Author: stuart taylor
Date: 05:24:50 01/14/03
Go up one level in this thread
On January 14, 2003 at 07:26:11, Uri Blass wrote: >On January 14, 2003 at 07:11:32, Terry McCracken wrote: > >>On January 14, 2003 at 03:31:31, Uri Blass wrote: >> >>>On January 14, 2003 at 02:51:27, Terry McCracken wrote: >>> >>>>On January 14, 2003 at 02:32:05, Uri Blass wrote: >>>> >>>>>On January 14, 2003 at 01:44:30, Luis Smith wrote: >>>>> >>>>>>On January 13, 2003 at 22:06:44, Terry McCracken wrote: >>>>>> >>>>>>>On January 13, 2003 at 19:08:23, stuart taylor wrote: >>>>>>> >>>>>>>>Let's get it over with already, (all this DB worship)! >>>>>>>>I'm sure we can already say that whichever comes at the top of the ssdf (When >>>>>>>>Deep junior/Junior 8 gets tested, and with necesary correction patches), and if >>>>>>>>you put that program on the best, but regular hardware, I think we can well >>>>>>>>assume it has gone a bit over the DB which embarrassed Kasparov. >>>>>>>>I think we can finally safely assume that this multi-million dollar enterprise >>>>>>>>which made headlines everywhere, is already exceeded with our bedroom (or living >>>>>>>>room) furniture. It was normally 2 years between top existing machine, and level >>>>>>>>of what is commercially available. This will now be a clean 6 years. >>>>>>>> How much more can we dream?! >>>>>>>> >>>>>>>>I think that it's safe to say that top commercial computer chess strength has >>>>>>>>absolutely outclassed itself atleast twice since then. We have gone from Hiarcs >>>>>>>>6 on 266 Mhz. to Fritz 8 on about 2.7 Ghz. I really don't think that deeper blue >>>>>>>>would pose any real threat today. OK, it might be very near the top, maybe >>>>>>>>second or third, and might stay around there for a couple more years due to its >>>>>>>>speed challenge. But what we have now, overall, is surely better. >>>>>>>>S.Taylor >>>>>>> >>>>>>>Where's your data to support your claims? >>>>>>>Stuart, you haven't a clue what you're talking about, and you know it...or you >>>>>>>should! >>>>>>> >>>>>>>Without concrete data, you're talking through your hat. >>>>>>> >>>>>>>Terry >>>>>> >>>>>>Say for instance, Fritz 5.32 and Shredder 7 played a series of games. Fritz >>>>>>5.32 was on a Deep Blue's hardware and Shredder 7 had an 8xprocessor computer >>>>>>with each processor running at 1 ghz. Despite Shredder's increase in knowledge, >>>>>>I would still pick Fritz 5.32 to win simply because of the vast searching power. >>>>>> I myself using Fritz 7 on the playchess server have played against a Fritz 5.32 >>>>>>on a Dual AMD MP 2200 against my P III 1ghz, despite having a better book and >>>>>>newer software, he still tore me to peices... >>>>>> >>>>>>I think Deep Blue has an edge over an 8x processor commercial engine...I think >>>>>>you all underestimate searching power, I think Deep Blue will be well ahead of >>>>>>its time for even 2-5 years to come...but then again I know absolutely nothing >>>>>>about chess programming and am probably wrong ^_^ >>>>>> >>>>>>Reguards >>>>>>Luis >>>>> >>>>>Yes >>>>> >>>>>I think that you forget some important things. >>>>> >>>>>1)The games that you play are blitz and on blitz better knowledge is not >>>>>important. >>>>> >>>>>I suggest you to try 120/40. >>>>> >>>>>2)I doubt if shredder has better knowledge than other programs. >>>>>There are a lot of positions when Fritz's evaluation is more realistic. >>>>> >>>>>3)There are logfiles of the game and I found nothing impressive there(I do not >>>>>know about a single idea that programs of today needs hours to find). >>>>> >>>>>My impression based on the games is that deeper blue was not better in tactics >>>>>than the programs of today. >>>>> >>>>>I think that you are overestimating the searching power of deeper blue. >>>>> >>>>>Uri >>>> >>>>No..he is not! Deep tactics with such a deep search and how it was accomplished, >>>>is very impressive, by any standards. >>>> >>>>Terry >>> >>>Can you show me one example for deep tactics that Deeper blue saw in the games >>>against kasparov when commercial programs needs hours to find. >>> >>>Can you show me one line in the pv of deeper blue that top programs need hours >>>to find? >>> >>>Uri >> >>Look Uri, this has been rehashed here for years, do some research! >>Robert Hyatt can send you plenty of data, if you wish. I'm not going to spend >>copious amounts of time looking up data for you, when for the most part you >>throw out "Blanket Statements" about Deep Blue. >>If you have _any_ sense at all it's quite obvious that Deep Blue could do a >>tremendous amout of calculation which still far exceeds _any_ MP Box of today. >> >>Terry > >I see nothing as obvious > >For me it is even not obvious that deeper blue searched more than 20M nodes per >second. > >Nothing is obvious for players who stopped to play. >If you play that they were good you need to find proof in the games and not to >be based on number of nps that I am not sure if it is right and even if it is >right the program may have a lot of other bugs. > >It is hard to do a program without bugs and if the program is complex you can >expect it to have more bugs. > >I do not say that the programmers of deeper blue were inferior but their task >was clearly an hard task and I see no reason to assume as obvious that they >could do things without significant bugs and significant bugs can cause a >program to be slower. > >Uri They also had plenty of leeway to get away with alot of wasted silicon power, and that might have made it much closer to our own systems in playing strength than we realize. Again, only a guess, but quite possible, I think! S.Taylor
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.