Author: Robert Hyatt
Date: 18:33:49 10/13/02
Go up one level in this thread
On October 13, 2002 at 15:47:14, Vincent Diepeveen wrote: >On October 13, 2002 at 15:35:12, Vincent Diepeveen wrote: > >In 2002 the depth 12 ply is perhaps small for some. >However i need to add some notes here. > >In 1997 i was at the world championship (in october) at >a PII-300 Mhz intel. that was one of the faster machines. I don't know what hareware era you were in, but in Jakarta I used a pentium pro 200. In the next WMCCC event (1997) I used a 500mhz alpha. Ferret and someone I don't recall used super-cooled alphas running at 750+mhz... > >Even then only very few hit 12 ply. I did 12 plies in 1996 in Jakarta. > >In fact no program with a decent leaf evaluation and >checks in qsearch and singular extensions (or even mate >extensions) got 12 ply. So... Only "crappy programs" could not get decent depth back then. That kind of reasoning cuts _both_ ways... You want to take "diep" as the model of perfect computer chess. It isn't, regardless of the outrageous statements you continually make. (Diep is tactically the strongest program in the world. Diep is the strongest program in correspondence in the world. Etc...) > >You cannot imagine how overwhelmed i was seeing the >search depths of ferret and darkthought there. > >I got with diep about 8-10 ply there. In positions with >a queen i normally got 8 or 9 ply. > >Please keep in mind that diep in 1997 did NOT use singular >extensions. I use *a lot* more extensions now than in 1997. > >I cannot remember 100% whether i used mate threat extensions >in 1997. > >Yet the majority of opponents i had: Fritz, Virtual chess >and such engines. Especially virtual chess. Incredible stupid >engines from 2002 standards. Just a piece square table. > >Really nothing that would score a single point now. > >No one got much above 9 ply. The few that got like 10-12 ply >were having huge forward pruning. very simple qsearches >and such. That is simply crap. Fritz was searching as deep as I was in Jakarta. We beat them that year but both confirmed that we were not out-searching nor getting out-searched by Fritz... Ferret and I were pretty equal in search depth, also, but you can ask Bruce to confirm that once again your imagination has run wild. > >No one got 12 ply with all those extensiosn deep blue used. > >In 1997 many games were still decided by huge blunders. >Tactical blunders every game. > >I remember a game diep-schach where i had a 8 ply search and >diep mated itself there kind of (thereby losing material) >whereas a simple king move would have prevented all that. > >Of course nowadays the program positionally sees that in evaluation, >but apart from that such tactical errors do not happen in blitz a lot >even anymore. That is a completely ridiculous statement. But not worth the effort to respond further to... > >Now if we look to when deep blue was developed from say 1987-1997 >we see that most games around that period were dominated by tactics. > >So creating a deep blue machine that was tactical very strong, >in itself that is not such a bad idea. Definitely in 1997 i remember >the zillions of statements that just getting more nodes a second would >solve chess soon. > >You must see it in the historic context. Admitting how easy it is to get >12 ply now, only shows how bigtime software progressed. Now please get >a pentium 133Mhz and then try to get 12 ply. > >Deep blue started long before the P5 was released. In fact 0.60 >micron technology was used for 486 computers (correct me if i'm wrong) >the last chip to see 0.60 micron was the pentium 60Mhz (with bug). Has nothing to do with deep blue development. They weren't trying to use the latest/greatest fab process. They used only what they could get quick turn around on, and that which could contain enough transistors to meet the design requirements. > >Now in all those tactical blundering computerchess events, deep blue >gets there with a 12 ply monster. > >Had you told me in 1994 that i would get 12 ply easily in the year 2002, >then i would have stared to you in disbelief, just like similar statements >of mine around 1997 were not taken serious by many that with so many nodes >a second getting 12 ply was not so impressive. > >>On October 13, 2002 at 14:40:12, James Swafford wrote: >> >>I wonder why. there is just one person ever in this whole >>planet who said 12(6) = 18 ply, and that's robert hyatt. >> >>there is 4 the same statements from hsu. >> a) page 5 at his paper. >> b) 3 in this talk >> >>there is many statements from hyatt in 98 and 99 that it was >>getting 12 ply, when in 99 at world champs many got 12 ply it was >>18 ply suddenly... >> >>But the clearest statement is next: >> >>EeEk(* DM) kibitzes: kib question from ardee: Does "12(6)" mean 12 total ply or >>12+6=18 total ply? This has the been source of huge arguments for years! >> >>directly the answer came a few seconds later: >> >>CrazyBird(DM) kibitzes: 12 total in terms of brute force. 6 is just the max >>partition in hardware. >> >> >> >> >>>On October 13, 2002 at 14:25:00, Vincent Diepeveen wrote: >>> >>>>On October 13, 2002 at 12:00:39, Andrew Williams wrote: >>>> >>>>It was answerred another 2 times that it's 12 ply simply, >>>>excluding qsearch+extensions. So the hardware is inside that >>>>12 ply, but the hardware depth FROM that 12 ply can be UP to >>>>6 ply. So it isn't always searching 6 ply in hardware. It's >>>>a variable depth which they can do in hardware. That explains >>>>probably extensions around mainlines with captures very well >>>>and do not forget that the chips could do up to 1 billion nodes >>>>a second in theory. >>> >>>Ok, if you say that was answered, I believe you, but that >>>doesn't sound like what the quote I responded to in a thread below >>>meant. Can you provide the transcript for this? Seriously, >>>I would very much like to see it. >>> >>> >>> >>>> >>>>It searched on average over a period of 3 minutes only 126 MLN >>>>nodes from that. that's only 10% effective usage or so each >>>>chip. That means that there are *always* loads of chips idling. >>>> >>>>So getting extra processors to the PV is a very clever thing to >>>>do then. Of course you give the chips a small search depth then. >>>>2 to 3 ply is what Brutus was using at world champs 2002. Bigger >>>>search depths in hardware were too inefficient. >>>> >>>>You cannot use killermoves in hardware and you cannot use hashtables >>>>and Hsu didn't use nullmove either. With his last so many plies pruning >>>>(whatever name you want to give it, razoring, futility pruning) that means >>>>that when searching the Principal variation, you have to give many processors >>>>small search depths, because the pruning around principal variation is >>>>a lot less than for the other moves. >>>> >>>>It would be interesting to hear in a next chat how many plies the forward >>>>pruning was in the hardware part. Here just doing 1 or 2 ply didn't >>>>reduce anything. Yet i remember some statement from a talk at M$ from >>>>Hsu recently where (was it Tom kerrigan) asked Hsu about forward pruning >>>>in the hardware chips and he answerred he was doing that. >>>> >>>>So what type of pruning he did there is not so interesting. Interesting is >>>>that it saved them up to 90% nodes in hardware. That's very clever, because >>>>chrilly concluded that without forward pruning in hardware, your tree gets >>>>just TOO huge (because move ordering is near random). >>>> >>>>Chrilly uses nullmove. If Deep Blue used something else there. Only >>>>interesting thing from my viewpoint therefore is to know how many plies >>>>they did this forward pruning in hardware. >>>> >>>>My own experiments are not relevant here too much, because i use hashtable >>>>and nullmove everywhere. So it is very well possible that without nullmove >>>>the effect of forward pruning is way bigger than without. >>>> >>>>>Several people asked: >>>>> >>>>>Question: What does "12(6)" mean in Deep Blue's logs? >>>>> >>>>>Dr Hsu said, "12(6) means 12 plies of brute force (not counting the search >>>>>extensions & quiescence). 6 means the maximum hardware search depth allowed. >>>>>this means that the PV could be up to 6 plies deeper before quiescence." >>>>> >>>>>Unfortunately questions were being fed through a third-party, so it wasn't >>>>>possible to get a follow-up. >>>>> >>>>>Andrew
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.