 >119. Mephisto Almeria 68020 12 Mhz +18 (1930 26) 1912 28 -28 634 60% 1838 >161. CXG Sphinx Galaxy 6502 4 MHz +119 (1899 28) 1780 17 -17 1713 37% 1874 >144. Novag Super Forte C 6502 6 MHz +36 (1896 15) 1860 18 -18 1555 46% 1891 >160. Psion Atari 68000 8 MHz +65 (1847 5) 1782 18 -18 1487 44% 1828 >175. Saitek Maestro A 6502 6 MHz +85 (1796 9) 1711 30 -30 547 45% 1746 >174. Novag Forte B 6502 5 MHz +45 (1761 28) 1716 24 -24 837 44% 1760 > +77.666666666666666666666666666667 Average 207Games > >127. Mephisto Roma 68020 14 MHz (1879 9) 1930 27 -26 712 65% 1819 -51 >135. Novag Diablo 68000 16 MHz (1875 9) 1905 21 -21 1123 41% 1971 -30 >151. Novag Super Expert B 6502 6 MHz (1792 9) 1819 29 -29 564 48% 1835 -27 >152. Mephiso Mega IV 6502 5 MHz (1766 28) 1816 17 -18 1584 44% 1860 -50 >140. Mephisto Dallas 68000 12 MHz(1766 15) 1871 21 -20 1177 61% 1794 -105 >165. ConchessPlymateVictoria 6502 5.5 MHz(1741 7) 1767 26 -27 701 40% 1841 -26 >166. Fidelity Excel Club 68000 12 MHz (1674 6) 1758 23 -23 931 51% 1749 -84 >139. Mephisto MM 5 6502 5 MHz (1657 6) 1874 19 -19 1374 47% 1897 -217 >153. Saitek Maestro D 6502 10 MHz (1622 18) 1816 26 -26 736 51% 1809 -194 >168. Novag Super Forte A 6502 5 MHz (1618 9) 1738 19 -19 1398 37% 1830 -120 >185. Fidelity Excellence 6502 3 MHz (1478 5) 1657 22 -23 960 42% 1715 -179 >188. NovagSuperConstellation 6502 4 MHz (1455 9) 1631 18 -18 1626 35% 1740 -176 > > -104.91666666666666666666666666667 Average 130Games >337 games total this would be +39 -38 if it were 1 Program > >One other example is Fritz and Junior. >1 Fritz 6.0 128MB K6-2 450 MHz 2630 26 -25 775 67% 2505 >2 Junior 6.0 128MB K6-2 450 MHz 2592 23 -23 971 65% 2481 > >First off Fritz attained 2393 at a Tournament in Jakarta Feb but 2457 overall on >a P3-500 so is that -237 or -173?? for those interested fritz took around 4hrs >to find a good move by IM Situru e4-e3. > >As far as Junior goes it is getting a rating on a K6-2450 that it took a >700Mhz*8 to achieve ????????? Junior on 700*8 got performance of more than 2700(I think the average of the rating of the opponents was 2702) hello if that doesn't speak the obvious i don't >know what will. 3.1* for 4 Processors, 5.9* for 8 essentially 4.13Ghz >doubling the speed of a 450 3 times nets 3.600Ghz , however trippling twice nets >4.050Ghz fairly close. According to Professor Hyatt each time you tripple you >net 90points so this would mean that there should be a 180 point difference in >the SSDF rating for Junior and the rating it earned at Dortmund. I do not know about this theory. Hyatt had no theory about human-computer games and I also do not think that he has a theory that each time you tripple you net 90 points. I know that there is an evidence for diminishing returns in computer chess. This isn't the >case so here are the options subtract by 237, 173 or 180. > >In my opinion subtracting by 180 seems right , just look at Rebel 10 it was in >the 2450+ range in many games against GM's and IM's on similar hardware. I think that nothing is right. I believe that the most popular programs will always be weaker against humans because they are more known so you can stop trying to guess rating against humans from the ssdf list. Uri
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.