"Why speed isn't everything"

icancam · Post by **icancam** » Thu May 20, 2004 5:41 pm

Here's an article from The Economist (understandable by a non-technical person such as myself) that discusses the future direction of computer chip technology.
http://www.economist.com/science/tq/Pri ... ID=2477063

Yomat · Post by **Yomat** » Thu May 20, 2004 9:58 pm

"Dr Pinfold says one solution his team is exploring is to use multiple cores, switching from one to another not to increase speed, but rather to minimise the total heat generated. When one core gets too hot, it is switched off for a while to cool down, while another core takes over."

Post by **Rusty075** » Fri May 21, 2004 4:24 pm

Switching from one core to another doesn't reduce the heat generated, it simply spreads it over a greater area. Kinda makes me wonder about Dr Pinfold's grasp of the problems involved

icancam · Post by **icancam** » Fri May 21, 2004 4:47 pm

Might it have more to do with exceeding a temperature threshold (per core) rather than the issue of the total amount of heat that would need to be removed from the micro processing system?

HammerSandwich · Post by **HammerSandwich** » Fri May 21, 2004 6:17 pm

Rusty075 wrote:Switching from one core to another doesn't reduce the heat generated, it simply spreads it over a greater area.

Doesn't resistance go up with temperature?

trodas · Post by **trodas** » Sat May 22, 2004 7:58 am

HammerSandwich - yes, resistance (when talking abour electric flow) increased in conductors with temperature

However resistance decreases with increasing temperature for SEMICONDUCTORS

And that is not all - thermal resistance decreased with increased temperature too

...after all, Dr Pinfold just need to learn that better cooling is allways better solution that chip with 4 cores - not to mention the 4 cores will be kinda expensive

Let's use the money on nice cooper heatpipe cooler rather