Ambient Temp & Core Temp

[mrfatboy]

It's really hot here today. Its 88F (32c) in the house. My cores (see sig) are running at 98F (37C)idle. I have a tuniq tower and I have a couple of questions:

1) How are these temp under these conditions?
2) In general how much is the tuniq supposed to cool under Ambient temp?
3) is there a ambient/cpu temp chart somewhere?
4) what's the temperature danger zone for the 6750?

thanks

 

[mrfatboy]

bump

 

[suszterpatt]

I'm out of the loop on the newest CPU's, but I can tell you that 5 degrees above ambient temperature is pretty good.

The temp limit on processors is usually around the 60 °C mark, and even then it's unlikely to do immediate damage to the core. It will reduce its longevity, but you'd need to go somewhere around the 70-80 °C mark to do damage in the short run.

But like I said, I'm not sure how all this applies to the core 2 duo's specifically.

 

[BonzaiDuck]

On the matter of ambient temperatures:

With a 1C-degree rise in ambient room temperature, there should be a corresponding rise of 1C degree in the motherboard idle and load temperatures, the CPU idle and load temperatures. Except for the fact that CPU and other coolers are designed to operate optimally over a range of temperatures so that they may depart slightly from this linear relationship with ambient, you can assume this 1-to-1 linearity is operative.

With a Core-2-Duo, there are three temperature sensors; with a Core-2-Quad -- there are five.

Intel explains that "TCase" is a sensor (and reported temperature) that resides between the cores, and effectively measures the temperature at the CPU heatspreader. There are "TJunction" sensors embedded in each core.

For the Core-2-Duo, there should be roughly a 15C difference between TCase and the average of the two core temperatures. I've observed about half that difference between the TCase temperature and the core values in the C2Q processor. That is, the TCase temperature should be always lower according to these guidelines than the individual core temperatures.

Please check the Intel web site for the thermal specifications (Chapter 5 of the specification and specification update) for your E6750. I THINK that the throttling temperature is about 65C for TCase, since that is what it is for an E6700. This means that the cores would approximately average about 80C before the CPU needs to throttle, or where it otherwise reaches its thermal limit.

Any incremental improvement in CPU temperatures, and of course any drastic improvement, will increase longevity and reliability. Further, any reduction in the idle-to-load temperature differential will improve stability. It could probably be said that in a rather cool environment, frequent power-up and power-down would decrease longevity because the CPU is subjected to greater temperature change.

To some extent, lowering CPU temperatures will improve performance, and may provide operational stability at lower VCORE voltages than at a higher temperature. The "thermal limit" of the processor is defined according to what happens to the metal conductors linked to the processor as opposed to the semiconductor itself. With increasing temperature, the semiconductor's conductivity actually increases, while the conductivity of the metal components decreases, and at the threshold temperature, there is a greater danger that the metal-to-silicone contacts will weaken or even separate.

The great risk of assuming that cooling the processor down to -- or below -- room temperature provides carte blanche to increase the VCORE for over-clocking, is that the effective cooling may eliminate temperature increases from over-volting, but over-volting itself may eventually damage or degrade the processor. So there may be a false sense of security about over-volting the processor if it is possible to keep temperatures low.

So I advocate over-volting below 10% in excess of the manufacturer retail box "maximum." If for instance the processor is being over-volted by 5% of that spec under air-cooling, and I were to change to water-cooling, I'd first want to see how I might drop the VCORE without changing the stability of the existing over-clock setting. Once I reached a minimum while keeping the over-clocked system stable, I would then raise the over-clock setting and sparingly increase the VCORE back to the original self-imposed air-cooling limit.

Choosing how much to over-volt the processor is a gamble based on the reasonable assumption that the manufacturer (INtel) chose the retail-box "maximum" voltage with hopes that they can completely eliminate RMA returns under warranty. So if the load voltage exceeds this limit by a few hundredths of a volt, there is still a low risk of damage to the processor, even if you have invalidated your warranty by either over-clocking or over-volting or both.

And obviously, increases in voltage mean a parabolic increase in temperature. But whether you choose to over-clock or run at stock, operating the computer at 110F-in-the-shade temperatures may encounter the limits I've discussed -- no matter how low the stable VCORE setting may be.

I suspect that the DOD has a rigorous replacement program in effect for laptops used in Middle-Eastern or Afghani battlefields during the warm months.

 

[imported_rod]

32C is "really hot"???

1) 37C is fine. Anything below 60C (under load) is fine.
2) You cannot cool your CPU below ambient with air/water - you would require a refrigerated/peltier system.
3) Huh?
4) ^^^See first answer.

 

[BonzaiDuck]

That's what I also said, in a manner of speaking, but I'd start getting worried if ambients were approaching 43C / 110F!! Any more than that for any length of time, and softy that I may be -- I might beg for death.