Author: William Kerr
Date: 09:19:34 08/04/99
Go up one level in this thread
On August 04, 1999 at 02:46:39, blass uri wrote: >I ask this question because I saw cases when Hiarcs did stupid blunders when my >Hiarcs could not reproduce the moves. > >My question is what is the probability that a program is going to do what it is >told to do? > >I know that one claim about proofs in mathematics by a computer programs is that >they are probabilistic proof because it is possible that one bit in the computer >was moved and the result is false. > >Of course the probability that humans did a mistake in the proof is bigger and >my question is what is the probability that a computer is going to do what it is >told to do. > >If the probability is 0.99999999999 for one machine instruction then it is >dangerous but if the probability is >0.9999999999999999999999999999999999999999999999999999999 >then practically there is no problem. > >The reply to my question may be machine dependent. > >Uri I suspect that cpu temperature can affect the probability of a processor performing an instruction(s) correctly. The hotter the cpu runs due to summer weather, poor cooling or over-clocking, can increase the probability that the cpu will start to produce incorrect results. If the cpu is near its temperature threshold where it will mis-behave, and the cpu is now required to perform many floating instructions or complex integer instructions, the cpu temperature may rise over the safe operating threshold and could start to give incorrect results. I believe Intel Pentium processors have a heat sensor to warn the operator or shut down the processor if too hot (the sensor may be part of the mother board instead). Anything is possible. No matter how improbable, given time and many attemps, it may just happen.
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.