Cpu Temp 15-20 Degrees Lower Than Cores, Is It Normal?

As the topic says my overall CPU temp is much lower than what the cores are, i'm wondering is this normal or something that i should worry about.

Temps
Overnight Idle Temps - This is when they actually match, they're both around 23-26
Light Use (Browsing Web/Watching A Movie/Texted Based Game) - CPU - Around 29-33ish Cores - 30-50ish
Moderate Gaming (Older Games) - CPU 34-40 Cores - 56-70ish
Heavy Gaming (Newer/CPU Heavy games like Attila/Rome II Total War) - CPU - 40-46 Cores - 60-84
Load Temps (Stress Testing) - CPU - 56 (rarely ever goes much higher) Cores - 90 -100

In the pic i'm running speedfan, realtemp and HW Monitor with the CPU load 43% (attaching it made it too small to read uploading elsewhere asap)

Any help is much appreciated

 

here's the picture

 

The problem is, different monitors use difference sensors. According to the Intel Ark for the 3570K, the maximum Tcase temp allowed is 67.4°C. But cores can get 10 - 15° higher than that with no problem - as long as they don't sit at those temps for hours on end. So the ones that bother me are your temps in the high 70s and above. I would not be panicky - assuming my system is still stable at those temps and I am not experiencing any sudden reboots, shutdowns, or system freezes. But those high temps need to come down a bit.

If me, I would first make sure the interior is clean of a heat-trapping dust. If clean and temps are still up there, you need to back off the overclocking. Try 4.0GHz, or even that CPU's turbo speed of 3.8GHz at least until your cooling issue is resolved. 4.2GHz is pushing your processor beyond what your case cooling can support at this point. Note I said, "case" cooling. It is the case's responsibility to provide a sufficient supply of cool air flowing through the case. The CPU cooler need only toss the CPU's heat up into that flow. If throttled down for noise, you may need to turn up the speed of your case fans, or add another or larger fans to get the necessary front-to-back flow through your case.

Contrary to what some may tell you, you NEVER EVER need to replace the TIM (thermal interface material) just because it has been a few years since you last did. The ONLY time you must replace the TIM is if the bond between the heatsink and CPU die is broken. Dried and crusty TIM around the edges means nothing. And even if the TIM between the mating surfaces dries, that is not a problem either. Those liquids are only there to make the application of fresh TIM smooth and easy. The solids that remain behind are still occupying the microscopic pits and valleys in the mating surfaces, preventing insulating air from getting inside. Remember, the best transfer of heat occurs with direct metal to metal contact. Any extra is just in the way.

Note too that, contrary to what some may believe, both Intel and AMD OEM coolers are quite capable of providing sufficient cooling, even with mild to moderate overclocking in a properly cooled case. If your case is clean and your case is providing adequate air flow, then you may need to consider alternative cooling solutions.

 

Note that Digerati's post refers to Intel Ark's Tcase - the temperature of the outside of the CPU, while in his second post the 'case' refers to cabinet in which the whole computer sits.

If you can stand another temp tool, download from the Intel website the Intel Processor Diagnostic Tool, which will tell you if your CPU is functioning ok, including a temperature check

Dumb_Question
6.June.2016

 

Huh?
What are you talking about? I only have one post above!

If you mean when I said "make sure the interior is clean...", then yes, I meant the case interior. I assume folks understand I don't mean to "de-lid" the processor.

The problem with the Intel tool is it really is not very informative for most user. For example, it does not tell you the actual temps.

Temperature Test
Module Version: 1.0.12.64b.W
Start Time: Mon Jun 06 10:26:46 2016
Test Result - PASS
Expected: Greater than 1 degrees below max
Received: 74 degrees below max
End Time: Mon Jun 06 10:26:46 2016
Total Time: seconds: 0​

It is great that it passed, but does not really say much.

 

I meant 2nd para, but post was what came out !
I agree that the o/p from the processor diag tool doesn't seem very informative...Received 74C below...
CoreTemp tells you stuff about (only the cores of the CPU) but it does at least give the actual temp and the max allowable temp of each core.
I thought there was only one source for every temperature on the m/b, and all the software used whichever source that was, eg the sensing diode in core 1. HWM measures the temps when it is invoked and does not change them until it is run again (?), speedfan updates continuously I think at a user set frequency ?), dunno about realtemp. Some of the temps given by speedfan are difficult to interpret, I agree, but by comparing the readings with other temperature monitors, some can be worked out. I haven't yet worked out how to use it as a non-interfering measuring tool though.

Dumb_Question
6.June.2016

 

Not maximum "allowable". It tells you the current temp of each core and the maximum temps each core reached during that monitoring session. It would be nice if it told use the maximum allowed but that would require the developers build and maintain a very large database of CPUs. And that would take considerable time and effort (read: $$$).

That's part of the problem. There are no industry standards for any of this. Motherboard and CPU makers can put sensors wherever they want. And they can label them anything they want too. This is why you cannot really compare AMD to Intel.

Speedfan has always had problems putting the correct label to sensor. Fortunately, you can edit the labels once you sort out which is which. I use Speccy mostly for that. HWiNFO64 is also excellent for properly matching labels to sensor but the amount of information HWiNFO64 provides can overwhelming.

Not sure what you mean there. I have a laser guided IR thermometer gun that I often use, but you cannot use them to measure inside processors. And since heatsinks typically cover the processors completely, you cannot even use them to measure the CPU case.

 

CoreTemp has a box labeled Tjmax - does anyone know the (meaning of the) number in that box ?

Sorry, I meant there was only one source for each temperature reading on a given m/b, so all software gives a result from that source for that particular m/b. This fact enables you correlate readings from each item of software.

Yes I use speccy for temp readings sometimes too, but it does not give rt readings, it has to be run again IIRC

I mean simply that I do not know how turn speedfan's speed control of the a fan off. I use it to make a fan work when I consider a temperature a temperature has risen too much. I do not know how to use it as a temp monitor only, leaving the o/s or whatever to control the fans.

sounds useful...and expensive ?

Dumb_Question
6.June.2016

 

Oh, I see what you are saying. Sorry. You are right.

That stands for the maximum junction temp as taken by the internal sensor in the processor (for Intels anyway). But the data sheets don't really define what/where that junction is. All I know is it relates to the temp at that point when the CPU is at the maximum thermal design power (TDP). And the TDP is the power dissipation and junction temperature operating condition limit, as specified in the data sheet for that specific Intel CPU.

The data sheet also says, "The maximum operating temperature implies maximum junction temperature." But again, there is nothing that specifically defines what or where that junction is. Some folks assume it is the core temp - but I note different cores have different temps at any given moment.

In general electronics, the junction temperature is the operating temp of the actual semiconductor in a device. But in a processor, there are 100s of billions of semiconductors in each CPU die. And temperatures across a single die vary all the time. So without knowing where the sensor is, it remains vague.

Clear as mud, huh? And AMD has their own language and way of measuring too.

I do think it is safe to say when the TJMax is reached, that will trigger the CPU's self-protection processes - the CPU will begin to throttle back in speed before burning up.

Depends on how much you want to spend. There are many to choose from. I use mine for all sorts of thing - checking the grill and frying pan, the grandkid's forehead (avoid eyes), dog's nose, and more.