Quote:
|
Originally Posted by pHaestus
RoboTech: C/W much lower on your die sim than I see on my test CPUs; perhaps die insulation isn't sufficient? That'd be consistent with attached graph too...
|
Yes, the C/W values will be much different because my new sim has a larger die surface area...
I went with this as I think it is more representative of current processors and what we will see in the future. The larger surface area is also more robust so should hold up better after lots of use.
"The thermal die consists of a copper block with a raised die area of 1,000 mm^2. (This is the approximate size of a P4 IHS - smaller than an Athlon 64 IHS but much larger than the exposed Athlon XP core.) Two Watlow cartridge heaters generate up to 300 watts of heat at 32 VDC. Waterblocks and heatsinks are mounted to a rigid phenolic top via threaded metal inserts arranged in Athlon XP, Athlon 64 and Pentium 4 mounting patterns. The die is well insulated with Cotronics ceramic fiber insulation."
Actaully the insulation is very good. "The thermal die simulator dissipates 1 watt for every 12.5ºC rise in temperature over ambient. At 100 watts input power this means that 2.2 W is being lost to the surrounding air and that 97.8 W is actually being absorbed by the waterblock." This statement is based on data from thermal loss calibration curves I ran on the sim.