Author: Melvin S. Schwartz
Date: 16:05:21 06/13/99
Go up one level in this thread
On June 13, 1999 at 14:59:22, Mark Young wrote: >On June 13, 1999 at 11:03:24, Melvin S. Schwartz wrote: > >> >>On June 12, 1999 at 17:28:58, Mark Young wrote: >> >>>On June 12, 1999 at 16:25:58, Melvin S. Schwartz wrote: >>> >>>> >>>>On June 12, 1999 at 12:34:02, Mark Young wrote: >>>> >>>>>On June 12, 1999 at 11:17:59, blass uri wrote: >>>>> >>>>>> >>>>>>On June 12, 1999 at 09:01:50, Melvin S. Schwartz wrote: >>>>>> >>>>>>> >>>>>>>On June 12, 1999 at 06:08:06, blass uri wrote: >>>>>>> >>>>>>>> >>>>>>>>On June 12, 1999 at 00:27:49, Tania Devora wrote: >>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>Hi guys, this is the number 12 game under tournament controls. I am AMAZED, >>>>>>>>>Fritz5.32 has lost again !! >>>>>>>> >>>>>>>>I am not surprised because Hiarcs7.32 is better than >>>>>>>Fritz5.32 >>>>>>> >>>>>>>After just 10 games? Tania does not give Fritz the ampount of Hash-tables he >>>>>>>needs. This deficiency of HT could and probably is affecting his play. According >>>>>>>to published data and my conversatons with ChessBase, Fritz depends a lot on HT, >>>>>>>and his playing will be affected by a drastic reducton of what is recommended by >>>>>>>the formula in Fritz's manual. >>>>>> >>>>>>Hiarcs7.32 is also affected by the fact that it cannot use >>>>>>the nalimov tablebases because tania has not enough memory in the harddisk. >>>>>>It is possible that it drew some games that it could win with nalimov tablbases. >>>>>> >>>>>>I know that mark young believes that Hiarcs7 is better than Fritz5.32 >>>>>>and Hiarcs7.32 is probably 25% faster than Hiarcs7. >>>>>> >>>>>> >>>>>>>>Hiarcs needs to play against better opponents. >>>>>>> >>>>>>>Did you say better opponents than the #1 rated program by the SSDF? What is >>>>>>>better than Fritz? CM6000? >>>>>> >>>>>>I hope that Junior5.4 is better >>>>>>It is not commercial >>>>>>The comercial version of Junior is Junior5 >>>>>> >>>>>>I also have problem to trust the 28.5:11.5 of Hiarcs7 against Junior5 in the >>>>>>ssdf games when all the other tests of public games showed better results for >>>>>>Junior5. >>>>>>Hiarcs7.32(better than Hiarcs7) won against Junior5 in the games that were >>>>>>posted here only 11.5:8.5(the sides did not use the original opening book and it >>>>>>was something similiar to the nunn test) >>>>> >>>>>I will get my copy of Hiarcs7.32 monday, the first match I will play against >>>>>Hiarcs7.32 will be Junior 5. I also seen the Hiarcs7.32 and Junior 5 results, >>>>>and they seem a bit odd to me. Junior 5 I have found to be a solid program and >>>>>it tends to crush other programs more then itself getting crushed. The 28.5 to >>>>>11.5 results in SSDF is pretty wide for to top SSDF programs playing each other, >>>>>but anything can happen in a 40 game match.:) >>>>> >>>>>It will be interesting to see if SSDF match results can be reproduced. I have >>>>>enough memory(256mb), and disk space(10GB free). To max out both programs it >>>>>term of Hash, and Table base space for a match at 40/2hr. >>>>> >>>>Mark, if I remember correctly, you recently told me the SSDF's rating system was >>>>very sound. >>> >>>I see you are confused again, I said the rating method they are using is sound. >>>It does not mean that they do not make mistakes in plugging in the data or >>>playing the games. That is why many of us look at the SSDF games here in CCC, >>>and if something strange is found, see if the game or results can be reproduced. >>> >>Hello Mark, >> >>I kind of think the reply I received from Brett is more meaningful than yours. >>First you say it's sound than you say it's not perfect. Well...I guess we have >>different meanings of what sound is! >> >>> Has your position on that changed? If not, how would you explain >>>>Fritz 5.32 being number 1? >>> >>>My position has not changed, but it help if the person I am talking to has a >>>common frame of reference in how ratings systems work and computer chess in >>>general. You sir do not. >>> >>Mark, >> >>You have a penchant for setting the wrong tone with me in debating an issue. You >>may think you are correct but I am entitled to disagree without your degrading >>remarks, If you feel I am so unintelligent about ratings and computer chess >>etc,. then I wish you would not reply to any of my posts. I don't need your kind >>of reples - go teach someone else! >> >> >>>>Of course the difference between the top three is very small and relatively >>>>insignificant, but if you feel so strongly that Hiarcs 7 is better than Fritz >>>>5.32, the SSDF testing will not back you up. Furthermore, their testing of >>>>Hiarcs 7 against Fritz 5.32 in 40 games reveals a score of Fritz 22 and Hiarcs >>>>18. This is not an indication of Hiarcs supremacy over Fritz >>> >>>Again you show your lack of knowledge about computer chess. I test programs on >>> >>Mark, >> >>My lack of knowledge...hmmm. Oh well, so much for the meaning of the SSDF's >>results. >> >>PII 400 or faster hardware. Just because Fritz 5.32 is the best program on P200 >>>hardware, this does not mean Fritz 5.32 is also #1 on PII 400 hardware. My That is correct. However, it also doesn't necessarily have to mean the opposite either. >>>results do not conflict or disagree with SSDF's results. I test under different >>>hardware conditions then SSDF. My testing shows, that Hiarcs7 is stronger then >>>Fritz 5.32 on PII 400 hardware. I do not "FEEL" this, my data shows this. Well, that certainly makes sense. If you've done extensive testing with Fritz getting his proper HT and F5 ctg book, then I would agree you have meaningful results. >>> >>>So you don't think I am the only one who finds Hiarcs7 is stronger then Fritz >>>5.32 I show you the SSR list. I have some reservations about this test for the reasons stated below. I believe they once had a tournament posted and the 10 game match between Hiarcs 7 and Fritz 5.32 was something like 6-4 in favor of Fritz. Of course that small amount of games doesn't prove much except that you couldn't call one clearly superior to the other. In fact, if you look at the difference in their ratings, it is only a 19 point difference - so what's the big deal? >>> >>>SELECTIVE SEARCH RATINGS >>>1. 253 2629 Hiarcs 7.1 - still top, beat Fritz532 by 8-4, Rebel 10 by 7.5-4.5, >>>Genius5 by 6.5-5.5 and drew 6-6 with Nimzo98 in latest results from my Magazine >>>readers. >>>2. 251 2610 Fritz 532 >>>3. 250 2602 ChessMaster 6000 - may not stay so high - most results in are >>>against weak opposition >>>4. 248 2590 Nimzo 99a - the 'a' upgrade finally moves Nimzo 99 above Nimzo 98! >>>5. 248 2588 Fritz 516 >>>6. 248 2585 Nimzo 98 >>>7. 247 2580 Junior 5 - the massive loss to Hiarcs7 in Sweden has hurt J5 - I >>>think it's better than this >>>8. 246 2571 Rebel 10 >>>9. 246 2571 Hiarcs 6 >>>10. 244 2556 Rebel 9 >>>11. 244 2551 Shredder 3 >>>12. 243 2547 MChess Pro 7 >>>13. 242 2539 Genius 5 >>>14. 242 2538 MChess Pro 8 - some poor results in Sweden, especially the crushing >>>defeat by Fritz 5.32, have hurt MCP8 >>>15. 241 2534 Shredder2 >>>16. 238 2508 Gandalf3 >>>17. 236 2492 Junior4.6 >>>18. 235 2483 Kallisto2 >>>19. 224 2398 CS-Tal DOS >>> >>Mark, >> >>The Selective Search rating list, I believe, is run by Eric Hallsworth who is >>the one responsible for the opening book in Hiarcs. It is just a little awkward >>that he is running that site and perhaps doing the testing. Anyway, you didn't >>state the time controls for that match between Hiarcs and Fritz. I have seen a >>10 game match between Hiarcs 7 and Fritz 5.32 with a score of 6-4 in favor of >>Fritz; furthermore, I have also seen a match between Hiarcs 7 and MCP8 at 40/2 >>with the score 6.5 - 3.5 in favor of MCP8. So what does it all mean - probably >>nothing. By the way, I am not a Fritz 5.32 fanatic! I have both programs and >>only respond as I see fit. > > >I DO NOT SAY FRITZ IS BETTER THAN HIARCS OR THAT >>HIARCS IS BETTER THAN FRITZ. I simply do not have the answer. I believe that >>much more testing be done before we come to any conclusion - I am not interested >>in mere speculation as being the foundation for a disagreement. > >We agree. Hold it right there! "We agree". I'm going to print that out right now!! I have had Hiarcs7 for many months and have played 100’s of games. I >have played the best commercial programs as well as the best and strongest >programs on the Internet, not to mention many strong human players GM and IM >included from around the world with Hiarcs7. Listen, Mark, I have stated before to Uri Blass that I prefer Hiarcs 7 over Fritz 5.32. It's just that I haven't seen conclusive evidence to make that kind of statement as fact. The testing you claim to have done is knowledgeable to YOU but until now - not to me. It wouldn't surprise me a whole lot if Hiarcs 7 is better than Fritz 5.32. I would not be surprised at all if Hiarcs 7.32 is better than Fritz. I just like to see fair testing done extensively in regards to any program vs. program. I do not care to see wild claims made about program A vs. program B without solid proof. > >I have reached some conclusions after looking at the performance data and >reviewing many of the games of Hiarcs7. It is the strongest program I have >tested to date. I do say this because I am a fan of Hiarcs7; my personal >favorite program is Rebel. I came to this conclusion because Hiarcs7 has the >best performance of any program in my testing. I don't have a problem with that at all. > >I have come to no conclusion about Hiarcs7.32, as the testing on this program >version has not started. > We agree. And I wouldn't find it hard to believe Hiarcs 7.32 is better than Fritz 5.32. It would be very logical to believe that if Hiarcs 7.32 is better than Hiarcs 7, then Hiarcs 7.32 should definitely be better than Fritz. The thing I like about Hiarcs is the way it plays chess - more chess theory than Fritz. Mel> >> >>Regards, >>Mel >>>>> >>>>>> >>>>>>I also found that some programs on pentium90 performed better than 11.5:28.5 >>>>>>against Hiarcs7(pentium200). >>>>>> >>>>>>These games are not public so I cannot check if there is no mistake. >>>>>> >>>>>>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.