AMD 8320 temperature problems

[scrubbergtx]

I have a 990FX board with an AMD 8320 (8320F written on the chip). My friend gave it to me because his board stopped working. I pulled the cooler off and the thermal compound had turned black and ran off of the CPU. There was NONE left on the CPU. It was all tar and around the edges. So I picked up a new 990FX board and popped it on there, put some arctic silver paste on it, and it works.

The problem now is heat. Right off the bat it was reaching temps of 60 degrees. Max temp is 61 degrees. The cooler is stock but it shouldn't be that bad. It's got copper heat pipes and I even opened the side of the case and pointed a big fan at it. That doesn't help at all.

My solution was under-clocking. I lowered the CPU frequency to 2.0GHz. Now the temperature remains between 40 and 50 degrees Celsius. CPU usage is way up there, getting nearly to 100%, as where before it was maxing out at 33%. Even 50 degrees seems way too hot for this CPU, but it could be related to CPU usage.

Is this CPU damaged? Everything runs flawlessly. The cooler is burned on the bottom. I'm not sure if I should keep it or get an 8350 to replace it since they handle more heat.

 

[Guest]

Scrubbergtx, It may be a bit difficult to wade through the posts But to answer your question go back to my post that has a link in it, the one where I mentioned to download AMD Overdrive and read the link provided. It will explain how AMD figures temperature by a mathematical algorithm. There is also a link inside the one provided that will go even deeper into the subject but may get a bit to tech heavy for the average person.

Hi again Bgunner
I have read that many many years ago. The post I linked is a quick and dirty explanation why Certain motherboards and software will interpret CPU temps differently.
As you already said it is "tech heavy" I completely agree. But this is not the discussion the simple point is that his (the Op) CPU cooler is not upto scratch and more then likely the motherboard (if has enabled turbo core will pump 1.4v to CPU NB which is not good and something AMD has never really resolved with any revision of BD.

 

[maxwellmelon]

information from sticky http://www.tomshardware.com/faq/id-2122665/understanding-temperature-amd-cpus-apus.html


Note: 61c is max socket temp for fx chip not max cpu temp according to amd I hate when people throw this number around. really this is not the cpu temp anyways this is just the temp of the motherboard under the cpu. "thermil margin" is the best gage for fx chips.

note :"Socket temperature is a great indication of your idle CPU temperature, but it is much less accurate during heavy loads. "

Note: "The second, and trickier temperature to understand, is thermal margin. "Thermal Margin indicates how far the current operating temperature is below the maximum operating temperature of the processor."..................This temperature is based off of precise calculations that are very accurate during heavy loads"

^^^why you should go by thermal margin

0c or negitave numbers is overheating. 0c means cpu is is max operation temp. this is also when throttling of the cpu will start. it is really best to not have a thermal margin lower then 10c under 100% load but also from the article which is also in the amd fx data sheet "It is perfectly safe to operate in the full envelope of thermal margin" that means as long as your cpu is not 0c or a negitave thermil margin your cpu is operating in a safe temp range. (though most people will not recommend operating close to the 0c limit)

also note: " This is also the signal that is most often erroneously interpretted by 3rd party programs like HWMonitor, ect. Use AMD Overdrive to measure thermal margin."

edited to make quotes from stickys in red

 

[Guest]

information from sticky http://www.tomshardware.com/faq/id-2122665/understanding-temperature-amd-cpus-apus.html


Note: 61c is max socket temp for fx chip not max cpu temp according to amd I hate when people throw this number around. really this is not the cpu temp anyways this is just the temp of the motherboard under the cpu. "thermil margin" is the best gage for fx chips.

note :"Socket temperature is a great indication of your idle CPU temperature, but it is much less accurate during heavy loads. "

Note: "The second, and trickier temperature to understand, is thermal margin. "Thermal Margin indicates how far the current operating temperature is below the maximum operating temperature of the processor."..................This temperature is based off of precise calculations that are very accurate during heavy loads"

^^^why you should go by thermal margin

0c or negitave numbers is overheating. 0c means cpu is is max operation temp. this is also when throttling of the cpu will start. it is really best to not have a thermal margin lower then 10c under 100% load but also from the article which is also in the amd fx data sheet "It is perfectly safe to operate in the full envelope of thermal margin" that means as long as your cpu is not 0c or a negitave thermil margin your cpu is operating in a safe temp range. (though most people will not recommend operating close to the 0c limit)

also note: " This is also the signal that is most often erroneously interpretted by 3rd party programs like HWMonitor, ect. Use AMD Overdrive to measure thermal margin."

edited to make quotes from stickys in red

Thank you.

 

[bgunner]

information from sticky http://www.tomshardware.com/faq/id-2122665/understanding-temperature-amd-cpus-apus.html


Note: 61c is max socket temp for fx chip not max cpu temp according to amd I hate when people throw this number around. really this is not the cpu temp anyways this is just the temp of the motherboard under the cpu. "thermil margin" is the best gage for fx chips.

note :"Socket temperature is a great indication of your idle CPU temperature, but it is much less accurate during heavy loads. "

Note: "The second, and trickier temperature to understand, is thermal margin. "Thermal Margin indicates how far the current operating temperature is below the maximum operating temperature of the processor."..................This temperature is based off of precise calculations that are very accurate during heavy loads"

^^^why you should go by thermal margin

0c or negitave numbers is overheating. 0c means cpu is is max operation temp. this is also when throttling of the cpu will start. it is really best to not have a thermal margin lower then 10c under 100% load but also from the article which is also in the amd fx data sheet "It is perfectly safe to operate in the full envelope of thermal margin" that means as long as your cpu is not 0c or a negitave thermil margin your cpu is operating in a safe temp range. (though most people will not recommend operating close to the 0c limit)

also note: " This is also the signal that is most often erroneously interpretted by 3rd party programs like HWMonitor, ect. Use AMD Overdrive to measure thermal margin."

edited to make quotes from stickys in red

Thank you.

Now, we all do realize I posted this same link long ago in this discussion? And the same one I referred the OP to look at to clear up how AMD deals with temperatures? I just didn't copy and post parts of damric's write up.

 

[Guest]

Now, we all do realize I posted this same link long ago in this discussion? And the same one I referred the OP to look at to clear up how AMD deals with temperatures? I just didn't copy and post parts of damric's write up.

Hopefully this will get through too.
61c is fine for the health of the CPU / not for the performance as it starts to throttle , IF bios and power settings are set to auto.
Simply put keeping the CPU under 61c at full load will improve performance is bios is reporting Auto/enabled power management.
Running the CPU at 61c is not a problem or detrimental to the lifespan in reality.
The safe long term temperature is any temperature within the specifications of the Processor.

 

[maxwellmelon]

sorry bgunner somehow I missed your post. I see it now. my apologies.

 

[scrubbergtx]

So long as it doesn't fry this new board I'm cool. I spent $120 on the 990FX board and another $80 on a 700W PSU just to throw this together. After under-clocking to 2.0GHz and disabling the boost frequency I'm still able to run Fallout 4 at a consistent FPS so it just needs to survive until I can replace the cooler.

 

[Rogue Leader]

So long as it doesn't fry this new board I'm cool. I spent $120 on the 990FX board and another $80 on a 700W PSU just to throw this together. After under-clocking to 2.0GHz and disabling the boost frequency I'm still able to run Fallout 4 at a consistent FPS so it just needs to survive until I can replace the cooler.

You're not going to fry the board however if you have to underclock the CPU that low just to keep it from throttling something is off. Whether its the cooler you have is damaged (probably based on your description) or at some point in the old system the CPU got burned up a bit something definitely isn't right there. But if you keep it within about 10 deg c of the thermal margin you are fine.

The thermal margin in overdrive is equivalent to approximately 71 degrees C. If you read through the link it says to try to keep the thermal margin at 10 degC or higher to avoid throttling. The temperature I quoted earlier (61) is the point that the CPU maxes out and starts to throttle. Its HIGHLY preferable for best performance to keep it at or below that number. Now using thermal margin this is easy, keep it at 11 or a higher number, 0 is bad. Throttling is not good either, its dropping your performance, your cooling system should keep you out of that, otherwise its not good enough. And eventually getting above that will wear on the cpu or the cooling system with its $5 fan will be working overtime and get to be less and less effective.

Apparently the General couldn't handle this info and decided to delete his account, but I want to make sure you fully understand what you're encountering here, and that allowing the CPU to run above the thermal margin can be bad for its long term health.

 

[kablesnet]

I lost a lot of time to investigate why my AMD processor is so hot. The Below picture summarizes my conclusion.

https://drive.google.com/file/d/0ByWZCAvfiuCAdHEwNTQ3YnN6Qmc/view

 

[bgunner]

I lost a lot of time to investigate why my AMD processor is so hot. The Below picture summarizes my conclusion.

https://drive.google.com/file/d/0ByWZCAvfiuCAdHEwNTQ3YnN6Qmc/view

You do not add temperatures together to get the total temp of the CPU. Thermal Margin shows how many degree's the CPU can heat up before it starts to throttle. Your math and conclusion make no sense at all when dealing with an AMD CPU. 119°C is equivalent to 246.2 °F Most CPU's Even Core 2 Duo's can not go that high before critical failure.

HWmonitor will at time show the wrong temperature for the CPU. Idle temps don't matter but load temperatures do, this is where the package temperature comes in. It is very inaccurate at low temps but under load it is very close to being. Please go back through all the posts and read the link given for Understanding AMD temperature readings.

 

[kablesnet]

I lost a lot of time to investigate why my AMD processor is so hot. The Below picture summarizes my conclusion.

https://drive.google.com/file/d/0ByWZCAvfiuCAdHEwNTQ3YnN6Qmc/view

You do not add temperatures together to get the total temp of the CPU. Thermal Margin shows how many degree's the CPU can heat up before it starts to throttle. Your math and conclusion make no sense at all when dealing with an AMD CPU. 119°C is equivalent to 246.2 °F Most CPU's Even Core 2 Duo's can not go that high before critical failure.

HWmonitor will at time show the wrong temperature for the CPU. Idle temps don't matter but load temperatures do, this is where the package temperature comes in. It is very inaccurate at low temps but under load it is very close to being. Please go back through all the posts and read the link given for Understanding AMD temperature readings.

I did not come out with sum without a logic. I added the thermal margin at the cpu temperature at a certain moment, in order to see what is the maximum temperature at which the processor fails. When AMD Overdrive shows 69 degrees C, that means that there are 69 degrees until the processor collapses. But if we have an ambient temperature of 25 degrees C (so the processor has at least that temperature) and there are other 70 degrees to collapse, it is impossible that the maximum temperature of the processor to be of only 74 degrees. It is obviously higher. And it is at least equal to the room temperature + the Thermal Margin. So, anyway, it is at least 100 degrees celsius.

 

[bgunner]

Now using your logic, which is flawed but I will go into that after, the average if an idling AMD's CPU thermal margin is in the lower 50's so for argument sake lets say 53°C. Lets take an average room temperature of 68°F (20°C) and add this to the thermal margin you get an average throttling temperature of 73°C. This is not critical damage temperature but the point that the CPU will lower the core frequency to cool itself off to prevent thermal damage to the CPU.

Now for the explanation of the flawed Logic and why the temp I just mentioned is not the same as AMD's data. The reason my temps are higher than the throttling temperature and your logic is wrong is because the thermal margin temperature is not accurate at idle/non loaded CPU. This means that at idle the temperature reading is not accurate and can be flawed and in my math above it seems to be off by about 12°C.

Now the 61°C OR in your math 74°C temperature is not the temp that the CPU will fail catastrophically and never work again but the temperature that the CPU will go into self preservation mode to try to prevent catastrophic failure. To calculate that temperature you can not use the Thermal Margin temperature at all because it is just used to tell when the CPU will enter a self preservation mode ( Thermal Throttling ) reducing performance to cool down.

Due to the differences in the silicon between each and every CPU the exact catastrophic failure temperature will vary by as much as 10°C but most likely it will be in the 90-100°C range. This would be the temperature that the transistors will fail and the silicon will scorch causing the CPU to never work again.