Author: Aaron Gordon
Date: 00:05:46 01/03/05
Go up one level in this thread
I wouldn't trust Tomshardware for much of anything, but he has to bring this to light.. as it is a major issue and not something that can be ignored (even by an Intel funded hardware site). His credability would go in the hole for the small majority that DO read his page. If you buy a Dell or any other pre-built P4-560 system you'll see it throttle down to 2.4-2.8GHz from 3.8GHz. It gets that hot. You need water-cooling or at the very least a high-end (and usually expensive/loud) air-cooling setup to bring the temperature down enough to just run 'normal' speed. Most people won't do this, or even know about the P4 overheating/throttling. Lots of the computers I work on (I do house calls) have the persons computer stuffed under a desk or in a cubby hole where it gets rather warm. They also never, EVER dust them. So once dust starts to clog it'll start to get hotter, and as a result it will throttle back even more. So the person with a "super" P4-560 chip supposedly now ends up having one running 1.8-2.2ghz, just so it doesn't kill itself. Also Prescott chips (compared with the Northwood) don't do as well in terms of MHz for MHz, they're generally slower. Some things can be faster, but MOST things are slower. Chess programs for example can be up to 20% slower. So lets say after 1 year your P4-560 3.8ghz is throttled down to 2GHz.. knock off 20% and we now have a Northwood equivalent chip @ 1.6GHz for some chess programs. Not only this, but electromigration will kill the chip faster. 80C+ sustained over long periods is horrible for modern chips. Here is a bit of information about the subject: http://www.csl.mete.metu.edu.tr/Electromigration/emig.htm Basically electromigration is always happening. Heat just accelerates it. It is like a stalactite and stalagmite in a cave, but in a modern CPU when it finally gets to the other side it'll short and cause circuit failure, which of course will kill the CPU on the spot. I got a formula a while back for figuring out the lifetime of a CPU, and ironically it was from an Intel engineer. Basically it stated for every 10 degrees celsius the CPU is brought down in temperature the processor will last twice as long. If you raise the temperature by the same the reverse is true. I'm just curious as to how long these new blazing hot P4s are going to last, and what is going to happen to Intels reputation if they start dying left and right on people after a year or two. Guess we'll see in a few years (assuming people keep the chips that long and don't upgrade).
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.