Author: Dann Corbit
Date: 14:34:25 11/17/04
Go up one level in this thread
On November 17, 2004 at 14:20:09, James T. Walker wrote: >On November 17, 2004 at 05:34:02, Uri Blass wrote: > >>On November 16, 2004 at 21:08:19, Dann Corbit wrote: >> >>>On November 16, 2004 at 20:37:09, Wayne Lowrance wrote: >>> >>>>Being a proud owner of Chess Tiger 14, it pains me to observe posting results >>>>here of Tiger 2004. It seems after a long pause since Tiger 15 that Tiger 2004 >>>>is not performing, so far, as well as I would have hoped. >>>>I passed on buying Tiger 15 waiting for Tiger 16 (2004 , guess) and now I dont >>>>expect to buy this version of Tiger for the present time being. >>>>Very sad 4 me because Gambit tiger was/is a fun engine... >>>>With respect and not wishing to harm Christopher. >>> >>>From the sound of it, it is a complete redesign. >>>I have ordered a copy (which comes with Chess Assistant 8). >>>I expect to watch how it plays, to see how much I enjoy it. >>> >>>I expect big things from Christophe. He has a much faster NPS now, and so he >>>has more slack to play with than before. >> >> >>I think that nodes per second are irrelevant here. >> >>Being twice faster is not more than 50 elo at the ssdf time control and the gap >>between shredder and tiger is clearly bigger. >> >> >> >> This is an engine that may challenge >>>Shredder before long. Sometimes, a step sizeways (or even backwards) is >>>necessary before we can take a big step forwards. >> >>I do not expect it to challenge shredder. >>It may be better than shredder8 after enough time but shredder also gets better. >> >>Uri > >The problem I see is that it's 2 times faster (Nps) but not 50 Elo stronger. In >fact at blitz time control G/5min it is showing only about 10 Elo higher than >Chess Tiger 15 after over 600 games. If you increase the speed by a factor of 2 >and do nothing else you should get a better improvement than what is showing up >in my test. If he had simply put CT on a machine twice as fast, we would see 2x faster and + 50 Elo or so. But he has (obviously) changed the internal representations in some way. Probably, he will have to rewrite lots of things involving evaluation (though search should not need much change). CT has sat at #1 on the SSDF. If he decides to put the effort into it, it would be very unsurprising to me if he sat there again.
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.