FX-9590 - What to Check for Temps ? :: Hardware and Operating Systems

All Discussions > Steam Forums > Hardware and Operating Systems > Topic Details

Niknokinater Mar 8, 2017 @ 4:30pm

FX-9590 - What to Check for Temps ?

Hi friends!

So I've been searching around for everything about the temperatures on this thing. I've learned that there is a significant offset in the actual CPU (as I imagined) which spooked me off to eyeing the another temp value that changed with the fan sound instead for some months. I come for feedback on what I should be looking at to determine a proper value.

I have the AMD FX-9590 @ 3.4Ghz no boost and Antec Kühler 650.

HWiNFO32 / ITE IT8721F / Temperature 1:
-Changes progressively, the one I look at
27-31°C Idle, 38-48°C Gaming, 59°C Prime95

HWiNFO32 / CPU [#0] AMD FX-9590 / CPU 0:
-Changes super quickly, actual CPU sensor? The offset one I read about?
2-10°C Idle, *don't look at it enough whilst gaming, 36°C Prime95 (same test as one above)

Antec cooler light:
-Changes at consistent values with Temperature 1 whilst at values below 50-ish°C
-When full load from Prime95 is given, the cooler light changes colors indicating higher temperatures faster than Temperature 1

AMD Overdrive gives me a Thermal Margin that looks like it's calculated from the CPU 0 reading and if that's messed up then it's as useful as CPU 0.

Any advice? :)
...that isn't a rant about why I bought this CPU? :)

http://steamcommunity.com/sharedfiles/filedetails/?id=880261150

Last edited by Niknokinater; Mar 9, 2017 @ 3:36pm

< >

Showing 16-28 of 28 comments

Niknokinater Mar 9, 2017 @ 7:31pm

Originally posted by upcoast:
Originally posted by Niknokinater:
In HWmonitor, the max reached 42°C for The Temperature -> CPU and 27°C for AMD FX-9590 -> Temperatures -> Package.

42c sounds normal also, it'll vary by game though.

I found that GTA 5 and Just Cause 3 give my 1.356v FX8350 a work out 45-47c but stuff like Farcry 4 I hover in the 40-42c. My cooler Tuniq Tower 120 is a bit dated and not as good as the cm212evo.

http://www.frostytech.com/articleview.cfm?articleid=2655&page=4

https://us.hardware.info/reviews/5981/8/17-cpu-water-coolers-test-better-than-air-cooling-better-coolingperformance-than-air

Random info for you on coolers incase you're interested ^.

Noted, and thanks for those links! I did enjoy myself looking at them.

#16

Xtenzor Mar 9, 2017 @ 10:22pm

your temps are good now. maximum recommended temp for the processor is around 62C(from what I remember), but easily can be handled to 70C.
your socket temperature is 42C and CPU core temperature 27C according to what you've said. Pretty cold

Last edited by Xtenzor; Mar 10, 2017 @ 6:02am

#17

Bad 💀 Motha Mar 10, 2017 @ 6:10am

I have done many tests where I did get Temp readings of 70-80*C (yes on AMD FX) and it was fine; however if the CPU is getting above that 62-65*C ranges that generally might also mean your VRMs on the motherboard are also VERY hot. And this is usually the part of the Motherboard that will fail before your CPU does. So just avoid running it so hot when using AMD FX series.

#18

MrMcSwifty Mar 10, 2017 @ 7:15am

Originally posted by upcoast:
Package temp gets ignored from me.

It's interesting to me that you say this because that's always the one I pay most attention to.

I'll have to see if I can dig up the email in which an AMD head states that this readout is what most accurately reflects what's going on with the CPU temps. I've posted it before so it's in my history somewhere...

To summarize, "package temps" are the ones being read directly off the diode, and they are accurate above 40C. The "CPU temp" (or Temp1, in OP's case) are actually socket temps and are influenced by too many outside forces to be considered accurate.

Package temps are also what are monitored AMD Overdrive, only displaying the thermal overhead rather than the acutal temps. If you take the thermal margin displayed in AMD OD and subtract it from 70, you'll get what is displayed under package temps in HWMon.

Of course, none of that explains why OP's is apparently 225C, lol...

Last edited by MrMcSwifty; Mar 10, 2017 @ 7:25am

#19

Bad 💀 Motha Mar 10, 2017 @ 8:01am

Still though; install a few of the of common monitoring apps; as sensor accurate will vary from not just Motherboard, but also CPU and the actual App you use.

So best bet is always have a few on-hand. They are generally all very small, so why not have them installed on each PC. If one is deemed completely inaccurate, then get rid of it.

#20

upcoast Mar 10, 2017 @ 9:01am

Originally posted by McSwifty:
Originally posted by upcoast:
Package temp gets ignored from me.

It's interesting to me that you say this because that's always the one I pay most attention to.

I'll have to see if I can dig up the email in which an AMD head states that this readout is what most accurately reflects what's going on with the CPU temps. I've posted it before so it's in my history somewhere...

To summarize, "package temps" are the ones being read directly off the diode, and they are accurate above 40C. The "CPU temp" (or Temp1, in OP's case) are actually socket temps and are influenced by too many outside forces to be considered accurate.

Package temps are also what are monitored AMD Overdrive, only displaying the thermal overhead rather than the acutal temps. If you take the thermal margin displayed in AMD OD and subtract it from 70, you'll get what is displayed under package temps in HWMon.

Of course, none of that explains why OP's is apparently 225C, lol...

OK, so I did a little test playing JC3 HWMonitor Package hit 37c and cpu temp hit 44c, I guess the big question is, does amd provide the 62c threshold temp against the Package temp or the cpu temp?

If package at 37c is accurate it would seem amd chips run pretty cool for something that sucks such a huge amount of electricity.

I've seen the last couple of big amd temp discussions and I figure this current one is still going to be just as confusing.

#21

Bad 💀 Motha Mar 10, 2017 @ 9:03am

It an app like HWMonitor it usually works like this for AMD:
CPU temp = Socket
Package temp = actual CPU cores < This is where it is not uncommon on better boards to see this safely reach 70*C+

#22

upcoast Mar 10, 2017 @ 9:23am

Originally posted by Bad-Motha:
It an app like HWMonitor it usually works like this for AMD:
CPU temp = Socket
Package temp = actual CPU cores < This is where it is not uncommon on better boards to see this safely reach 70*C+

This is where it gets confusing ^ I'd technically have a 33c leeway on my package temp which is weird because my understanding is that at 1.356v I'm over the normal fx8350 volt but on a mediocre older tuniq tower my temps according to package read outs of 37c are spectacular.

This is why I go by the cpu temp at 44c and that I'm fairly poor on the cash end so blowing my system just isn't something I want to do but if it's really 37c and the threshold is package 62c then I'm clear ether way.

Why I don't use a bunch of monitoring programs at the same time is that I started getting wild readouts. I used HWinfo, HWM and AMD overdrive what a mess it made, so I use one.

#23

MrMcSwifty Mar 10, 2017 @ 10:10am

Originally posted by upcoast:
OK, so I did a little test playing JC3 HWMonitor Package hit 37c and cpu temp hit 44c, I guess the big question is, does amd provide the 62c threshold temp against the Package temp or the cpu temp?

I found that post I referred to earlier:

Concerning your question regarding the temperatures with your processor. The maximum temperature threshold is 62 Celsius which set for the internal die (core) temperature of the chip.

The core temperatures have an equational offset to determine temperature which equalizes at about 45 Celsius thus giving you more accurate readings at peak temperatures. The hindrance in this is the sub ambient idle temperature readings you speak of.

The Cpu temperature is read form a sensor embedded within the socket of your motherboard causing about a 7-10 Celsius variance form the actual Cpu temperature, which may be what you are reading about on the net.

As long as your core temperature has not exceeded the high side of the 60 degree mark for extended periods of time you should be ok. 62 degrees holds a generous safety net to begin with.

Alex Cromwell
Senior Technology Director
Advanced Micro Devices

He doesn't mention "package temp," but that is what his reply was in regards to, and what he is referring to when he says "internal die (core) temperature." So 62C should be counted against the package temp, which can be verified by what I said earlier, by comparing it against AMD's own Overdrive software. For me, the two have always lined up.

My question has always been why OD uses a 70C thermal limit rather than 62C for FX. I would guess it just uses a universal figure to account for FM2 APUs and others with higher limits.

FWIW, the CPU "socket" temp has always been a few degrees hotter than package temps on my 8350/990FX too, so this doesn't seem so unusual, and would be in line with what he says about a 7-10 degree variance. Remember that this sensor is in proximity to the CPU socket, VRM, etc and is not in contact with the heatsink like the internal diode is, which may account for the slightly higher temps. Or perhaps lower temps for someone with a different hardware config and/or better airflow.

#24