AMD FX 8320 too hot

zrossk · Jan 10, 2015

Hello, I have just recently finished building my first PC and everything works great except for one thing, the CPU runs way too hot. Idle, it runs at about 50 degrees, but when doing more intensive gaming, it gets up to 70+ in a couple minutes, and at the rate at which it goes up, i know it could get up to 80 if i had not ended the game. I am using the stock cooler, which i read could keep it at decent temps. I also did not install my own thermal paste as i read it came preapplied. I will most likely be purchasing an after market cooler. What's the reason it is running this hot? Did i install something wrong? Specs: AMD FX 8320 8 core @3.5 GHz, Gigabyte GA-970A-DS3P ATX AM3+, Sapphire Radeon 280x, 8 gigs RAM, and a cooler master HAF 912 case. Any help appreciated, thanks.

holyrage · Jan 10, 2015

yes u didn't seat the cooler well seriously that's way too damn hot for a FX chip

Max safe core temp recommended by AMD is 62 degrees celsius.

maximum socket temperature is 70 degrees celsius.

its best to keep the core temperature under 55 degrees though to keep the lifetime longer

it is hot!

zrossk · Jan 10, 2015

holyrage :

Is there any way i can reposition the cooler without ruining the thermal paste so that i can use the PC before my aftermarket cooler arrives?

rulejunior · Jan 10, 2015

Morning Zrossk. So this is a bit of an odd one, but I do believe I know what it is. So, the AM3+ Socket had three chipsets which are 970, 990X, and 990FX. Your board is using the 970 Chipset which was designed for budget systems that don't use super powerful/power hungry cpus like the FX-8320. So, after some digging around I found that the board you have doesn't have Heatsinks across the 4+1 power phase (It does if the board is a rev 1.3/1.4, it should tell you this on the board itself). This lack of heatsinks can and will contribute to a spike in north bridge temps which will cause your CPU to heat up since it is on the North Bridge. My suggestion is to get an aftermarket cooler for now, and to eventually replace that board with something from either the 990X or 990FX chipset as they are better suited for running a cpu with a TDP of 125 watts like the Fx-8320. Also, you are technically killing off your cpu with the heat, even though it isn't your fault. The one thing about AMD cpu's is that they really aren't designed to go above 60 degrees Celsius for long periods of time, and the higher the temp, the shorter that time.

zrossk · Jan 10, 2015

rulejunior :

Morning Zrossk. So this is a bit of an odd one, but I do believe I know what it is. So, the AM3+ Socket had three chipsets which are 970, 990X, and 990FX. Your board is using the 970 Chipset which was designed for budget systems that don't use super powerful/power hungry cpus like the FX-8320. So, after some digging around I found that the board you have doesn't have Heatsinks across the 4+1 power phase (It does if the board is a rev 1.3/1.4, it should tell you this on the board itself). This lack of heatsinks can and will contribute to a spike in north bridge temps which will cause your CPU to heat up since it is on the North Bridge. My suggestion is to get an aftermarket cooler for now, and to eventually replace that board with something from either the 990X or 990FX chipset as they are better suited for running a cpu with a TDP of 125 watts like the Fx-8320. Also, you are technically killing off your cpu with the heat, even though it isn't your fault. The one thing about AMD cpu's is that they really aren't designed to go above 60 degrees Celsius for long periods of time, and the higher the temp, the shorter that time.

Thanks for the reply. Do you recommend a cooler that would keep it at a safe temperature for now?

rulejunior · Jan 10, 2015

Normally I would suggest the Hyper 212 Evo, but with these temps, you are going to need some serious horsepower. Noctua's NH-D15 is a pretty powerful cooler and should take care of it. An All in One Liquid Cooler is a good idea, but I wouldn't recommend it just because they can be a pain in the ass to install, and if you don't know what you are doing, something can break. If I had to recommend one of them, something from the Corsair H series, but I would check to see what will fit in your case from it.

mdocod · Jan 10, 2015

zrossk,

List of possible sources of your problems:
1. CPU fan not plugged into the CPU fan header on the board (or not plugged in correctly).
2. Plastic protective cover on HSF was not removed before installation.
3. Failure to apply thermal compound to heatsink before installation. (brand new boxed heatsinks will normally have thermal compound already on them)
4. Accidental introduction of debris to the surface of the CPU or CPU HSF before installation.
5. incorrect use of software to observe CPU temps (improper support for your hardware, improperly calibrated, or simply reading the wrong temp readout. Socket temp instead of core temps, for example)
6. Insufficient pressure between HSF and CPU from installation error.

rulejunior's input on this thread is riddled with misinformation. Ignore.

--------------

rulejunior :

Actually there have been at least 10 different chipsets used on AM3+ socket boards. Including, for example, the 760G, 770, 785g, 880g, 890FX, 890GX in addition to those you have listed. Technologically the AM3+ socket can be supported with BIOS modifications to many older chipsets, which has been done by many board makers.

Your board is using the 970 Chipset which was designed for budget systems that don't use super powerful/power hungry cpus like the FX-8320.

The chipset has nothing to do with the VRM design on the board. There are 970 chipset boards with awesome CPU VRM's, and there are 990FX boards with terrible CPU VRMs.

So, after some digging around I found that the board you have doesn't have Heatsinks across the 4+1 power phase (It does if the board is a rev 1.3/1.4, it should tell you this on the board itself). This lack of heatsinks can and will contribute to a spike in north bridge temps

The lack of VRM heat-sinks contributes to elevated VRM component temps primarily, which can effect the temperature of anything in close physical proximity. I'm not sure why you are singling out the "north bridge" as a component that would suffer from elevated temps. The north bridge on the AM3+ platform is integrated on the CPU. North bridge temperature is not heavily influenced by VRM temps as the north bridge is cooled by the CPU cooler.

which will cause your CPU to heat up since it is on the North Bridge.

The north bridge is integrated on the CPU. The CPU is not "on" the north bridge. You seem to have some confusion over what the north bridge is.

My suggestion is to get an aftermarket cooler for now,

Most aftermarket coolers will actually make things worse for VRM temps, as the OEM cooler provides the best airflow over the VRMs on the motherboard. Making sure the OE heatsink is properly installed and working is a far better solution in this application than an aftermarket heatsink, unless the aftermarket heatsink is a down-blowing variety.

and to eventually replace that board with something from either the 990X or 990FX chipset as they are better suited for running a cpu with a TDP of 125 watts like the Fx-8320.

Again, chipset and VRM quality are not a mutually guaranteed relationship. Yes there are many 990FX boards with great VRMs, but there are also 970 and even 760G boards that can run an FX-8320 at stock speeds without any problems. Some of those 970 boards have fantastic VRM configurations that can overclock these chips pretty well.

Also, you are technically killing off your cpu with the heat, even though it isn't your fault. The one thing about AMD cpu's is that they really aren't designed to go above 60 degrees Celsius for long periods of time, and the higher the temp, the shorter that time.

AMD core temp reporting is actually on a scale that measures "remaining thermal margin," and does not actually produce a real temperature reading. When software is properly calibrated, the inverted remaining thermal margin readout is supposed to align "0 margin remaining" with a temperature reading of 70C, which is the maximum recommended long term operating temp for piledriver. When calibrated to this scale, 90C is the critical shutdown temp.

-------------

Normally I would suggest the Hyper 212 Evo, but with these temps, you are going to need some serious horsepower. Noctua's NH-D15 is a pretty powerful cooler and should take care of it. An All in One Liquid Cooler is a good idea,

There's obviously something wrong with the implementation of the stock cooler in zrossk's build. These CPUs do dissipate a lot of heat, but the stock cooler is more than capable of keeping up when properly implemented. Without first identifying why the stock cooler isn't working, there's absolutely no good reason to go replacing the HSF... The 212 EVO, D15, and AIOCLC's you are throwing out as good cooling solutions are all bad options for the motherboard in question here, as they will all reduce VRM cooling and introduce MORE problems than they will solve. We don't need CPU cooler capable of dissipating 300W here, we need the 140W cooler working correctly.

zrossk · Jan 10, 2015

mdocod :

zrossk,

List of possible sources of your problems:
1. CPU fan not plugged into the CPU fan header on the board (or not plugged in correctly).
2. Plastic protective cover on HSF was not removed before installation.
3. Failure to apply thermal compound to heatsink before installation. (brand new boxed heatsinks will normally have thermal compound already on them)
4. Accidental introduction of debris to the surface of the CPU or CPU HSF before installation.
5. incorrect use of software to observe CPU temps (improper support for your hardware, improperly calibrated, or simply reading the wrong temp readout. Socket temp instead of core temps, for example)
6. Insufficient pressure between HSF and CPU from installation error.

rulejunior's input on this thread is riddled with misinformation. Ignore.

--------------

rulejunior :

Actually there have been at least 10 different chipsets used on AM3+ socket boards. Including, for example, the 760G, 770, 785g, 880g, 890FX, 890GX in addition to those you have listed. Technologically the AM3+ socket can be supported with BIOS modifications to many older chipsets, which has been done by many board makers.

Your board is using the 970 Chipset which was designed for budget systems that don't use super powerful/power hungry cpus like the FX-8320.

The chipset has nothing to do with the VRM design on the board. There are 970 chipset boards with awesome CPU VRM's, and there are 990FX boards with terrible CPU VRMs.

So, after some digging around I found that the board you have doesn't have Heatsinks across the 4+1 power phase (It does if the board is a rev 1.3/1.4, it should tell you this on the board itself). This lack of heatsinks can and will contribute to a spike in north bridge temps

The lack of VRM heat-sinks contributes to elevated VRM component temps primarily, which can effect the temperature of anything in close physical proximity. I'm not sure why you are singling out the "north bridge" as a component that would suffer from elevated temps. The north bridge on the AM3+ platform is integrated on the CPU. North bridge temperature is not heavily influenced by VRM temps as the north bridge is cooled by the CPU cooler.

which will cause your CPU to heat up since it is on the North Bridge.

The north bridge is integrated on the CPU. The CPU is not "on" the north bridge. You seem to have some confusion over what the north bridge is.

My suggestion is to get an aftermarket cooler for now,

Most aftermarket coolers will actually make things worse for VRM temps, as the OEM cooler provides the best airflow over the VRMs on the motherboard. Making sure the OE heatsink is properly installed and working is a far better solution in this application than an aftermarket heatsink, unless the aftermarket heatsink is a down-blowing variety.

and to eventually replace that board with something from either the 990X or 990FX chipset as they are better suited for running a cpu with a TDP of 125 watts like the Fx-8320.

Again, chipset and VRM quality are not a mutually guaranteed relationship. Yes there are many 990FX boards with great VRMs, but there are also 970 and even 760G boards that can run an FX-8320 at stock speeds without any problems. Some of those 970 boards have fantastic VRM configurations that can overclock these chips pretty well.

Also, you are technically killing off your cpu with the heat, even though it isn't your fault. The one thing about AMD cpu's is that they really aren't designed to go above 60 degrees Celsius for long periods of time, and the higher the temp, the shorter that time.

AMD core temp reporting is actually on a scale that measures "remaining thermal margin," and does not actually produce a real temperature reading. When software is properly calibrated, the inverted remaining thermal margin readout is supposed to align "0 margin remaining" with a temperature reading of 70C, which is the maximum recommended long term operating temp for piledriver. When calibrated to this scale, 90C is the critical shutdown temp.

-------------

Normally I would suggest the Hyper 212 Evo, but with these temps, you are going to need some serious horsepower. Noctua's NH-D15 is a pretty powerful cooler and should take care of it. An All in One Liquid Cooler is a good idea,

There's obviously something wrong with the implementation of the stock cooler in zrossk's build. These CPUs do dissipate a lot of heat, but the stock cooler is more than capable of keeping up when properly implemented. Without first identifying why the stock cooler isn't working, there's absolutely no good reason to go replacing the HSF... The 212 EVO, D15, and AIOCLC's you are throwing out as good cooling solutions are all bad options for the motherboard in question here, as they will all reduce VRM cooling and introduce MORE problems than they will solve. We don't need CPU cooler capable of dissipating 300W here, we need the 140W cooler working correctly.

Thanks for the help. I took the CPU fan off the CPU and went through the things you listed. The thermal paste looks like it's applied evenly, no plastic or debris, and the latch has it pretty tight. Is it possible something on the motherboard is faulty?

mdocod · Jan 11, 2015

What software are you using to read these temps? (note #5 in list)

-------

My advise at this point, would be to download AMD Overdrive and compare the thermal margin remaining to the temp readings in the software you are using to observe this.

I'd be willing to bet the software you are using is not calibrated correctly to generate a correct reading from the thermal margin output.

zrossk · Jan 11, 2015

mdocod :

Ok, will try Overdrive. I tried HWMonitor and OCCT. OCCT measured a couple degrees cooler than HWMonitor, but still not an acceptable temperature. Also, something that tells me that the readings could be right was that on the thermal paste there was a small burn mark. Is this normal?

zrossk · Jan 11, 2015

mdocod :

Just tried AMD Overdrive. I'm getting a thermal margin of around 30-40 while idle. Is this dangerous? Also, i noticed on OCCT that my CPU fan was running under 1000 rpm while idle. Could this be increased and if so would it help?

mdocod · Jan 11, 2015

Thermal margin is the amount of headroom between the current temp, and the maximum temp recommended for continuous operation.

0 thermal margin remaining is *supposed* to be calibrated to a 70C temp reading in software for piledriver based CPUs. When people talk about "max temp" for these CPUs being 62C, 60C, 61C, 70C, whatever, it's important to understand that any of these could be correct if that's how the software is calibrated. I believe it should be calibrated to 70C. The linux kernel developers for the K10temp driver which supports family 15h CPUs and lm-sensors packages think it should be calibrated to 70C, and the enterprise world that uses the linux kernel all use that standard. By calibrating it this way, we should never get a negative reading on an inverted "human readable" scale, the CPU temp will read from 0-70C in the usable thermal margin, which is fine. Seems like almost all 3rd party windows software deviates from this standard by anywhere from a little, to a LOT.

When the thermal margin reads 40C, the temp reading in another piece of software should be inverted and read 30C (70C - 40C = 30C).

"idling" at a 40C thermal margin does seem high and suggests to me that something is out of whack (unless you have abnormally high ambient temps). When vishera truly idles down (low power state via cool/quiet, reduced voltage, C1E and C6 gating enabled) the thermal margin should increase to ~60C or more, (which causes a correctly calibrated piece of software to read ~10C core temps) Are you absolutely SURE the CPU is under no load when you read that thermal margin?

That "burn mark" does not sound good to me; however, are you sure you aren't just observing the natural discoloration of the thermal paste spread over the copper? Or the text engraved on the top of the CPU? An actual "burn" mark seems highly unlikely, though the point I made about possible debris (4) before may apply here. If debris were preventing the CPU cooler from sitting flat on the CPU, I think it's possible you could have an isolated hot spot that could discolor the paste.

zrossk · Jan 12, 2015

mdocod :

Thermal margin is the amount of headroom between the current temp, and the maximum temp recommended for continuous operation.

0 thermal margin remaining is *supposed* to be calibrated to a 70C temp reading in software for piledriver based CPUs. When people talk about "max temp" for these CPUs being 62C, 60C, 61C, 70C, whatever, it's important to understand that any of these could be correct if that's how the software is calibrated. I believe it should be calibrated to 70C. The linux kernel developers for the K10temp driver which supports family 15h CPUs and lm-sensors packages think it should be calibrated to 70C, and the enterprise world that uses the linux kernel all use that standard. By calibrating it this way, we should never get a negative reading on an inverted "human readable" scale, the CPU temp will read from 0-70C in the usable thermal margin, which is fine. Seems like almost all 3rd party windows software deviates from this standard by anywhere from a little, to a LOT.

When the thermal margin reads 40C, the temp reading in another piece of software should be inverted and read 30C (70C - 40C = 30C).

"idling" at a 40C thermal margin does seem high and suggests to me that something is out of whack (unless you have abnormally high ambient temps). When vishera truly idles down (low power state via cool/quiet, reduced voltage, C1E and C6 gating enabled) the thermal margin should increase to ~60C or more, (which causes a correctly calibrated piece of software to read ~10C core temps) Are you absolutely SURE the CPU is under no load when you read that thermal margin?

That "burn mark" does not sound good to me; however, are you sure you aren't just observing the natural discoloration of the thermal paste spread over the copper? Or the text engraved on the top of the CPU? An actual "burn" mark seems highly unlikely, though the point I made about possible debris (4) before may apply here. If debris were preventing the CPU cooler from sitting flat on the CPU, I think it's possible you could have an isolated hot spot that could discolor the paste.

I'm pretty sure it's idle, AMD Overdrive is only thing up. Today i have been getting 1 long beep before my normal short post beep, which from what i have read means motherboard error. Not sure what to think at this point.

zrossk · Jan 12, 2015

Never mind about the motherboard issue, i believe something plugged into it was coming loose.

mdocod · Jan 13, 2015

My advise would be to acquire a new tube of thermal compound, carefully and thoroughly clean off the heatsink and CPU with cotton (q-tips/pads etc). Use a touch of alcohol for final cleanup. Carefully apply a small pea sized dollop of compound to the center of the CPU, then carefully place the CPU cooler on the CPU and latch it in place, making sure to completely articulate the cam latching mechanism, and of course, making sure no debris get on the CPU or heatsink before installation.

If you're still having abnormally high idle and operating temps then it's time to explore your BIOS settings for improper configuration. Did you buy the board new or used or refurb?

zrossk · Jan 13, 2015

mdocod :

Ok, i will try that. I bought the board brand new from Newegg. Something else i have noticed is very dramatic jumps in CPU frequency often going from 1400 to 3700 MHz, which is overclock for my CPU. Also, very low frames in Minecraft, not too intensive of a game, but i figured the reason for that was the hot CPU. Could that be the case?.

mdocod · Jan 13, 2015

It's perfectly normal for the CPU to run at speeds from 1.4ghz - 4.1ghz depending on workload. From 1.4 -3.5ghz there are numerous power states available. When the CPU is in the highest power state (3.5ghz base) APM and turbo core take over with the option to scale as high as 4.1ghz when temperature and power levels allow.

When a motherboard overheats, it can use those low power states as part of it's self-preservation method, (forcing the CPU into a low power state even when the workload calls for maximum CPU performance).

zrossk · Jan 17, 2015

mdocod :

Alright, thanks for the help, I have ordered Arctic Silver paste + cleaner. One major thing i have noticed is that even when the CPU is around 60, other players and load times in games are very slow while my frames are at 60+, and very slow internet browsing. I did a speed test and it read that my ping is 85 ms, and from what i can remember, that's not too bad. I'm using a Net gear genie wireless adapter. Could this be a problem with the CPU or something else?

bmacsys · Jan 17, 2015

zrossk :

If the heatsink wasn't seated properly the cpu would overheat and shut down literally within seconds. It would barely make it out of POST.

zrossk · Jan 24, 2015

Update: Got new thermal paste and 212 evo installed and it runs 20 degrees cooler. Thanks for all the help.

mdocod · Jan 24, 2015

Hopefully the 212EVO doesn't cause more problems than it solves on your motherboard... I suspect you'll have MORE problems as you've reduced airflow over the VRMs with that switch.

bmacsys · Jan 24, 2015

zrossk :

If you want to help your board out get a couple 80mm fans. Attach one to the vrm heatsink with zip ties. Then place another 80mm fan on your cpu socket with double sided tape. 3M or Scotch. Your board doesn't have many fan headers. So you may have to get some fan Y splitters. They are cheap on eBay. If you ever have the extra money I would look into getting a GA-990FXA-UD3 Revision 4. Cheap as heck on eBay due to them being scrapped from failed bitcoin miners. Has to be R4 though. Scads better phase power than your board.

Blake_17 · Jun 9, 2016

Hi, I know I am resurrecting a dead thread here, but here's what I have discovered. The AMD Fan that comes already mounted to the CPU is pushing air INTO the heatsink rather than using induction and pulling air from its edges through the heatsink. I got my fan and noticed the same thing, it was hitting 62-65*C. So was the MOBO. The Powersupply was getting so hot it was shutting off randomly and restarting despite having two case fans pulling 65cfm a piece running full bore. I flipped the stock fan around so it pulls rather than pushes air, put some rubber spacers in where the screws go in to lift it off a little bit, and viola! I rarely hit above 52*C. But that damn thing is noisy as shit. I am putting a 70-120mm adapter onto it and putting in a big 76cfm case fan to handle the cooling now. The stock cpu doesn't have a very large heatsink, but it seems to get the job done at the very least until I can order a bigger CPU Heatsink.

AMD FX 8320 too hot

Honorable

Distinguished

Honorable

Honorable

Honorable

Honorable

Distinguished

Honorable

Distinguished

Honorable

Honorable

Distinguished

Honorable

Honorable

Distinguished

Honorable

Distinguished

Honorable

Honorable

Honorable

Distinguished

Honorable

Commendable

Share this page