Author: Henrik Dinesen
Date: 01:27:42 08/23/05
Go up one level in this thread
On August 22, 2005 at 19:04:08, George Speight wrote: >On August 22, 2005 at 04:36:09, Henrik Dinesen wrote: > >>On August 22, 2005 at 04:17:27, George Speight wrote: >> >>>On August 22, 2005 at 03:23:44, Jonas Cohonas wrote: >>> >>>>>Hi, Kurt I know you are not going to believe me, but your hash is set way too >>>>>low to be running 40/40 , unless your processor mhz. is as low as 700! George >>>> >>>>George, that is simply not true. >>> >>>I really would not know, to tell u the truth. I just know this guy at chessbase, >>>and he asked the programmers of Fritz,Shredder, and Junior that question. They >>>said the perfect setting for their programs in 40/40 was 128mb, assuming u have >>>more than a 1000 mgz processor. I just assumed they would be ina position to >>>know. Of course alot would depend on available ram. George >> >>That sounds weird. The programmer, of all, should "know", and one thing seems >>clear: Those 3 engines doesn't work optimally with the same hashsize. They work >>at pretty different speeds. Ie. Junior is the fastest seacher and therefore fill >>the hash quicker than the two others. >> >>To set an average for all for a tournament, like Kurt does, is reasonable, and >>64, 128, 256 or something different probably doesn't make much difference, >>except that some engines might benefit a little from one size, while others >>might suffer a bit. >> >>Henrik > Hi, Henrik. I certainly would not be in a position to know. Hash settings have >been one of the big arguments over the years. I was just passing on what i was >told. Regards, George. Tho i tend to agree that as long as u dont set it too >high, the gain by increasing it would seem minimal to me. It has been argued by >some who u think would know that if u have it, use it. Ive heard so many >conflicting thoughts on this subject that i have about given up trying to figure >out the right approach. But your thoughts do make sense to me. Regards again, >George Hi George, Shortly: Yes, there's many conflicting statements and oppinions. The best to do, is to test and test... Like Kurt did, but unfortunately lost hes data :-( Regards Henrik
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.