I7 4790K: Temperatures??

DutchMike

Distinguished
Oct 21, 2013
298
1
18,795
Howdy great peeps,

Before changing anything in the BIOS, at stress test it got 59c maximum.
With all power saving options etc disabled it becomes about 72c - NO overclock yet..

Are these temperatures safe? I intended to do overclocking but it seems out of the picture.
However, what ARE safe temperatures for the I7 4790K?

I read that Intel says 72c, on the forums some say 65c, & others say 80c ..

Hoping for a quick response,
Mike

CPU Cooler:
DeepCool Neptwin
 
Solution
The maximum junction temperature on Ivy Bridge was 105C but Intel reduced that a bit for Haswell, don't remember if it was 98C ot 102C but if you are going to overclock, you obviously do not want to get anywhere near that.

80C is a generally safe core temperature but at the upper end of what most overclockers feel comfortable with.

Less than 70C core temperature under load is a safe area many OCers like to be in - very low likelihood of causing any damage to the chip.

The 72C you read is the max package (IHS) temperature.

intel specs are 72 degrees at the ihs the cores can go to about 90 before throttling
you can leave power saving on it makes no difference my i5 4670k is at 4.6ghz with 1.248v with all the c-states enabled
make sure your voltage is fixed and not adaptive/dynamic or the board will give the cpu too much voltage and therefore overheat
 


Howdy,

To what temperatures can I overclock?
Since I know that Prime95 tests it 100%, whilst gaming any game - it will usually be about 5c-10c less?
 
Give this a read:http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

Tcase refers to the temperature on the IHS (NOT the core temp!!!). Anything above 72c will either throttle or shut down your system.
 


Howdy,

Should I put my Voltage at according to what HWmonitor says it uses at 100% stress test?
 
The maximum junction temperature on Ivy Bridge was 105C but Intel reduced that a bit for Haswell, don't remember if it was 98C ot 102C but if you are going to overclock, you obviously do not want to get anywhere near that.

80C is a generally safe core temperature but at the upper end of what most overclockers feel comfortable with.

Less than 70C core temperature under load is a safe area many OCers like to be in - very low likelihood of causing any damage to the chip.

The 72C you read is the max package (IHS) temperature.
 
Solution


Howdy,

Well I see a package temperature, the temperatures of the cores & at motherboard section "CPUTIN"?

So for which one should I be looking out not to exceed 72c-80c?
 

wrong my 4670 used to frequently go to 89 degrees when I was on the stock cooler and never failed
 

89C junction temperature is fine. 89C IHS temperature on the other hand is not since unless the CPU is idle, the junction temperature will be well over 105C.

The CPU does not have an actual sensor for package temperature. That value is estimated based on core temperatures, VRM load and the CPU's thermal model. In other words: it cannot be trusted.
 


Here is a picture of HWMonitor & Core Temp whilst running Prime95 test:

http://i.imgur.com/83OcCTz.png
(excuse the low resolution, lol)

So what temperature should I be looking for to keep it down under 70c-80c?
 


So the current temperatures are already maxed out? :??:
Overclocking is out of the question?

And I must keep the cores under 72c?

Thanks for your great help,
Mike
 

You must keep the cores well under 100C (Tj max)

Every degree cooler increases your chances of achieving a higher stable OC. This is why extreme overclockers use things like phase-change cooling, dry ice or even liquid nitrogen.
 


I will be sure to do a full stress test :)

But you said to keep IHS under 72c, which of the temperatures is it on this picture?
http://i.imgur.com/83OcCTz.png

Or should I keep all temperatures listed there under a certain temperature?
Since I already have 72c whilst not even overclocking..
(My old FX-6300 overclocked to 4.4Ghz did not exceed 61c)

This should be my final question,
thanks again!
 

None of them since there is no actual sensor for it.

If you wanted to monitor Tcase for real instead of using VRM monitoring and thermal models to ESTIMATE it, you would need to drill a small hole through the heatsink or a small groove under it to install a thermocouple or thermistor under the heatsink in direct contact with the middle of the IHS.
 


So there should be some of estimate on how much celsius it is?
I take it then it IS a lower temperature then listed on the picture?

Mike :)
 

Tcase will always be cooler than Tjunction since Tcase is located between the CPU die (hottest - where all the heat is generated) and the heatsink. (coolest - almost room temperature by the time heat reaches the heatsink or radiator's fin stack.)

For every thermal resistance you add in the path from ambient to Tj, you add P * Rth to ambient temperature where P is the power being dissipated by the device being cooled and Rth is the thermal resistance of a given piece of the thermal path. Between the CPU die and air, you have the die itself (the die is mounted face-down on the LGA substrate so heat has to go through the die to reach the IHS side), then there is the thermal interface material between the die and IHS, the IHS itself, the thermal compound between the IHS and HSF, and then the HSF itself.

Since the thermal spec allows Tj to reach 100C while Tcase is only allowed up to 72C at 84W, we can calculate that the worst-case Rth(j-case) is 28C / 84W = 0.33C/W. So if you work backwards from there starting from Tj=65C (average across all cores) while the VRM reports a 60W load, you get Tcase = Tj - Rth * P = 65 - 0.33 * 60 = 45C.

Keep in mind this is just an estimation based on worst-case power with the smallest allowable gap between Tcase and Tjmax. Real chips should rarely be anywhere near that bad, which means Tcase will be at least a few degrees closer to Tj.
 


Do you even know what adaptive means??? You are actually misleading. Adaptive is the safest way to overclock. It means it uses up to the given voltage and not more.

Fixed/Manually voltage is more dangerous because it gives the voltage constantly.

Please read

 


In a perfect world, you would be right. But unfortunately, the "adaptive" systems are not yet advanced enough to make sure the cpu is running at the lowest possible voltage it can. If you have done any overclocking , you would know that leaving your voltage on adaptive or dynamic will almost always give your cpu more voltage then it needs to be stable, thus giving you higher temps.

However, you are right in that it is safe, but you can only achieve small overclocks with high temps.
 

The problem with adaptive is that offset calculations are based on a CPU running at stock voltage and stock clock. Once you start overclocking and overvolting, that goes out the window as a reliable power optimization technique.
 

Wrong adaptive let's the board detect how much voltage the cpu needs and almost always overdoses it with volts on adaptive my mobo will put voltage up to 1.3 when with tweaking it is fully stable at 1.22 I'd rather have 1.22 fixed than let it throw it up to 1.3 and risk frying it

 
I am running mine with 4.8 ghz and after an good houer of stress testing it gets around 80-85c.
I am watercooling it but if havent got a good cooler leave it. 4ghz ins plenty of power!
 

Yes well the thing is, after 0.5Ghz OverClock,
it tends to get a LOT more hotter 😛