News Focus
News Focus
icon url

alan81

12/02/05 1:31 AM

#67070 RE: oc detective #67068

OK, so here is the task...
Go figure out how to read this value out of the CPU. Test a few hundred of them and let us know what they are. BTW, why do they program this value into the CPU. What is done with it and by whom?
--Alan
icon url

wbmw

12/02/05 2:14 AM

#67083 RE: oc detective #67068

Re: Tcase max is not something adjustable - it is a fixed value for every single cpu. What I am saying is that whilst the actual value can vary from cpu to cpu the fixed value is pre-programmed into the individual cpu at testing in Singapore.

Not quite. Tcase is a specification derived from thermal validation. The only numbers programmed into the CPU are the ones that relate to the temperature at which the CPU is programmed to throttle or shut down. I'm not sure what Opteron does at higher temperatures, but AMD includes Tcase just like Intel as a specification for systems builders. It is a threshold that the cooling solution should be designed not to cross. The number itself is not programmed in during "test and sort", like other fusable values.

Re: So you cant have an increase in the Tcasemax that you describe in your second paragraph as a consequence of increased power dissipation.

If you under-design your cooling mechanism, then you certainly can exceed Tcase,max. When you do, the processor has several options. It can try to throttle or turn off the clocks until it cools down, or in extreme cases, it can trip the catastrophic thermal trigger and shut itself down completely (requiring a reboot to reawaken).