|
Many OS hats...
|
20/04/2007
|
| |
Multicore processing is very much the ‘in’ thing and many say it’s the future of processing in general. It would be hard to dispute this in the pc world, but is the same true for embedded devices?
The primary reason for migrating towards multicore processing is physics; greater transistor density and switching speeds mean a commensurate increase in heat. Getting rid of that heat is governed by the laws of thermodynamics and there comes a point when heat dissipation methods will no longer keep up. At that point, the silicon substrate would be compromised.
That ‘point of no return’, arguably, arrived a little while ago. Intel drew a strategic line in the sand; the heating effect of transistors had to be capped. Naturally, processor throughput had to continue to increase, so the challenge was how to use slower transistors without losing performance. The solution was multicore processors.
Of course, physics isn’t the only reason for multicore processors. Neither is having two or more processors working alongside each other new – multiprocessor topologies date back to the first mainframes. Then, the reasons for doing so were pretty much the same as today; to maximise performance.
|
| |
Author
Philip Ling
|
| |
| |
|
This material is protected by Findlay Publications copyright 2008.
See Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the sales team.
|
| |
|
| Email this article |
| |
|
|
|
|