Author: Robert Hyatt
Date: 07:28:32 09/11/99
Go up one level in this thread
On September 11, 1999 at 02:05:20, Ed Schröder wrote: >On September 10, 1999 at 18:36:12, Fernando Villegas wrote: > >>On September 10, 1999 at 11:50:50, Ed Schröder wrote: >> >>>On September 10, 1999 at 07:50:36, Micheal Cummings wrote: >>> >>>>On September 10, 1999 at 06:04:42, Ed Schröder wrote: >>>> >>>>>>Perhaps Ed should clarify this issue. >>>>> >>>>>My only goal is to find out how Rebel will do against GM's at tournament >>>>>time control. Because of the attention the event gets I can afford to put >>>>>some money and time in it. >>>>> >>>>>As for the question, "count this game or not" I on my abacus will not >>>>>count this game because the game was not representative to the above >>>>>mentioned goal. >>>>> >>>>>Officially this game counts of course. A PC can break any time. A pity but >>>>>it is simply all in the game. A human can go to the bathroom and lose on >>>>>time, no win on my personal abacus then but the win officially will count. >>>>> >>>>>Anyway in 8 days Rebel can try again against Lithuania (1 x GM and 3 x IM). >>>>> >>>>>Ed Schroder >>>> >>>>Hi Ed >>>> >>>>Do you think that all the problems are fixed, are you changing any program or >>>>hardware for this upcoming match. And if so do you think that it is worse or >>>>better for your chances at winning. >>> >>>I will change hardware. I am after the Athlon-650 at the moment. I don't >>>think the Athlon-650 will run faster than the K6-III-600 I am having now. >> >>Perhaps Bruce is right and you are again putting Rebel riding a wild ponny. I >>would love to see how Rebel does using a more conventional hardware as that we >>generally have. I mean, a 233, 266 or so Mhz stuff. >>fernando > >Of course Bruce is right and you have a point too. But I am sure the >majority wants to see the best software + the best (fastest) hardware. >And I definitely belong to that majority :-) > >A chess programmer's main wish and goal is to see his baby to perform as >strong as possible. That is his driving force. For me that is 18 years >now and I am still not cured. Will it ever stop? :-) > >Ed > > Nope. I've been doing it for 31+ years and I feel exactly the same. However, one suggestion: put yourself together a test suite, that you can run on stable hardware to get precise node counts, then run this on your bleeding edge hard- ware to verify that (at least at that point in time) it is producing correct results. Harry and I did this for Cray Blitz, because (a) we always ran on prototype hardware; (b) we ran on development machines which means that the compiler/libraries could be flakey. we could at least confirm that the code we were running produced the exact same results as what it would in normal hardware. Of course, the hardware can overhead after the game is in progress and none of this helps a bit. Happened to me in more than one ACM event. But we did catch a _lot_ of developmental compiler bugs and occasionally a hardware bug (we used to tear up the new gather/scatter vector load/store hardware when it was first done.) > >>>I also think book preparation must be done better as the Benko-gambit line >>>was not good for Rebel, too much strategic involved. I am not so optimistic >>>anymore than I was after the 2 nice draws. >>> >>>Ed
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.