i3-4170 80C within a few seconds of Prime95 :/

[Ollie Dixon]

Just built a new PC. The i3 is on the stock cooler. I just stop the test when it hits 80C

http://i.imgur.com/kVyWpiV.jpg

I wasn't expecting great cooling but this is the highest I've ever seen a CPU go.

Is this normal? Cooler seems to be really snug..

 

[CompuTronix]

The reasons that most readers find processor temperatures so confusing are;


(1) not recognizing the vast number of variables in hardware, environments, software and test methods which influence temperatures and measurements, and ...

(2) the glaring lack of clarity within Intel's Thermal Specifications, especially the disconnected relationship between Tcase (CPU temperature) and Tj Max (Core temperature).


Most owners don't know that Tcase is CPU temperature, NOT Core temperature, and that there's a 5C difference which Intel doesn't seem to want anyone to know about, apparently because it provides them with an extra 5C thermal cushion for warranty. Most users don't realize that when Core temperature reaches 100C (TjMax), CPU temperature is right behind it at 95C (Tcase)! Since most users monitor Core temperature and not CPU temperature, I think the terminology by which Intel expresses processor temperatures in their spec's is obsolete.

Unfortunately, Intel currently regards these two specifications (Tcase and TjMax) as separate and discrete entities; one for desktop processors (with Integrated Heat Spreader), and the other for mobile processors (without Integrated Heat Spreader) ... which is confusing for the end user, when we know full well that a defined and documented relationship exists.

Further, I think the fact that the 5C difference between Core temperature and CPU temperature can only be found in a few engineering documents, and is not clearly written into a single, understandable specification, is a HUGE failing. Add to this when you chat online with Intel and ask them about the relationship between Tcase and TjMax, their representatives rarely provide a straight forward and informative answer!

Also, Intel is inconsistent with the use of their own terminology! For example, the Thermal Specifications show "Tcase", but in many White Papers and other engineering documents, Tcase is correctly expressed as "TcaseMax". Intel's definition of Tcase is; "Case Temperature is the maximum temperature allowed at the processor Integrated Heat Spreader (IHS)". So as you can see, the term "TcaseMax" is not used in their Thermal Specification, which as an engineer, I find completely inadequate.

If Intel's Thermal Specification was "TcaseMax" instead of "Tcase", as it is shown in other documents, it would give people a completely different impression. Here's the clarity in all the confusion:

Tcase is CPU temperature.
TcaseMax is a specification.

Tjunction is Core temperature.
TjMax is a specification.

As Intel has it, owners can't make heads or tails out of the woefully lacking, vague, ambiguous, contradictory and confusing Thermal Specifications. I find it amazingly disappointing that a giant corporation who manufactures and markets the best processors in this world doesn't do a better job of writing Thermal Specifications that are clear and understandable, which apply equally well across desktop and mobile platforms.

If Intel would commission me to rewrite their Thermal Specifications, I would eliminate "Tcase", "TcaseMax" and all references to "CPU" temperature. These would be replaced by a new term called "TjunctionMax", and "TjMax" would be redefined as "Tthrottle", both of which refer to "Core" temperatures.

The new and simplified Core temperature spec's for the processor in this thread would look like this:

i3 4170

TjunctionMax 77C: The maximum Core temperature allowed at the transistor junctions in the processor die.

Tthrottle 100C: The Core temperature at which the processor will reduce clock speed for overtemp protection.

Very simple, straight forward and understandable for everyone. No more confusion.

CT

 

[photonboy]

What??
"Very simple, straight forward and understandable for everyone. No more confusion."

You preceded that by saying the max core temp allowed is 77degC, but stated that the cores throttle at 100degC.

Again, you are saying that Intel won't throttle until 23degC above the maximum recommended core temperature? That makes no sense. Why have the specification listed, and the ability to throttle if they don't match up?

(For the record, if you read my LAST response I actually did get the terminology correct. Nowhere did I state that Tcase was core temperature.)

*Long story short is that if the Tcase max temperature was supposed to be 72degC, which meant that Tjmax was 77degC AND the CPU actually throttled at 77degC this would all make perfect sense to me. It doesn't throttle that way, so I hardly find this not confusing.

At least mobile throttles based on Tcase which I do understand.

I am doing more research so no need to reply yet if I have explained the situation correctly. I don't like explaining things to others if I'm not fully understanding the situation and I HAVE definitely made mistakes (like when I said RAID1 was no faster than a single drive... which it was when I was taught. )

 

[photonboy]

FYI, I'm reading some stuff so I'll be a while:

Slightly older article but still useful: http://www.intel.com/content/www/us/en/embedded/testing-and-validation/cpu-monitoring-dts-peci-paper.html

Intel: http://www.intel.com/support/processors/sb/CS-033342.htm

 

[LookItsRain]

So, from what i understand, everything that intel list is entirely useless OR
From what ive understood myself here over the years, the tcase max and tjmax are the max safe temps that intel recommends for that specific cpu, but it will not throttle until 100c etc. to prevent outright damage.

Just like how some cars are limited to 98mph, but they can do 120 without an issue but 125 will explode the tires.

 

[photonboy]

CompuTronix has a good link here: http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

Everything I've read takes me back to the following conclusion:

1) SAFE temperature for the i3-4170 is 77degC max for any single core.

2) The CPU throttles at 100degC (hottest core) which is 23degC above Intel's own recommended maximum.

I do NOT get the logic behind this since I would assume throttling in laptops would regularly happen suggesting they are regularly well above the recommended temp (Linus from Tech Tips submerged an Apple laptop in water and got 2X the performance as it didn't throttle... so it is throttling a lot). For desktops it's much less likely. Those who overclock generally understand there is a reduced lifespan and those who don't are much less likely to run too hot.

So again, I think the CONFUSION has to do with Intel not throttling down at the point they themselves recommend not exceeding for a reliable, long operation.

*Do I have this correct now CompuTronix?

 

[CompuTronix]

Yep ... you got it.

Just for fun, let's compare quad core processor temperatures to a four cylinder car with five temperature gauges; one gauge we're all very familiar with, which is the coolant temperature for the entire engine, and the other extra four gauges are for the coolant at each cylinder head.

The cylinders are of course the heat sources, and any good mechanic knows that each cylinder runs at a slightly different temperature, yet each must be a little warmer than the coolant for the entire engine, thus a comparison to CPU temperature and Core temperatures.

Regardless, the manufacturer says that normal on the engine gauge is 195F, but the cylinder gauges max at 260F. Even the wife can guess that it's probably not good for those needles to be pegged, 😀 thus a comparison to Tcase and TjMax. Let's just add a note that the cylinder gauges run 5 degrees higher than the engine gauge, and it all starts to make sense ... except for how Intel writes their spec's.