AMD FX9590 running waaay too hot. Running 20 degrees hotter than usual at idle for no apparent reason.

Gundalfthegreat

Prominent
May 25, 2017
12
0
510
I have an amd fx9590, with a full open loop cooling system. I have a 280x140 45mm thick full copper radiator, 2 decnt push fans and on 140x37.5 thick high power pull fan (only had room for one on the back for the rad). It's got pure distilled water in and a bit of biocide in to stop it getting gunged up. And it has a 16 watt 800lph water pump with a 4m pressure head. So all the kit is good and i've just completely dissassembled the whole loop, taken apart and thoroughly cleaned everything including the pump internals. So it's all like new. It used to keep the idle temps really low, between 28 and 34 degrees Celsius. But for the last couple of months I've noticed the idle temps jumped up massively.. So my first thought was maybe the water block is clogged, so i took it all apart, cleaned it all and put it all back together. It's dropped the temps by about 8 degrees but it's till idling at about 55 degrees which is waay too high. Used to be a lot lower. It's not that the heat is being dissipated, there's a lot of heat coming from the radiator so the heat is being taken away from the cpu it just seems to be running stupidly hot. The fans are all set up correctly, the pump is wired directly to the power supply so is always on full. There's no reason why it should be running this hot. I've also checked task manger to see if there's any background apps causing the cpu to work harder, and accorfing to speedfan the cpu is only being used 1-2% so it doesn't seem like there's a dodgy program making it overheat. And of course when i try and play any games the fans spin up and make a huge amount of noise moving tonnes f air and the computer heats up my room like a furnace. I've no idea what could be causing it.
 
Solution
No i'm not getting mixed up. I don't have the 8350 and my Core voltage is set at 1.4V on the core for 5Ghz.
I have different profiles on both systems and have never had my Core voltage to 1.5V. It is at that voltage that would be creating the excessive heat.

I can only assume the MB or PSU are not regulating voltages well enough however, I have never used the ASRock board but I have read about problems and that it's not the best MB for OCing. The CPU would start to throttle above 75-80C and the VRMs over 80C would shut the MB down.

Having read your MB manual, you do not have a Bios with much scope for voltage regulation and other Bios Tweaks unfortunately, so you will have to resort to an underclock to be able to drop your VCORE.

If...
Victem number 9999 to amd fx 9*** series.

what are your system specifications? just to be clear dont feel bad we get alot of problems from that chip.
you need an excellent cpu cooler an motherboard in order to support it, worst part is, it gets out peformed by a intel pentium 60 dollar chip with a free stock cooler.
 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510
As for thew spec, i've got a asrock 970A-G3.1 motherboard, it says it's compatible with this motherboard that's why i chose it, but the heat sinks on the motherboard do seem to get incredibly hot sometimes despite having a lot of air blowing over them. I Have decent water cooling loop and as i said it used to stay pretty cool. On a cool day it's idle as low as 28 degrees Celsius. I have 2x8 GB corsair vengeance ram sticks (2400MHZ), a 500GB toshiba SSD which has the os on (win 10), and a 3TB HDD, And i have a Sapphire radeon r9 380x gpu. As for the performance, it's pretty decent for the price. i got a score on cinebench of 816 for the cpu, my friend has a intel i7, not sure which one but it cost about £300 more than mine (Mine was £160 brand new his was well over £400) and considering the price difference his cpu only managed about 120 more on the cinebench test. So the performance is decent, it's always run hot as you expect with on of these processors but for some reason in the last couple of months the temps have gone way up for no apparent reason. It's not overclocked, it just runs at the standard 4.7ghz and can boost to 5Ghz under load.

Could a bios update help? Or could it be something to do with windows 10?
 

DSzymborski

Curmudgeon Pursuivant
Moderator
One problem with FX+ series motherboards is that companies play fast and loose with "compatible" for the 125W/220W CPUs. I've seen a *lot* of throttling on 760/970 motherboards that claim compatibility. Though this may not be the issue you have right now.

How have you been monitoring temperatures? Reporting of FX temperatures is quite inaccurate, you really need to be looking at the AMD Overdrive thermal margin.

What is your power supply? I'm starting to wonder if your CPU is getting more voltage now than it should be (MOSFET failure is typically a bit more spectacular). There are a lot of fail points for this platform (I abhor this CPU with the heat of a thousand suns because of all the problems that pop up).

 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510
Yeah i'm having a lot of trouble with it. Regretting my life decisions right now. It would be fin if it wasn't for the insane heat, it's literally like having an electric fan heater on in my room.

Okay, the power supply i have is an Aerocool Interrogator 850w 80plus bronze power supply. And i've been using speedfan to monitor temps, but i do have amd overdrive, i don't know what the thermal margin is though. I've just had another look on the bios though, and it's saying the cpu temp is 57.5 degrees Celsius and i've only just turned it on. Which is pretty much same as what speedfan was saying. As i say it used to idle as low as 28 degrees, usually around 32, but now the temps have gone though the roof. It's now at 61 degrees and isn't even doing anything, it's just on the bios screen. I don't know anything about voltages, or what the normal voltages should be. There's a thing that say's vcore: 1.472v. I think these processors are meant to be around 1.5v?
 

DSzymborski

Curmudgeon Pursuivant
Moderator


Bah, I was hoping something weird was going on with the voltage. You're in a kinda awkward situation in that your PSU is a terrible choice for this build -- hec-made low-end junk inappropriate for a gaming rig, let along one with this CPU -- but I can't say for certain that the power supply is directly causing *this* problem. I'd still replace that PSU at some point, it's a serious mistake to have that running with a 220W CPU.

Have you tried underclocking it yet? Need to see how normal the heat looks when you turn it back into an 8350 (it's just a binned 8350, after all). And also, that's one fo the last resorts, turning a 9590 into a functional 8350, which is better than nothing.

I'm hoping MeanMachine pops in this thread, I think he's the enthusiast here that has had the most success playing around with this CPU (I think he's the one).
 

DSzymborski

Curmudgeon Pursuivant
Moderator


Not really, no. The Kaby Lake Pentiums fare very well in game benchmarks against the 9590, thanks to its significantly superior single-core performance. The KIA doesn't have more power per cyclinder than the Cadillac. It's not that the Pentiums are amazing -- they're budget -- it's just that the 9590s are lousy when we're talking gaming. Productivity, of course, is another story.
 
Unfortunately you have chosen a budget build around a Beast of a processor. The PSU is also an inadequate low efficiency and budget quality unit which does not help.
The VRMs on the MB would get hot with anything above 4.7GHz stock frequency and 1.5 volts on the core using the 970A-G3.1

You should not need your core voltage at 1.472V at 4.7GHz and should only require 1.3V max at stock frequency.
The FX-9590 requires a liquid cooler of at least H110 twin fan radiator or equivalent for anything above 4.8GHz and without Turbo Boost tech enabled.

I recommend you set CMOS to default at 4.7GHz fixed frequency and start your core voltage at 1.275V. If its unstable then increase core voltage in .010 increments till stable then test for stability under load.

Your current temperatures are way too high and should be around 10-15C above ambient room temperature and 60-65C under load.

I trust you have a decent tower case and plenty of case fans to extract heat.

 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510


Thanks for the reply.



Have you had a look it my cooling system specs? I did start off with an arctic 240mm closed loop cooling system with 4 puch pull fans, but it wasn't up to the task even at stock settings.

It's a good cooler. EK copper waterblock, 280mm (45mm thick) full copper Alphacool radiator, Phobya DC12-400 water pump, 250mm cylinder reservoir, I have 2 NZXT high static pressure push fans on the radiator and a Silverstone 140mm (37.5mm thick) large high static pressure fan on the back of the rad. I only had room for one pull fan. I have 2x 120mm cortair fans in the ceiling exhausting air out of the top and an arctic f12 exhausting out of the back. The case i have is an Aerocool Aero-800, it's not massive, It's a mid gaming tower so it's a decent size. I have all the air coming in through the radiator at the front of the case and exhausting out of the top and the back. I had considered getting another 240mm radiator in the roof of the case, but as i said, it used to idle and run at really good temps for a 9590.

This was my first build so I didn't think the motherboard would be that important as long as it was specified as being compatible with the 9590. I wouldn't have thought the power supply would have been the issue because while it was pretty cheap it has decent specs and has plenty of power. The power supply never gets hot or anything. I haven't meddled with the clock speeds or anything, it's just at its default settings. I think the motherboard might be the problem because the vrm's do get extremely hot.

How would i go about turning down the voltage? Should i do it through the bios? Or using AMD overdrive?

 
I dont know where that guy is getting his information, but its a fact the 9590 peforms with all those expensive water cooling and high end boards the same as a budget g4560 chip with a stock cooler.

not to bully but dude
your cpu 120 dollars
your cpu cooler 120 dollars?
motherboard that cna handle that cpu 120 dollars?

total 360 dollars

pentium= 60 dollars
stock cooler = free
motherboard = 70 dollars

same peformance?+ upgrade path to an i5 or i7.

the motherboard is very important for that cpu, only a handfew can support the nearly 200w cpu draw(theyre expensive).
you can turn down the voltage with amd overdrive.

what motherboard have you got?
 
That guy if you mean me is getting his information from experience. I have two systems built around the FX-9590 and clocked at 5.2GHz and stable with proof. Not that I would recommend a build around the FX-9590 cause it's EOL and superseded by the RYZEN 7 1800X on the AM4 platform which I also have.
Yes everyone knows that the FX-9590 has a 220W TDP draw and will run hot if it's requirements are not met.

The only MB I could recommend running the FX-9590 is the Crosshair V Formula Z.

Yes it's just not worth throwing money at I agree however it is a very capable processor for rendering and video editing apart from the cut and paste BS that many detractors here love to add.

I would not recommend the FX-9590 for gaming alone as single core performance is not a match with even the Intel i3 however with the current price and 8 cores it will still play most AAA games at max settings and with some games being developed and utilizing more core intensive games it can be an attractive proposition baring in mind the above at it's current price.

@ Gundalfthegreat, If you wish to continue using this processor then I suggest you replace your TIM with AS5 or MX4 as from what you describe you have sufficient water cooling capacity. If anything you can mount a fan blowing onto the VRMs which will help.
As far as voltages are concerned you have the ability in Bios (I do not recommend using Windows based software to do this) to reduce your Frequency even below stock and then reduce the Core voltage to minimize heat. Every minor decrease in core voltage will exponentially reduce your temperature.

Go into your Bios>OC Tweaker menu and set OC mode to "Manual"
Enable APM.
Set CPU voltage to 1.275V initially.
If system does not boot then up the voltage in .010 steps till it boots then conduct a stress test.
Download AIDA64 and HWInfo monitor for testing and put them side by side on your desk top. Run the stress test from AIDA64 Tools menu and monitor temps. Run the test for 10mins and take screen shots at the 10min mark and post your results here. If temps reach 75C then stop the test.
I will be able to check to see if your temps and voltages are OK.

 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510


I did a bios update which didn't make any difference to the temps. then i strated messing with voltages.

I tried to use AMD overdrive first, Initially the NB voltage was 1.4v, and the cpu voltage was sett to 1.5v. I dropped the cpu voltage down to 1.3 and it dropped the cpu temp down to 40 degrees Celcius then started going awol. It crached and wouldn't restart for ages. When it eventually came back on it had reset itself in ad overdrive, the nb voltage reset to 1.1250v, and the cpu voltage was back to 1.5v.

Also, i tried a stress test after it came on and seemed to have reset itself to stock settings, i used prime 95, cpu usage was 100% only did it for about 5 minutes, the cpu temps initially rose to a maximim of 90 degrees then started to fall back down to about 71 degrees. The idle temps are at 53 degrees now but thats with all the fans on full, i still need to reset the fans back to normal through the bios. I also noticed that during the stress test the vcore voltage dropped from 1.5v down to 1.4v. I don't know if this is normal?

I've just tried resetting the voltages through the bios. I can't set the cpu voltage below 1.5v without it completely messing up. You sure you're not getting mixed up between a 8350 and a 9590 for the voltages? Because this cpu won't work below 1.5 volts for some reason (9590). I think the only way to lower the temps is going to be to dial back the clock speeds a bit which is a bit disappointing because I didn't want to have to sacrifice performance.

 
No i'm not getting mixed up. I don't have the 8350 and my Core voltage is set at 1.4V on the core for 5Ghz.
I have different profiles on both systems and have never had my Core voltage to 1.5V. It is at that voltage that would be creating the excessive heat.

I can only assume the MB or PSU are not regulating voltages well enough however, I have never used the ASRock board but I have read about problems and that it's not the best MB for OCing. The CPU would start to throttle above 75-80C and the VRMs over 80C would shut the MB down.

Having read your MB manual, you do not have a Bios with much scope for voltage regulation and other Bios Tweaks unfortunately, so you will have to resort to an underclock to be able to drop your VCORE.

If the changes in Bios voltages that you make will not retain then the system would revert to default and if your voltages are not stable then your PSU is suspect as the FX-9590 demands a stable and efficient voltage. It really is a beast of a CPU.
BTW; If your rail voltages are not within +or- 5% under load then forget it. They can be monitored with either HWinfo monitor or AIDA64 in the voltage graph.
 
Solution

Gundalfthegreat

Prominent
May 25, 2017
12
0
510


Right I've just had a look over everything. I uninstalled AMD overdrive because I wasn't sure if it was interfering with the settings. I went back into the bios and checked everything, I realized i hadn't disabled the turbo boost so i've disabled that, and with the apm enabled and everything else you've mentioned i've managed to get the voltage down from 1.5 to 1.4375v at 4.7Ghz. That is the absolute lowest i could get it at that clock speed. It's brought the idle temp down to 46 degrees which is a massive improvement and it seems to be stable at the moment. I just did a short stress test, the temps rose up rapidly then went back down to between 70-80 degrees Celsius with the cpu at 100% usage.

The ambient temp today (according to the weather forecast for my area) is 27 degrees celcius today, so i'm not sure if these temps are good considering. Also, I do already have mx4 thermal paste on, used the pea method apply it making sure i didn't use too much. So i don't think reapplying the heat paste would make any difference unless i tried using a bit less and spread it out using a credit card? Everybody seems to have very mixed views on how to apply thermal paste.

So unless you think it would be worth giving that a try, I don't think I can get the temps any lower without lowering the clock speeds. If the thermal throttling kicking in when the cpu is under load, will reduce the clock speeds anyway I'm thinking it might be worth lowering them slightly along with the voltages again to reduce the idle temp further.

Also I installed hwinfo but i'm not sure how to check the rail voltages? So i'm not sure what they are at. Also does the northbridge voltage make mych of a difference? Because according to hwinfo it's at 1.25v despite being set at 1.125 in the bios.

Thank you very much for your help and input by the way!
 

DSzymborski

Curmudgeon Pursuivant
Moderator


That's the problem. The listed wattage and efficiency rating of a PSU is important information to know, but it's like the gas mileage/gas tank size when you're shopping for a car. Knowing the quality of the parts inside -- will they last -- and knowing how the PSU deals with ripple and how well it regulates voltage are among the far more important things to know when buying a PSU. Any power supply can slap "600W" on the side. There's no regulatory authority and it's not uncommon for low-end power supply manufacturers/brands to just make up the numbers.

This is why you see PSU recommendations collected in various curated lists. An unknown power supply is guilty until proven innocent; unless it's on a known platform and/or dissected and examined with an oscilloscope and other specialized parts, it's not a recommended purchase. Unfortunate, this PSU isn't even an unknown. It's a platform made by hec, one of the least recommendable manufacturers out there. So you get low-quality Chinese capacitors, low-quality sleeved fan, mediocre ripple suppression, etc. The voltage regulation actually isn't that bad and this one at least has protection circuitry (not always a thing on these low-end PSUs), but this is not a quality PSU.

And when essentially dealing with a heavily overclocked version of a 125W CPU, these things become even more important, because you're not asking for the power of an office build or even an average gaming build, but an extremely high-stress scenario for a power supply. While this PSU may have been fine, though still a curious choice, for an office build that uses the web and Microsoft Office, it's not a recommendable situation for a high-level overclock on a hungry CPU. If you're going to run a 9590, you need to have top-notch component support.

You're very fortunate MeanMachine saw this thread, as I noted above, he's the poster that I know of with the most experience and success wrestling with this CPU. Personally, I'm more pessimistic about 9590 builds and both don't counsel using a 9590 in a build or encourage people to move to a better platform as soon as possible rather than sinking more money into what is essentially an underperformer for the trouble if you're gaming.

 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510


Well I've learned a lot today. I just thought the quality and expense of the power supply meant it'd be more reliable and powerful. As i say it was my first build, I wanted watercooling and was on a tight budget so when i saw the fx9590 come up cheap, i knew they had decent power for the money (at first glance anyway) knew they ran a bit hot but figured that wouldn't be too much of a problem with watercooling.... ha. It's a learning curve eh. I thought the power supply was a good one, shows how much i know. 850w plus 80 bronze (I don't know what plus 80 Bronze means tbh). Sounded good. Now i know why people spend so much on them now. Like i say though, I've had this system a year and it's only the last couple of months that the temps went through the roof. It used to idle at very low temps.

And I was a bit baffled when MeanMachine41 was saying the voltage needed to be set at no more than 1.3v. As all the other things i'd read about it were saying 1.5 volts and every time i tried to lower the voltage it wouldn't load windows. Only managed to reduce the voltage slightly after doing as Meanmachine said. So it must be down to the power supply as you and mean machine said. He must have a very good power supply!

I already knew the motherboard was struggling to cope as soon as i got it because of the insane heat coming from the heatsinks on it. I didn't realise manufacturers would bend the truth when it came to comparability.

In the end i've had to dial down the clock speeds to 4Ghz.... ahem, and put the voltage down to 1.2v. I might be able to get the voltage a bit lower But i don't want to make it unstable. It's the only way i could get the temps low so it doesn't heat up my whole room while gaming. It still might for all i know i haven;t tried it yet! Only had time for a short stress test. It now idles at 39 degrees Celcius when the air outside is 27 degrees. So, bit of a blow to performance but I guess that's just a hrash lesson for me about power supplies and overclocked CPU's.
 
DSzymborski is quite correct in what he says and I consider the PSU to be the single most important hardware component in an extreme system. Read my PSU fact in my signature.

You can check your 12V, 5V, 3.3V Rail voltages in HWinfo by expanding the pages (Two green arrows bottom left) under your MB heading. Check voltages when system is under load.

It's good to see at least your temps have come down and hopefully more stable AND that your now fully understanding what your dealing with. :)

As an added note. I noticed you have your top radiator fans exhausting from your case to cool the radiator. I suggest you change the fan orientation to bring in ambient air past the radiator. It can make a difference if pockets of hot air are building up inside the case.
 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510


Yeah I didn't realize the psu was so important with a highly strung cpu. At least now i know! I am just going to keep my current set up for the next couple of years i reckon then eventually upgrade to a ryzen 7 set up or something similar. But for now i'm just going to make do with what i've got. It's good enough for gaming and with the gpu I've got i don't need to worry about the cpu bottle-necking gaming performance so it'll be fine for the next couple of years.

I'm just doing a stress test now and i've found the voltages info you were talking about. Here's a screenshot:
tIcpfnn

http://imgur.com/tIcpfnn

I'm not sure exactly what this all means. But i think it's within +- 5%. Also i'm not sure why all the cores aren't at 100%? I think it might be because i had other things open while doing it. As you can see it's cooler on a stress test than it used to be at idle.

And here are some pictures of my setup so you can see how the cooling system is set up:
kuxdgBB

http://imgur.com/kuxdgBB
pz8Xidk

http://imgur.com/pz8Xidk

As you can (hopefully if its worked) see, I have a 280x45mm thick radiator in the front, 2 decent push fans on the front of that then 1 large 37.5 mm deep fan pulling to help bring as much air though the radiator as possible and blow air across the motherboard. So ambient air come in through the rad first then is exhausted out of the top and back. The large pull fan is brilliant by the way:

http://www.silverstonetek.com/product.php?pid=366&area=en

It moves a ton of air with very good static pressure but is actually pretty quiet for the amount of air it moves.

Thank's for taking the time to read my thread and help me with my problems! Very much appreciated! I'm going to keep tweaking it until i find the best balance between performance and temps then i'll post how i've set it all up.


 

Gundalfthegreat

Prominent
May 25, 2017
12
0
510


As I've said many time already. I have 2 high static pressure fans pushing air in, the front of the radiator and 1 very large pull fan on the back, so 3 fans on it in total. There's just about room to squeeze another one on the bottom half of the back of the rad but only half of it would stick out over the hard drive box bit at the bottom of the case, there are gaps in the hard drive box thing to let air through but i figured the air would be very restricted so it wouldn't really be worth bothering with a pull fan on the bottom.

Since i Do have a spare 140mm fan though i might just try it and see if it makes any difference. But it'll be a pain to install it because of how tight the space is.
 
Well you really have underclocked a considerable amount down to 4GHz and a Vcore of 1.16V.
Your temps are maxing out at 52C so you can bring it back up.

Your BCLK ratio is 200 so set your multiplier at 22.5 for 4.5GHz and Vcore at 1.260V - 1.35 and test that.
Which stress tester are you using.? Hope it's not P95.

HWInfo is not either reading or showing your 12v or your 5v Rails. Try HWMonitor to see if that will read your voltages during the stress test. Just don't have two monitoring systems open at the same time.