Advanced Micro Devices Inc (AMD): OC, Alan, you're both wrong. All the...

Reply Private New

Replies (1) Next 10 Prev Next

Send PM Follow Ignore

Followers	29
Posts	25865
Boards Moderated	0
Alias Born	09/11/2002

wbmw

Re: oc detective post# 66908

Thursday, 12/01/2005 3:51:05 AM

Thursday, December 01, 2005 3:51:05 AM

OC, Alan, you're both wrong.

All the tables do is perform a little math for you. Take the column row, subtract the ambient temperature, and then divide the result by the thermal resistance, and you'll get the value in the resulting column.

For example, the first result from profile A, at the row marked 49C degrees:

49 - 42 = 7
7 / 0.34C/W = 20.6 W

All this says is that if you are using the recommended thermal solution with total Tcase to ambient thermal resistance of 0.34C/W, and your product is giving off 49C degrees of heat (7C degrees over ambient), then you can effectively dissipate up to 20.6W of processor power before your processor temperature will begin to rise. It DOES NOT SAY that Opteron will dissipate this much if the core temperature reaches that high. In fact, depending on the application and what kind of code is run, and how much idle time occurs, the processor wattage will exceed this, and the processor temperature will rise.

The graph is helpful to determine how hot your system is going to get, given the thermal resistance of the cooling solution. So let's say you are using thermal profile B (most likely, dual core Opteron) with the system built to recommended thermal resistance of 0.27C/W. Now, you can look up on the table how hot the processor will be if it's dissipating, say, 70W of power. According to Table 9, it will be ~61C degrees (assuming 42C degrees ambient), or more accurately, simply 19C degrees over ambient.