• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD FX 9590 overheating

T

Theairwolf

n00b
Joined
Apr 9, 2014
Messages
1
Hey, i have a custom build pc, specs:
AMD FX 9590 4.7 GHZ
8 GB DDR3 2133 Mushkin RAM
2 GB VRAM, GV-N660OC-2GD
2 1200 RPM Fans, 2 2800 RPM FANS,
Scythe Samurai ZZ Rev.B CPU cooler
M5A99FX PRO R2.0 Motherboard
Corsair 500 W Power Supply

When i play heavy games, like Battlefield 4 etc. the temperature raises to 85°C and my pc shuts down. I know, i should be able to play BF4 on ultra, but he just shuts down. Is there a solution, or does someone of you had it yourself and what was the right solution for you? Thanks

Theairwolf
 
rennyf77 said:
what video card? at first blush, i'm betting insufficient power supply.
Click to expand...

video card its listed.. its a Gigabyte GTX 660. that isn't a issue nor a reason for overheat the CPU unless the heat produced its extremely high and the CPU cooler draw part of that of that air making worse the cooling performance of the CPU but thats hardly this case..

Theairwolf said:
Hey, i have a custom build pc, specs:
AMD FX 9590 4.7 GHZ
8 GB DDR3 2133 Mushkin RAM
2 GB VRAM, GV-N660OC-2GD
2 1200 RPM Fans, 2 2800 RPM FANS,
Scythe Samurai ZZ Rev.B CPU cooler
M5A99FX PRO R2.0 Motherboard
Corsair 500 W Power Supply

When i play heavy games, like Battlefield 4 etc. the temperature raises to 85°C and my pc shuts down. I know, i should be able to play BF4 on ultra, but he just shuts down. Is there a solution, or does someone of you had it yourself and what was the right solution for you? Thanks

Theairwolf
Click to expand...

fast answer.. the cooler.. its extremely insufficient for that monster of processor

in what head fits the idea of put that kind of low profile cooler in a 220W chip??.. man, thats a extremely hot processor, and that cooler its a low profile cooler with 3 heat pipe + 92mm fan built to UP TO 130W chips, and with a poorly cooling performance at those numbers... thats your reason of overheating... that chip need a big badass Air cooler or AIO Water cooler kit and trust me of the big ones.. the absolutely minimal AIO Water cooler kit i recommend for a chip like that its the Corsair H80i with preference on H100i-H110.. nothing less..
 
a cpu approaching 85'c shouldn't be reason enough to cause a shutdown unless we're talking about an unstable overclock. i've primed and ibt'd i7 chips beyond 85'c plenty of times.
 
rennyf77 said:
a cpu approaching 85'c shouldn't be reason enough to cause a shutdown unless we're talking about an unstable overclock. i've primed and ibt'd i7 chips beyond 85'c plenty of times.
Click to expand...

AMDs arent as happy with high temps. Either way, the BIOS probably has a setting for temp shut down. 85 is extremely hot for just gaming.
 
asus recommends plugging in the 8 and 4 pin cpu power eps connectors for boards that have them because the centurion cpus (fx 9370, fx 9590) can draw more than the 300w from the 8 pin eps. essentially insufficient power delivery can produce shutdowns. i've been reading about this a lot the last few days since acquiring an fx 9370 and asus sabertooth board.
 
rennyf77 said:
a cpu approaching 85'c shouldn't be reason enough to cause a shutdown unless we're talking about an unstable overclock. i've primed and ibt'd i7 chips beyond 85'c plenty of times.
Click to expand...

Yeah thats intel chips.. They work completely different from AMD chips.. In intel what people have to pay more attention are for individual core temperatures, and those have a tjmax of 100C in average (depending of generation it vary for example 98 for a 2600K and 105 for a 3770K).. However thats the core temps, the Tcase or entire processor temperature its another whole story, you will never see a intel running at such high Tcase temp and they are way more tolerable to heat for example a 4770K have a max safe Tcase temp of 72C while in a FX8350 its if i can remember correctly 61.5C. And at the moment a FX8350 its near of that temp it will start to throttle badly or cause major instability issues.. Just as example my 3770K have a tcase of 67.2C but even overclocked it barely touch the 48C line in the CPU temp under heavy IBT stress test.. But core temps in the other hand are in the 70C-72C line under same test..

Core temps for AMD FX chips are just ignored because generally are just inaccurate readings.. The FX9590 run at a high voltage from factory and run at a near maximum safe temp, those tends to run at 55C average with good coolers so with crappy coolers it might run hotter.. I guess due to the high TDP nature of the FX 9590 they might be have more tolerance to the temperatures but even i think 85 its extremely high for a FX chip, nothing strange that the system just shut-down to prevent damage to the chip..
 
rennyf77 said:
asus recommends plugging in the 8 and 4 pin cpu power eps connectors for boards that have them because the centurion cpus (fx 9370, fx 9590) can draw more than the 300w from the 8 pin eps. essentially insufficient power delivery can produce shutdowns. i've been reading about this a lot the last few days since acquiring an fx 9370 and asus sabertooth board.
Click to expand...

Thats correct.. In fact isn't even recommended to use cheap boards with a good VRM cooling and with 8pin EPS to support the high power they draw.. Also its recommended to have good cooling in the VRM section.. Those tend to run way hot even in some high-end boards.. A bad VRM cooling can just shut down the machine to prevent damage, however in this case i can be almost sure the issue its due to bad cooling in the processor
 
cheapest solution for the op is to buy another cpu cooler. the scythe samurai is rated for cpu's up to 130w. the centurion chips are 220w parts.
 
rennyf77 said:
a cpu approaching 85'c shouldn't be reason enough to cause a shutdown unless we're talking about an unstable overclock. i've primed and ibt'd i7 chips beyond 85'c plenty of times.
Click to expand...

Umm the max temp rated on these is 70c lol. They throttle to about 1ghz at anything over 80.

To the OP - get nothing less than a Evo212 or noctua D14 if you want air and if your going AIO nothing less than an H100i.

With a 9590 with a 100mhz oc I am stress testing and hitting 70c which is the max recommended temp. This is with an H100i.

Great chip - has a super low Vid and I'm sure I could manage 5.2 out of it with a proper watercooling kit but this is my backup PC :p

Due to the misinformation on the web I actually expected the h100i to do quite well considering I lapped the chip and used liquid metal TIM.

This is not the case. Don't expect to overclock much with anything short of a custom loop.
 
klepp0906 said:
Umm the max temp rated on these is 70c lol. They throttle to about 1ghz at anything over 80.

To the OP - get nothing less than a Evo212 or noctua D14 if you want air and if your going AIO nothing less than an H100i.

With a 9590 with a 100mhz oc I am stress testing and hitting 70c which is the max recommended temp. This is with an H100i.

Great chip - has a super low Vid and I'm sure I could manage 5.2 out of it with a proper watercooling kit but this is my backup PC :p

Due to the misinformation on the web I actually expected the h100i to do quite well considering I lapped the chip and used liquid metal TIM.

This is not the case. Don't expect to overclock much with anything short of a custom loop.
Click to expand...
Are you running at turbo speed + 100 (5100) or base clock 4.7 + 100 for 4.8?? Becouse my chip wont break 50C at 4.8 non stop stress testing. This chip runs so much cooler than my past 8350
Edit... and when I say 50c. I mean I don't like my temps to go much past 40C 45 max during a stress.
 
Go into the BIOS and manually set the voltage to 1.375V and see how that runs. When the voltage is set to AUTO it is crazy high and contributes to way high temps especially on low end cooling which unfortunately you have. A Cooler Master Hyper 212 is the minimum youre gonna want with these. The stock coolers are also surprisingly adequate but for these 9590 monsters, you want some beefy cooling. Also switch the CPU/NB voltage from AUTO to manual and set it to 1.20V.

That will help a bunch in lowering your temps but youre gonna have to invest in some better cooling. Try to get something like at least that Hyper 212 but youd be really doing yourself a favor by getting something like a Corsair H80i instead. Theyre gonna cost you a little more but theyre up to the job of handling cooling duties for a 9550 running at stock settings.
 
9590 with an Antec 620.. no problems here. I had a 212 Evo previously, it also worked fine, but re-purposed it for another machine so decided to try a watercooler.
 
I'd go for at minimum a 212 evo. Or a h100i if you have the $ . I had a FX6350 @ 4.8ghz with an evo and it never went over 60 in games. Make sure your voltage is under control if you haven't already download cpu-z and use your load line calibration settings in your bios to keep the votage from jumoing around. @4.7ghz you will probably need 1.45-1.47ish volts. Good luck. If that doesn't work try re-seating the HSF. Open up your case while its running and see if the heat drops, if it does you need a better cooling solution. Try pointing a house fan in the general direction of the CPU and monitor the temps.
 
Last edited:
klepp0906 said:
Umm the max temp rated on these is 70c lol. They throttle to about 1ghz at anything over 80.

To the OP - get nothing less than a Evo212 or noctua D14 if you want air and if your going AIO nothing less than an H100i.

With a 9590 with a 100mhz oc I am stress testing and hitting 70c which is the max recommended temp. This is with an H100i.

Great chip - has a super low Vid and I'm sure I could manage 5.2 out of it with a proper watercooling kit but this is my backup PC :p

Due to the misinformation on the web I actually expected the h100i to do quite well considering I lapped the chip and used liquid metal TIM.

This is not the case. Don't expect to overclock much with anything short of a custom loop.
Click to expand...

lol you probably have the cooler bad mounted.. that cooler shines with big surface chips like the 9590 and make an exceptional work.. my 8350 with a H110 work all the day at 5ghz at 1.51v with a sabertooth 990FX and temps are never over 59C. so you have a problem with your cooler really.
 
45nm Phenom II C3 locks up or crashes ~75°C. It's a good thing the 32nm chips have a bit more thermal headroom.
 
lots of bad information in this thread.

first TJmax for the FX processors is 90c. When they reach 90c they will auto shut down. In most cases you will never reach 90c because the motherboard bios will usually promote throttling to reduce temperature, but not always.


That CPU cooler is a joke for this cpu. AMD recommends a Corsair H80 or better. If you wish to stick with air cooling, then Notcua NH-D14, ThermalRight Silver Arrow, Phanteks PH-PC12DX would all be suitable there are few others.

AIO liquid coolers anything that is better than the H80 or simliar. Such as Antech 920, Corsair H100, H80I,H90 Thermaltake Water 3.0 Pro,Cooler Master Seidon 120XL or the bigger versions of these.

Replace the cpu cooler and you will fix your problems.
 
I had an FX 9590 recently, and I struggled to cool it at stock with a custom 360 radiator loop. I can't imagine trying to run that thing on a dinky little air cooler. You need a BIG air cooler, Phanteks, Noctua, etc. Even then those will struggle with that chip if you turn off APM.
 
The CPU cooler is definitely underrated for that CPU's TDP. The cooler is rated for dissipating up to 130W, and AMD says your CPU can generate 220W of heat. Downdraft coolers are not generally the best for high-wattage CPUs because they have a tendency to also heat up the area around the CPU socket, which can adversely affect the performance of all of those power delivery elements around the socket. A tower cooler rated for less dissipation may hold up well enough because it is at least moving the hot air away from the CPU socket area.

I would check into the airflow in your case as well. The downdraft cooler and a large dead-air space will just become a big heat trap.

The PSU may be a little on the light side for that particular setup, given the enormous power draw that FX 9590s are capable of. You can always check your draw at the wall with a Kill-A-Watt or similar for a crude method of calculating something in the neighborhood of your total system power draw using the formula: wattage from the wall * PSU efficiency = approximate system power draw. If you're anywhere near 500W from the wall, then the PSU may be struggling to keep up. Irregular voltage/ripple has the potential to heat things up, which is one of the reasons why good PSU testers like [H] will give them a hot-box torture test to see how the components fare when they are outside of their comfort zone.

Adjusting the voltage in the BIOS is also worth looking into. Some motherboards do push too much voltage to the CPU. Voltage (V) X Current (A) = Power (W or VA). Therefore, more voltage = more heat to dissipate if current stays the same. I have personally had mixed results with under-volting AMD CPUs, but when it has worked, it definitely reduced temperatures by a significant margin. You could also try disabling any turbo-core multipliers and just running the CPU at stock to find a stable minimum voltage for constant use.
 
LigTasm said:
I had an FX 9590 recently, and I struggled to cool it at stock with a custom 360 radiator loop. I can't imagine trying to run that thing on a dinky little air cooler. You need a BIG air cooler, Phanteks, Noctua, etc. Even then those will struggle with that chip if you turn off APM.
Click to expand...

I also have a custom loop and had an issue with heat. Even more so because I was cooling the VRMs on the Asus CH5 Formula-Z. Maybe it was silly of me thinking I could cool the VRMs and the CPU with an EK Coolstream RAD XT 360 and Gentle Typhoon 1450 RPMs in push/pull, who knows.

I replaced both with a Haswell 4770K and MSI z97 Gaming-9 and couldn't be happier.
 
Last edited:
Good grief it's not that big a monster. If you leave it all on AUTO I guess it could be cause it uses way too much voltage for its stock and turbo speeds. I'm running 4.8 GHz at a mere 1.4V and running my H100 on medium fan speeds and having no problem. The auto voltage for turbo is near 1.50 if I remember right which is WAY too high. The 9590 is just a high binned 8350. You don't have to have a $600 custom loop to cool one of these especially at stock speeds. A H80i would be great or a H100i would be even better. Buy a better cooler and dig into the BIOS to manually set the voltage and you'll see a big difference.
 
ohms said:
I also have a custom loop and had an issue with heat. Even more so because I was cooling the VRMs on the Asus CH5 Formula-Z. Maybe it was silly of me thinking I could cool the VRMs and the CPU with an EK Coolstream RAD XT 360 and Gentle Typhoon 1450 RPMs in push/pull, who knows.

I replaced both with a Haswell 4770K and MSI z97 Gaming-9 and couldn't be happier.
Click to expand...
I don't understand how that setup wouldn't be enough. I'm using a Swiftech H220 so I have a thinner 2x120mm radiator. Did you have heat issues only while stress testing with Prime95/IBT or with general use as well?
 
1/2 bent said:
Are you running at turbo speed + 100 (5100) or base clock 4.7 + 100 for 4.8?? Becouse my chip wont break 50C at 4.8 non stop stress testing. This chip runs so much cooler than my past 8350
Edit... and when I say 50c. I mean I don't like my temps to go much past 40C 45 max during a stress.
Click to expand...

I mean 4.8. This is with all cores enabled and turbo disabled. Takes me a constant 1.5v last I checked.

Running it at 5.0 now with 1.54vcore which is prime stable but hits 79-80c which I'm fine with. Doesn't come close to that under regular useage and idle is around 35c. All cores running at 5 w that kind of voltage just generates more heat than the h100i can remove. This is with 2 extra fans in push/pull fwiw.

In contrast my naked haswell runs about the same temps at 1.5v/4.7ghz but is under a monster custom loop. Difference being the haswell has headroom lol.
 
Araxie said:
lol you probably have the cooler bad mounted.. that cooler shines with big surface chips like the 9590 and make an exceptional work.. my 8350 with a H110 work all the day at 5ghz at 1.51v with a sabertooth 990FX and temps are never over 59C. so you have a problem with your cooler really.
Click to expand...

Yea I thought to myself - is it possible after 20 years of building and an insane custom setup on my haswell that I botched screwing in two screws...... You have turbo core on probably.

That wasn't sarcasm btw. Really. 2 options - either people online are full of it and don't understand how to stress test OR their was a problem with my install.

So here we are a month or so later. Half a dozen reseats and a 25 dollar tube of liquid metal later - no difference.

A few things id like to add. First and foremost, when I say temps I mean under 100% load not via something like hyper PI, via IBT on maximum. Even LinX falls a few degrees short of that. As does prime.

Secondly, I added two more fans for push/pull and certainly dropped the temps quite a bit. Not sure if people are listing their temps with upgraded fans or what but 50c under full load with a stock h100i on one of these chips just isn't realistic.

With the extra fans and the lap job, as well as liquid metal and the fan speed set to max performance - my 9590 hits 80c in IBT at 5.0ghz.

Said temps and speed are the result of 1.5375v and ultra high LLC etc. anything less and IBT locks up. I haven't got to trying prime yet which will likely yield a few degrees but may require more voltage.

This chips power requirement goes through the roof post 4.8 or so. As I mentioned previously, this is with turbo core and all the silly power saving stuff off so all cores are running at 5.0 which needless to say is dramatically warmer (no doubt constitutes some of the wonky "baseline" numbers you see on the net)

I'll have to try it back at 4.8 and a much lower voltage to see how much of a difference adding the two fans made.

(Now that the thread jacking is complete) lol
 
Last edited:
LigTasm said:
I had an FX 9590 recently, and I struggled to cool it at stock with a custom 360 radiator loop. I can't imagine trying to run that thing on a dinky little air cooler. You need a BIG air cooler, Phanteks, Noctua, etc. Even then those will struggle with that chip if you turn off APM.
Click to expand...

This

Except - what's APM do? I turned it off as it sounded like power management mumbo jumbo (first time using AMD since Athlon thunderbird)

At this point, whatever I can do to get my temps down so I can keep it at. 5.0 24/7 would rock.

For the other guys I was talking with. Just did a quick run with turbo core back on and at 4.7 and it's maxed at 51c. So 30c difference from 300mhz and enabling all cores simultaneously. Decisions decisions. Who runs one of these at stock (people who can't afford a second custom loop I guess lol). But really, 4.9 sounds ugly - and if it's advertised as 5.0 id like to think it has Atleast a little head room.
 
Last edited:
I don't think that the AMD chips really like IBT as a stability test. It just overheats them for no apparent reason. Someone else can explain it much better than me. I'm idling on a 4.7GHz OC with my FX-9370 @1.416v. My temps are 34c. Running 24/7 with 4 fans on a H100i. I did install an extra 120mm fan on my VRM modules since the stock cooler fan would have normally blowing on them, and with the H100i there it's missing a fan to cool them. Also I have a 120mm fan on the back of the motherboard blowing on the CPU socket again to keep the VRM's cool. I got a pack of 6 fans for free from TigerDirect after rebate so I figured I would use them. ;)

In the bios I have Turbo cut off. I manually set my memory to what Corsair recommended for me. Everything else is stock for my motherboard except I set the processor to 4.7GHz. I even run the power saving modes since the computer never cuts off unless the electricity is interrupted. It is stable @5.0GHz if I want it to be. I don't really see the point in 5.0GHz as my room is significantly warmer at those speeds for very little benefit.

Ask some of the guys what to use to check stability. Mine has been stable since November of 2013, so I don't even bother with those programs. I have the launch Asus Sabertooth 990FX version 1 motherboard that doesn't officially support the FX-9370 because of the VRM's. That's why I took 5 minutes to make sure that they stay cool. And the number one reason that you would have stability problems is that the VRM modules are overheating because of a lack of airflow over their heatsinks.

Hope I didn't ramble too much. Need a nap, but wanted to help before I took one. :)
 
ohms said:
I also have a custom loop and had an issue with heat. Even more so because I was cooling the VRMs on the Asus CH5 Formula-Z. Maybe it was silly of me thinking I could cool the VRMs and the CPU with an EK Coolstream RAD XT 360 and Gentle Typhoon 1450 RPMs in push/pull, who knows.

I replaced both with a Haswell 4770K and MSI z97 Gaming-9 and couldn't be happier.
Click to expand...

I know - I have considered pulling the loop in my sig off and putting it on the 9590 but realistically it's a backup PC. I was just trying to see what it was capable of but was thermally limited much sooner than I anticipated.

I took people's info online to heart and didn't realize most of them were either stress testing with something like blend or aida coupled with leaving turbo core on and other "tech" which leads to very very skewed info.

To be fair, I'm happy with the chip though. It feels quick (though benchmark numbers would beg to differ. Hyper PI takes literally twice as long) I'm just hoping it's as robust as the intel chips are.

For those following the data I've been tossing up. Just ran with turbo disabled at stock to compare to my enabled to show the difference. At stock voltage and temperature simply enabling all cores added 5c maxing at 56c. Mind you this is lapped, with liquid metal, and with 2 extra fans. I'd imagine with the h100i stock without the lap and some AS5 the temps would be a 10c difference if not better. ie: in line with my former temps.

Best as I can tell a h100i is plenty for a factory 9590 (which to be fair generates a ton of heat and uses more voltage than most have the nuts to put into their intel) so it's no slouch.

But an enthusiast level 9590 (ie: one with all the consumer trappings turned off) and a mild overvolt/overclock eats it for breakfast.


Which no one should find surprising or take offense to. What intel chip do you know that you can put those kind of volts thru and run at those speeds on an AIO?

Just figured id give some specifics and another PoV for those planning to buy a 9590 and go the cheap route with cooling. Had it not been a backup PC which I somehow convinced myself I wouldn't care about the overclock, and my budget - I would have never considered anything but a custom loop.
 
klepp0906 said:
This

Except - what's APM do? I turned it off as it sounded like power management mumbo jumbo (first time using AMD since Athlon thunderbird)

At this point, whatever I can do to get my temps down so I can keep it at. 5.0 24/7 would rock.

For the other guys I was talking with. Just did a quick run with turbo core back on and at 4.7 and it's maxed at 51c. So 30c difference from 300mhz and enabling all cores simultaneously. Decisions decisions. Who runs one of these at stock (people who can't afford a second custom loop I guess lol). But really, 4.9 sounds ugly - and if it's advertised as 5.0 id like to think it has At least a little head room.
Click to expand...

APM makes your motherboard slow your processor to a lower speed at a significantly lower voltage. My 4.7 OC @1.416v becomes 1.410GHz while idling @0.864v with APM on. The only disadvantage to this is that it will sometimes mess with overall processing power when doing things like rendering video. I can best describe it as getting stuck at a lower speed. Cutting off APM fixes that, but at the cost of extra heat. Since I don't render video all day, and when I do I like to use my R9 290 anyways, I leave APM on. But if you're doing benchmarks or stability tests leave it off.
 
Another thing I didn't consider is what people are using to check their temps since amd insists on being about as confusing and un calibrated as one can imagine.

Some may be listing package temp as core temp. Or even worse using something which has like a 20c offset as I found out real quick realtemp vs fx9590 = nice cool CPU lol.
 
If you want to benchmark Pi, you have to use a special program to enable all of the AMD features on the chip. Your Pi score will soar. I don't remember offhand what it was called, but I'm sure once again that some of the guys can assist you in finding it.
 
cageymaru said:
APM makes your motherboard slow your processor to a lower speed at a significantly lower voltage. My 4.7 OC @1.416v becomes 1.410GHz while idling @0.864v with APM on. The only disadvantage to this is that it will sometimes mess with overall processing power when doing things like rendering video. I can best describe it as getting stuck at a lower speed. Cutting off APM fixes that, but at the cost of extra heat. Since I don't render video all day, and when I do I like to use my R9 290 anyways, I leave APM on. But if you're doing benchmarks or stability tests leave it off.
Click to expand...

I figured it had something to do with something like that. On the formula z it's pretty well hidden and on by default but I tend to turn everything like that off bar virtualization.

Im actually testing the difference now with it on and off to see temp wise. Wonder what makes that any different than cool n quiet or the other 2 or 3 throttling/temp/power adjusting options under CPU in the UEFI.

I'm used to the intel stuff like eist but it's the same stuff just named differently. Being my first and rig in a decade+ I forget the names but I'm sure you know what I'm talking about.

Thanks for the reply :p
 
cageymaru said:
If you want to benchmark Pi, you have to use a special program to enable all of the AMD features on the chip. Your Pi score will soar. I don't remember offhand what it was called, but I'm sure once again that some of the guys can assist you in finding it.
Click to expand...

Ah! That is very good to know. I wondered :p

To the guy who posted the link, thank you very much as well! (Didn't want to make another post just to say Ty lol)
 
Oh and read post number 26 in the thread. I don't think you want to be using IBT on an AMD FX chip. Ask some of the guys about it for an explanation.
 
You must log in or register to reply here.
Back
Top