Author: Aaron Gordon
Date: 14:48:40 07/22/05
Go up one level in this thread
On July 21, 2005 at 18:11:12, Mark Ryan wrote: >On July 21, 2005 at 14:34:08, Robert Hyatt wrote: > >>On July 21, 2005 at 14:05:47, Kurt Utzinger wrote: >> >>>On July 21, 2005 at 12:13:27, Gian-Carlo Pascutto wrote: >>> >>>>Do those CPU's really exist? >>>> >>>>I couldn't find any - seems to me like another nice example of "why you should >>>>not overclock" :) >>>> >>>>-- >>>>GCP >>> >>> You are right I think. >>> Kurt >> >> >>I would agree. Hash collisions simply will not cause this. Even if you use 32 >>bit signatures... > >Could the blunder have been caused by an over-heated chip? Uri mentioned in >another thread that Rebel made some bad moves (in one of the "GM Challenge" >games) that were caused by an over-heated chip. > >Mark A lot of the time people do not run the proper tests. They'll run their favorite game/app for an hour and think all is well. In reality, they could have massive memory errors. I've actually seen this happen before. I had a stick of ram start dying on me. My rating on FICS kept dropping and dropping, finally the program itself started to crash. Tested with memtest.. sure enough, ram kicked the bucket. Put in a new stick and all was fine, rating went back up to normal. It is a good idea to verify that your overclocked settings are alright with burnk7, prime95, memtest86+ at 15C over your normal running temps. Once that is done clock it back 5% or so and you'll be fine. As far as overheating goes.. never seen it cause a problem. Usually any overheats cause massive sudden problems. Completely locking up the computer or crashing the program. Generally unstable/bad ram corrupts the hash tables and thats what will do it.
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.