FX 9590 stability issues at stock clocks

C

Colonel Sanders

Supreme [H]ardness
Joined
Sep 26, 2001
Messages
5,590
I've been trouble shooting a friend's computer for stability issues. At first I thought it was because the RAM was set too high (without other appropriate changes in BIOS) but I discovered WHEA errors even with everything set to stock clocks. So here's some more detailed info:

FX 9590
Asus Crosshair V Formula-Z (latest BIOS, default settings)
8gb x 2 2400mhz GSkill RAM (running 1333mhz right now)
Corsair 850w PSU

Right, so like I said, with "optimized default" settings loaded from BIOS, I'm seeing stability issues running mainly games. When I run P95 blend test for more than 5 minutes or so, I start getting WHEA errors and CPU cache L1 and L2 errors. Temps are fine AFAIK for this CPU, reaching around 68c max. Cooling is a Corsair H110 by the way.

This has me kind of stumped except for bad hardware, maybe RAM or CPU?

edit: OK, I've been out of the AMD game for a very long time and I guess 68c is actually quite hot for these? I've gotten used to Intel...
 
I've got the same setup. Optimized defaults are not stable.

You have to disable all of the auto overclocking settings and set the bus frequencies manually.
 
Spirit_Retro said:
I've got the same setup. Optimized defaults are not stable.

You have to disable all of the auto overclocking settings and set the bus frequencies manually.
Click to expand...

Thanks for replying! Sorry to be a pain, but could you let me know which settings to change and what they should be? Like I said, I've been out of the AMD loop for so long
 
Those things run hot too. Keep an eye on the temps.
Edit- And running two threads is a no-no obviously.
 
Colonel Sanders said:
I've been trouble shooting a friend's computer for stability issues. At first I thought it was because the RAM was set too high (without other appropriate changes in BIOS) but I discovered WHEA errors even with everything set to stock clocks. So here's some more detailed info:

FX 9590
Asus Crosshair V Formula-Z (latest BIOS, default settings)
8gb x 2 2400mhz GSkill RAM (running 1333mhz right now)
Corsair 850w PSU

Right, so like I said, with "optimized default" settings loaded from BIOS, I'm seeing stability issues running mainly games. When I run P95 blend test for more than 5 minutes or so, I start getting WHEA errors and CPU cache L1 and L2 errors. Temps are fine AFAIK for this CPU, reaching around 68c max. Cooling is a Corsair H110 by the way.

This has me kind of stumped except for bad hardware, maybe RAM or CPU?

edit: OK, I've been out of the AMD game for a very long time and I guess 68c is actually quite hot for these? I've gotten used to Intel...
Click to expand...

Yes 68c is very hot. My processor tops out at about 45c with the same cooler. VRMs can on the board can get close to 60c. They get hotter than normal since water cooling creates less air flow around those heat sinks. Might want to look into increasing the case airflow.

Also- be careful some software reports the VMR temp as the CPU temp. I'd take the temp provided by the corsair link software and add a few degrees to determine proc temp.

As for the changes to bios, I can't remember them at the moment and do not have access to that machine at the moment. However, if you look at the settings they are fairly apparent. Just set everything marked "auto" to the proper value.
 
70c is nothing for silicon to run at.

He probably has bad ram (nothing to do with clocks to be honest) or power supply. Simple fixes.
 
CAD4466HK said:
It is for an FX.
My son's FX 6300 will start to throttle at 66c, and my cousin's FX 8350 will BSoD if it goes over 70c (thermal throttling disabled).
Click to expand...

Really? Never owned one. Possible the tim is just junk. The silicon is probably surpassing 100c then and the thermal sensor is just picking up the package temp. Interesting.
 
yeah 70c is lock up time with the FX series. OP, id set everything manually as suggested for both the cpu and ram. and add a bit of voltage to the ram too, like .1v. youll also need to crank up the fans on the aio and get air moving over the vrms like mentioned above.
 
tangoseal said:
Really? Never owned one. Possible the tim is just junk. The silicon is probably surpassing 100c then and the thermal sensor is just picking up the package temp. Interesting.
Click to expand...


The FX's don't actually have thermal sensors, iirc. It's algorithm based.
 
  • Like
Reactions: Nobu
like this
ManofGod said:
Yes but, is that socket or core temp. Do not forget that they are both monitored but they are not the same temp though.
Click to expand...

I believe it is the core temp, but I could be wrong since it's been awhile.

Iirc, When it comes to monitoring temps. You don't use Core temp with an offset for FX cpu's as your load temps will be way off the true value. Use either HWinfo64, HWmonitor or Aida64 and they will all show 2 values, 1 for the socket and 1 for the core temp. The socket temp is what you should look to if you want an idea of idle temp's and the core temp for load temp's. The reason for this is as stated before AMD use some odd way of measuring the cpu temp and it's not accurate below 40-45c.

And I believe there is a 10-20 degree offset between the socket and the core. Once again iirc.
 
Last edited:
yeah there is usually 18-20c difference between cpu and socket, socket being the higher one. on my system afterburner reads my cpu temp correctly, just another option if you have it.
also, lost of people cool the backside of the socket with the highend or oc'd chips. I did, before my old board blew up.
 
pendragon1 said:
yeah there is usually 18-20c difference between cpu and socket, socket being the higher one. on my system afterburner reads my cpu temp correctly, just another option if you have it.
also, lost of people cool the backside of the socket with the highend or oc'd chips. I did, before my old board blew up.
Click to expand...
Thing is, both core and "cpu" temp are derived from the same sensor on FX cpus. One is the thermal junction temperature (rather, distance from tjmax) iirc, and the other is "core" (which is what AMD recommends using for fan profiles). Neither are very accurate except under load (when temp goes up).

https://help.argusmonitor.com/index.html?TemperaturemeasurementforAMDCPUs.html

there was a thread on the net somewhere that had an AMD employee (or someone in contact with one) explain this better, but I can't find it.
 
The FX9370/9590 was AMDs attempt to try and stay relevant in the processor market with their garbage Bulldozer architecture. They basically threw caution to the wind and went balls to the wall pushing the architecture to the limit without concern for power usage.

I've never come across an FX9000 series chip that worked out of the box. The samples I've dealt with had to be significantly underclocked or had core voltage bumps to get working at the stock settings.
 
You must log in or register to reply here.
Back
Top