Intel CPU max temp question...!

[Pipps]

Hello

I have an Intel E5300.

On the attached page above, it is stated as possessing a thermal specification of 74.1°C.

Does this mean that, so long as I make sure that SpeedFan does not report its temperature exceeding or closely approaching that temperature, then my CPU should continue to operate healthily and happily?

I ask, because currently it is running at around 60°C, with only passive cooling via a Thermalright HR-01 Plus heatsink, in a well ventilated case.

On this basis, is 60°C ok for a continuous temperature reading?

 

[unclewebb]

That 74.1C specification can only be measured by cutting a groove into the top of your heatsink and mounting a thermocouple to the geometric center of your heat spreader. Translation: that spec is a meaningless number for most end users. It was designed to guide system builders with the necessary tools to correctly measure this. Any software, including SpeedFan can not accurately tell you what that temperature is.

For consumers, Intel built sensors into each core. As long as your Distance to TJMax as reported by RealTemp has not counted down to zero and your CPU is not thermal throttling then it is running within the Intel design spec and should live a healthy long life. If an Intel CPU gets too hot, it will automatically slow down to cool and protect itself from damage.

Here's the latest version of RealTemp.

http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

The Thermal Status area monitors for thermal throttling. If it says OK then you don't have to give your temperatures a second look.

 

[TheLaughingMan]

If an Intel CPU gets too hot, it will automatically slow down to cool and protect itself from damage.

That explains some stuff for a friend of mine. Thanks.

 

[unclewebb]

Once an Intel CPU reaches the thermal throttling point even for a millisecond, RealTemp will report that as LOG in the thermal status area which means that at least one thermal throttling episode has been logged. It is very rare to see that because during normal use, it rarely happens.

Even when doing some crazy overclocking, it still rarely happens.

On a CPU like my E8400, when thermal throttling starts, it will rapidly cycle the multiplier hundreds of times a second between 9.0 and 6.0. As soon as the CPU cools down a degree, it immediately goes back up to the higher value. I've run my E8400 for several hours and watched both cores sit there mostly at 98C. I went and watched TV for a while expecting it to blow up or for something bad to happen but nothing exciting happened. It just kept running Prime95 100% reliably with zero errors. After that test, I learned that most people are way too concerned about core temperatures. These Intel CPUs do a great job of looking after themselves.

You can start up RealTemp at any time and it will check the thermal throttling bit that is stored within the CPU. Even if RealTemp wasn't previously running, it will still let you know if any thermal throttling has occurred since power up. I like that feature.

If the word LOG changes to HOT, that is a sign that thermal throttling is in progress. When the CPU can't control its temperature, it will keep the multiplier mostly at 6.0 and it will also try to reduce the core voltage if the motherboard lets it to try and control temperatures. These things work incredibly well even when pushed way beyond sanity.

Lots of people claim to be Prime stable but not many people are Prime stable with the CPU fan unplugged.
Now that's Prime stable.

 

[Black Panther]

not many people are Prime stable with the CPU fan unplugged.
Now that's Prime stable.

So that's how you got it to 98 deg C?
(I was wondering, I got my own E8400 @ 4.2Ghz on air but it still never goes over 60 on Prime95.)

 

[RejZoR]

Anyone dares to sacrifice a Core 2 Duo CPU for an endurance high heat run till it dies, just to see how long that would take. We all know too much heat is not good. But just how too much for how long? That would be fun to observe

 

[Black Panther]

Perhaps Unclewebb's ready to take the test? With CPU fan unplugged?
Me, I wouldn't with my own E8400... (it's not that I don't have faith, just that I am not rich enough to sacrifice it esp since I'm planning other hardware to buy)

 

[Mussels]

On this basis, is 60°C ok for a continuous temperature reading?

depends if its idle or load. if thats what you get sitting at the desktop, then no - when you load it, its going to get hot and throttle. put a fan on there.


Dont forget that whil the CPU may have a safety mechanism, the VRM's and capacitors near the CPU socket do not - and they're designed to be cooled by the airflow from the stock CPU cooler. you're likely cooking those.

 

[unclewebb]

I'm still using the above E8400 many months later. It runs stable at 4000 MHz. It's an early C0 stepping. It has always needed more core voltage than the newer E0 stepping CPUs. I've been running it mostly at 1.40 volts and I've run it without a heatsink up to about 110C or so and have never had a problem with it. I've pushed this to the edge harder than most people do. If I can't kill it, most normal users never will.

After 3 hours of torture at 98C, I was afraid that the CPU was going to burn its way through the motherboard.
You're definitely right that the VRM needs some airflow and protection.

 

[sneekypeet]

Ok Im going out on a limb here....

Essentially if I was to water cool a mobo and get alot of cooling with fans to the back side of the mobo, I could essentially run an Intel indefinately, say crunching, passively? of course leaving throttling and Cstates all on.

 

[Mussels]

Ok Im going out on a limb here....

Esentially if I was to water cool a mobo and get alot of cooling with fans to the back side of the mobo, I could essentially run an Intel indefinately, say crunching, passively?

yeah. but the speed would drop as it overheated and throttled.

 

[sneekypeet]

yeah. but the speed would drop as it overheated and throttled.

understood, and realized. I'm almost willing to run one at say 98*C for a while as an experiment. Just not too sure I wanna spring for the blocks. I do have "extras" that I wouldnt miss

 

[Pipps]

For consumers, Intel built sensors into each core. As long as your Distance to TJMax as reported by RealTemp has not counted down to zero and your CPU is not thermal throttling then it is running within the Intel design spec and should live a healthy long life. If an Intel CPU gets too hot, it will automatically slow down to cool and protect itself from damage.

Here's the latest version of RealTemp.

http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

The Thermal Status area monitors for thermal throttling. If it says OK then you don't have to give your temperatures a second look.

Dear Sir

Thank you for your incredibly helpful advice. I would've had to pay money for something so insightful and detailed!

Thank you also for explaining that this will also apply where a CPU is overclocked. I like to run the E5300 (2.6GHz) at 3.4GHz, given the choice.

I now look forward to using RealTemp as the superior software for monitoring my CPU temps. Thank you for introducing me to it.

Much obliged

Pipps

 

[unclewebb]

You're most welcome.

I've done a considerable amount of testing using my own parts. You should be quite safe at 3.4 GHz.

 

[Pipps]

Sir! Thank you for confirming that my 3.4GHz overclock would still be fine. Much appreciated!

And one further question, if I may...

You and Mr Mussels mention the associated risk of overheating to motherboard VRMs and capacitors located near the CPU.

I have an Asus P5Q Pro motherboard, with a a Thermalright HR-01 Plus heatsink heatsink.

Would you consider that, under the circumstances, I should hold any concerns as to the risk of heat damage to the wider motherboard?

 

[unclewebb]

I don't think Thermalright would still be in business if their heatsinks were damaging people's motherboards. People trying to set new world records at 5+ GHz would be well advised to put a fan on those components to be on the safe side but for everyday users like yourself, there's no need to be concerned.

Most users can use the default core voltage, or very close to that, to run a 45nm CPU reliably at 3.4 GHz. This might sound like a big overclock but for your CPU, it's more like a minimum overclock.

There is a huge amount of unused headroom built into these CPUs. There has never been any need for Intel to push this design really hard because AMD has been playing catch up ever since the Core 2 was first introduced. My first E6400 could be reliably overclocked by over 60% and it still runs just as good today. You have chosen a very safe overclocking level. I don't see any reason to worry about this.

 

[[I.R.A]_FBi]

My e6400 went to 3.2 ghz on stock voltage easilywith not much extra heat

 

[unclewebb]

My E6400 B2 stepping was very similar. It wasn't until I pushed it past there up to 3600 MHz and a little beyond that it started to get voltage hungry.

 

[Pipps]

Unclewebb, that is brilliant! Thank you!