This was a great thread. I know its ancient now, but I think it was worth resurrecting.
The info from RGE was excellent, but it seems most ignored it (even while giving him great kudos). This must be an example of the "metric system" attitude he is talking about.
How can people possibly say they are going to choose some arbitrary value to treat as "max" when they dont know the accuracy of the reading itself?
TJMax is unpublished. That is a fact. There is no reason to assume that the published 100C for MOBILE procs is relevant to desktop procs and, in fact, a good bit of reason to assume it actually isnt.
TDS reports distance to TJMax and nothing more. If TJMax is a true unknown, you CANT say "Ill NEVER go above 70C!!!" because, without a Fluke, you have NO IDEA what the real temp of your proc is.
The only realistic measure is to say, "Ill never allow TDS to be greater than X". But to me, if TDS=0 is where throttling occurs, than having some mental need to be 20, 30, or even MORE away from TJMax is a kind of a sickness. LOL... No offense.
I mean really... Above someone said that "the temperature readings we get at home are a fair estimate" Actually, they arent. Not even remotely. That was RGE's entire point. They are a CALCULATED NUMBER based on pure assumption.
Intels "spec" is merely for TCase Max and is a guideline, again, as RGE said and even then, there are "official" reports of multiple versions of that truth based on rev, stepping, or whatever. "TCase Max" isnt even an absolute. You'd need to really check exactly what the value is for your specific part IF you cared, but I dont see how its relevant at all really.
For annecdotal evidence, Ive had an OC'd QX6700 running at as close as TDS=9 or 10 nearly 24x7 since it was released and everything is absolutely fine, so most of you guys shooting for TDS=30 or 40 or more (soon you'll need to put it in a fridge as TDS will be greater than the actual operating temp, I guess) will be all set with procs that last probably until after you're long gone
The info from RGE was excellent, but it seems most ignored it (even while giving him great kudos). This must be an example of the "metric system" attitude he is talking about.
How can people possibly say they are going to choose some arbitrary value to treat as "max" when they dont know the accuracy of the reading itself?
TJMax is unpublished. That is a fact. There is no reason to assume that the published 100C for MOBILE procs is relevant to desktop procs and, in fact, a good bit of reason to assume it actually isnt.
TDS reports distance to TJMax and nothing more. If TJMax is a true unknown, you CANT say "Ill NEVER go above 70C!!!" because, without a Fluke, you have NO IDEA what the real temp of your proc is.
The only realistic measure is to say, "Ill never allow TDS to be greater than X". But to me, if TDS=0 is where throttling occurs, than having some mental need to be 20, 30, or even MORE away from TJMax is a kind of a sickness. LOL... No offense.
I mean really... Above someone said that "the temperature readings we get at home are a fair estimate" Actually, they arent. Not even remotely. That was RGE's entire point. They are a CALCULATED NUMBER based on pure assumption.
Intels "spec" is merely for TCase Max and is a guideline, again, as RGE said and even then, there are "official" reports of multiple versions of that truth based on rev, stepping, or whatever. "TCase Max" isnt even an absolute. You'd need to really check exactly what the value is for your specific part IF you cared, but I dont see how its relevant at all really.
For annecdotal evidence, Ive had an OC'd QX6700 running at as close as TDS=9 or 10 nearly 24x7 since it was released and everything is absolutely fine, so most of you guys shooting for TDS=30 or 40 or more (soon you'll need to put it in a fridge as TDS will be greater than the actual operating temp, I guess) will be all set with procs that last probably until after you're long gone