May be a dumb question but...
- Thread starter Alphane
- Start date
You can't just make the chips bigger, there's a maximum viable limit above which the time taken for signals to travel from one side of the chip to the other is significant enough to impact on the maximum frequency of operation.
As for cooling, there was talk of adding bore holes into the die to allow for in-chip water cooling.
As for cooling, there was talk of adding bore holes into the die to allow for in-chip water cooling.
Bigger = Bigger latency = Less efficient for the same amount of work
Adding holes would reduce the surface area for heat transfer unless the CPU cooler had the opposite and had lots of bumps to perfectly fill the holes. Easier to just be flat and whack a big cooler on top I expect (which has a massive surface area from fins usually) and allows for after market solutions much more easily.
Adding holes would reduce the surface area for heat transfer unless the CPU cooler had the opposite and had lots of bumps to perfectly fill the holes. Easier to just be flat and whack a big cooler on top I expect (which has a massive surface area from fins usually) and allows for after market solutions much more easily.
Soldato
- Joined
- 19 Oct 2007
- Posts
- 3,739
- Location
- Delete *.*
Didn't listen to your own sig
No I did it's something I've thought about for a while, our CPU's seem to be the same little blocks of Silicon they where back in the day of the first spectrums, yes 10,000* more complex but essentially using the same architectural principles, now we've got to the point where cooling the electron channels has become more important does this not mean we should reconsider the way we are designing our processors to take advantage of this.