Sheesh, could it just be my chip? (Temperature)

[Erick Aguilar]

So I've been having an extremely hard time with my chip and temperatures.
It never overheats, the system has never frozen on me or crashed (Without me being the reason due overclocking). But the way the chip reacts doesn't make any sense to me and it's frustrating me a lot recently.

I am running my A10-5800K at 4.3GHZ on a Cooler Master N520 cooler, which was enough for my previous A6-3670K, and it would never break 65C.

But this thing even with the same TDP breaks the 70C barrier in less than 10 seconds on an OCCT stress test. Or so the sensors say. I have reapplied TIM to make sure everything was okay. And one of the things that worry me is that the heatsink does not feel -FIRM- sure, if I wiggle it a bit after it's bolted down it stays in place, but if I twist it a little, it does budge, making me believe there is not enough pressure between the cooper plate and CPU, which could be the cause of the issue, the bolts are tight, and I dont dare to tight them more as Im afraid the motherboard could crack, but this wasn't an issue before. I am seriously at a loss.

Now the leap I took from 3.8 to 4.3GHZ is fine, it is a stable overclock and it works. But not knowing my temperatures is really frustrating as the hot months are coming, and I can sure you room temperature is going to go up to the 110's soon enough.

When gaming which is the most intensive task I do, I run hwmonitor and after a session I check temperatures, I am overwhelmed by the fact that my max temperature is 72C. Just 2 degrees short of AMD's max temperature recommended for the chip. Bear in mind I have only ran hwmonitor with Tomb Raider 2013, and it -is- kind of a demanding game, but the temperature readings I get are unacceptable. The air being pushed by the heatsink is indeed warm, but it doesn't burn.

I'm just at a loss, I want better cooling, but what's the point if I can't know if my temperatures are good or not?

Any advice is truly helpful, specially if you own the same chip.

Thank you.

 

[kbrady1979]

All chips are not made equal, even the same model chips. Yours might just be at it's limit. Also, I understand that you don't want to crank on the bolts holding the CPU cooler on, but have you tried seeing if they had backed off any? 1/8 of a turn won't hurt anything, and will let you know if they are as tight as they need to be, or have loosened up some.

Also, have you checked to see what your temps are(idle and load) without any overclock?

 

[Britton30]

My cooler will turn a bit to if I try it and it is completely tightened. Have you tried disable the on-chip GPU and run only the GPU card?

 

[Layback Bear]

First could you fill in your system specs completely. It will help us help you. Look at my specs and the other good folks that are helping you.
Next I would put the computer back to default and play the game while monitoring the temps. That will give you a temp base line to work with.
Like kbrady1979 has posted no two chips are equal. We need the temp base line to work with.

Also when you post temps please also give us ambient temp to compare with.
The object of all of this is to get your cpu temps as good as possible under load with no over clock using normal test and methods. At that point the good folks here can help with the over clock.
I don't recommend over clocking anything in a computer before knowing the facts and figures of a stock clock under load.

 

[Erick Aguilar]

All chips are not made equal, even the same model chips. Yours might just be at it's limit. Also, I understand that you don't want to crank on the bolts holding the CPU cooler on, but have you tried seeing if they had backed off any? 1/8 of a turn won't hurt anything, and will let you know if they are as tight as they need to be, or have loosened up some.

Also, have you checked to see what your temps are(idle and load) without any overclock?

Thank you for the reply. At the moment I am getting ready to go work, and I will be able to answer some questions that don't require system tweaking.

And yes, as a matter of fact yesterday I removed the heatsink completely to do some cleaning, and I turned the bolts as far as they could go please a little push, but I don't want to go any tighter or else Im afraid I'll hear a creak.

 

[Erick Aguilar]

My cooler will turn a bit to if I try it and it is completely tightened. Have you tried disable the on-chip GPU and run only the GPU card?

As far as I know, there is not way to -disable- integrated GPU. I guess I could downclock it down to 0? never really thought about it, I would need to check the motherboard manual/BIOS.

 

[Erick Aguilar]

First could you fill in your system specs completely. It will help us help you. Look at my specs and the other good folks that are helping you.
Next I would put the computer back to default and play the game while monitoring the temps. That will give you a temp base line to work with.
Like kbrady1979 has posted no two chips are equal. We need the temp base line to work with.

Also when you post temps please also give us ambient temp to compare with.
The object of all of this is to get your cpu temps as good as possible under load with no over clock using normal test and methods. At that point the good folks here can help with the over clock.
I don't recommend over clocking anything in a computer before knowing the facts and figures of a stock clock under load.

Thank you, ambient temperature today is much cooler than yesterday.

Today's temperature is 73.4
I lowered the clock to stock (3.8hz(, and there was no temperature change.
I then took the clock up to 4.2GHZ and lowered the VCORE one value down to 1.4325 and there was a difference. My CPU took 5 minutes to reach 72C under OCCT, but still. Too much for my liking. And it didn't use to heat up like that before.

As I said, my 4.3GHZ overclock was stable on OCCT for 24 hours without going past 65C, I don't know where those 7C are coming from, possibly ambient temperature.

Perhaps my cooler is just not cutting it for the job anymore?

 

[Erick Aguilar]

..Interesting.
Today I read an article about actual temperature readings with AMD chips.
Basically it was telling how I should focus on Core temperatures instead of CPU temperatures due the location of the thermal sensors. Core temperature being the one inside the processor, and CPU temperature actually being the socket temperature.
Here is the thread:
AMD Temp Information and Guide

It is a little dated, but it does make a lot of sense even today.

After some more testing I took my Voltage a little lower and left the clock at 4.2GHZ, then turned on Core Temp and ran Tomb Raider 2013 for an hour or two. Highest Core temperature registered by core temp peaked at 50C, and CPU temp as dictated by HWmonitor was 70. So there is a big temperature discrepancy between both values, as stated in the URL I previously mentioned.

 

[ICIT2LOL]

Hum I'm interested to know how the HW showed up the CPU temps Erick as I have been playing with Intel processors with it and I have three diode readings.
Diode 1
Internal
Diode 2

Now from what I can gather the 1 & 2 are the diodes for reading / regulating fan speeds for the Tcase temp (where the CPU "lid" meets the heat sink)

The internal diode on my readings is usually a few degrees lower than the 1 & 2 and I think is one that is embedded in the silicon chip itself and related to the Tjunction temp. But this is Intel and AMD might have a different set up and I also stand to be corrected in my statement re above.

This is the reference I used Guide to Understanding Intel Temperatures | TechREACTION

This goes some way to explaining AMD temps ALCPU Forums • View topic - Core Temp - FAQ

 

[Britton30]

Tcase: Technically this refers to the temperature that you can measure using a thermocouple embedded in the centre of the heat spreader, but this is only done in the factory by Intel or by users willing to drill their heat spreaders open to insert a thermocouple. Therefore Intel provided a diode between and below the cores with a reading calibrated by the BIOS which can be used instead. This reading can vary greatly when the BIOS version is changed, but will not necessarily change if the BIOS calibrations were not altered between versions.

This is Tcase, not measurable unless the CPU lid is drilled into.

 

[ICIT2LOL]

Tcase: Technically this refers to the temperature that you can measure using a thermocouple embedded in the centre of the heat spreader, but this is only done in the factory by Intel or by users willing to drill their heat spreaders open to insert a thermocouple. Therefore Intel provided a diode between and below the cores with a reading calibrated by the BIOS which can be used instead. This reading can vary greatly when the BIOS version is changed, but will not necessarily change if the BIOS calibrations were not altered between versions.

This is Tcase, not measurable unless the CPU lid is drilled into.

Hum Gary I thought it was measured by the dual diodes as picked up by that HW app. It's a pity there is no real good sound way to measure either it or the Tjunction unless as you say you drill in and embed some thermo devices yourself.
The only way that I can think of is to use one of those rather expensive infra red spot guns but you would have to get very close I suspect.

I did find this but it is only mentioning one diode and I ma left wondering if that means just one core or all cores in mutli core CPU.
http://en.wikipedia.org/wiki/Thermal_diode


Am currently looking for info on how the CPU's are put together brick by brick as it were but it is quite difficult to find any constructional details.

 

[Britton30]

According to the temp article you linked, Tjunction is the same as what we call core temp and arrived at with diodes between and under the CPU die. It is all very confusing.

 

[ICIT2LOL]

According to the temp article you linked, Tjunction is the same as what we call core temp and arrived at with diodes between and under the CPU die. It is all very confusing.

Yep I know mate I found this in the meantime which is a little more simple

http://www.arcticsilver.com/PDF/WhtPr/CalibratingPIII_InternalThermoDiode.pdf

 

[Layback Bear]

I still suggest you need a base line. Just changing a bunch of things a bunch of times really doesn't tell you much.
Keeping records of the specs including anbient temp, volts, component temps, and over clocks ect. Using the same test in the same fashion and recording the changes will give you the proper information you need.
When you get your computer where you want it you will know what got you there and what didn't.

 

[ICIT2LOL]

Yep I guess you are right Bear one could go nuts trying to find out the actual meaning of life eh??

 

[Layback Bear]

Do these types of things in methodical fashion and keeping records in my opinion is the only way to do it. When one does a bunch of Hodge Podges things found on the internet you don't even know how you got a BSOD while over clocking.

 

[essenbe]

From Core temp FAQ: I can only see a single temperature reading.
AMD processors based on the Phenom and Phenom II (Athlon II, Sempron II, Turion II, etc.) and all of AMD's new CPUs such as the FX and APU series only have a single thermal sensor.
Thus Core Temp will only display a single CPU temperature reading. There is no way of getting a per-core reading on these processors.

What I don't know, is that single diode measuring Tcase or TJ Max? There is a big difference. TJ Max can run 10-15°C higher than Tcase. Which is that single temp reading with AMD? I don't know.

 

[Erick Aguilar]

Now.. the question. I believe the temperature I am getting by coretemp to be more adequate after performing tests while gaming and OCCT. BUT in case that I am wrong.

What happens when a chip does exceed it's temperature? I've always believed that if you do it one time it's all over forever. Or is there a moment when the chip will override all settings to protect itself or just BSOD if it gets -too- hot?

 

[Erick Aguilar]

From Core temp FAQ: I can only see a single temperature reading.
AMD processors based on the Phenom and Phenom II (Athlon II, Sempron II, Turion II, etc.) and all of AMD's new CPUs such as the FX and APU series only have a single thermal sensor.
Thus Core Temp will only display a single CPU temperature reading. There is no way of getting a per-core reading on these processors.

Yeah, I read that. I also read that at over 45C Coretemp does display the most accurate readings for AMD chips as voice by an AMD representative apparently.

Yesterday I tested the chip with Coretemp and cpu temp respectively. While Cpu temp reached 75, core temp was displaying 67 all night (ran the tests as I slept) No problems or errors. But Im still downclocking until I get a better cooler.

 

[kbrady1979]

Now.. the question. I believe the temperature I am getting by coretemp to be more adequate after performing tests while gaming and OCCT. BUT in case that I am wrong.

What happens when a chip does exceed it's temperature? I've always believed that if you do it one time it's all over forever. Or is there a moment when the chip will override all settings to protect itself or just BSOD if it gets -too- hot?

I know Intel chips will start to throttle their frequency back when they hit their heat threshold, in order to keep the chip from burning itself out. I would assume AMD chips work in a similar fashion.