Topic Title: AMD FX-8350 hardware failure
Topic Summary:
Created On: 02/27/2014 08:37 AM
Status: Post and Reply
Linear : Threading : Single : Branch
<< 1 2 3 Previous Next Last unread
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 03/05/2014 02:24 PM
User is offline View Users Profile Print this message

Author Icon
QB the Slayer
Master Looter

Posts: 329
Joined: 11/27/2012

The CPU temp while under load is 56 not the 77...  I am thinking the 77 is the socket temp which is a bit on the high side of things.  Don't forget, OCCT uses HWMonitor as the built in sensor (so readings will be the same).  Although it doesn't hurt to run that app as well and look at the graphs like this:

 

Also our OP has an Asrock 990FX Extreme 4, not a gigabyte, that's me

 

Has the 1/4 LLC setting helped your stability?  Have you been able to run Prime95 with that setting?  What are your new load temps with that setting (BTW, it's the AMD FX-8350 Package temperature you want to look at).

 

QB



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

The MONSTER HTPC:

CPU: AMD FX-8350.||.Cooler: Corsair H80i
MB: Gigabyte 990FXA-UD7.||.RAM: 8 GB Mushkin Blackline DDR3 2000MHz (7-10-8-27-1T)
Case: CoolerMaster HAF 932.||.PSU: Corsair HX750
GPU:Asus R9 270X DirectCU II TOP.||.Audio: Creative X-Fi Titanium Fatal1ty Pro w/ Logitech Z-5300e (5.1, 280W-RMS)
System Drive: 2xSamsung 840 Pro 128GB RAID0.||.Working Drive: 2xMushkin Chronos 60GB RAID0

 03/05/2014 04:12 PM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

Yes I am aware that OCCT uses a version of HWMonitor and that it has had some issues over the years with certain mobos and sensors.

56C for package temp may not be true core temp. 77C for the thermister is also very high. Core temps of 56C are still quite close to the reliable thermal limit depending on what he's using for a HSF. It's very possible if the CPU core temp is spiking with load and that's causing his system to freeze.

The AM3+ vcore VRM issues exist on many mobos. That's why I like to see the OCCT graphs for the core temps and vcore voltages under load as the first two parameters when a working system suddenly has issues. Gigabyte has had a lot of VRM issues but so has Asus and others. Asrock specifically recommends the 990FX Fatality series for the FX-8350. The Extreme 4 is probably borderline but it should work even if the VRM runs a little on the hot side. 

Your vcore dropping as much as it does is proof of a bad VRM design for sure. For two years Gigabyte claimed there were no problems even when we made them run P95 stress tests to see the VRM overheating and vcore issues.



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.

 03/05/2014 04:37 PM
User is offline View Users Profile Print this message

Author Icon
QB the Slayer
Master Looter

Posts: 329
Joined: 11/27/2012

For sure the socket temp is very high and even the package temp... I would like to know more about the OP's total system cooling (case, fan sizes & placement, CPU cooler). Today's systems run very hot even at default settings.

I also noticed that the OP seems to have Turbo or APM enabled (x20.5 multi in CPU-Z).  It might be worth while to disable all CPU power features (APM, C1E, Cool-n-Quiet, etc etc) while we look at getting him stable.

As for my board...  yeah it's a revision 1.0, the VRM is actually quite robust, but it just lacks the LLC that the FX line seems to demand, which Gigabyte did address in future revisions of the UD7...  just wish I had waited to get one of those.  But it all worked out, I have very good cooling and can afford to use the v-core settings need to be stable.

 

QB



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

The MONSTER HTPC:

CPU: AMD FX-8350.||.Cooler: Corsair H80i
MB: Gigabyte 990FXA-UD7.||.RAM: 8 GB Mushkin Blackline DDR3 2000MHz (7-10-8-27-1T)
Case: CoolerMaster HAF 932.||.PSU: Corsair HX750
GPU:Asus R9 270X DirectCU II TOP.||.Audio: Creative X-Fi Titanium Fatal1ty Pro w/ Logitech Z-5300e (5.1, 280W-RMS)
System Drive: 2xSamsung 840 Pro 128GB RAID0.||.Working Drive: 2xMushkin Chronos 60GB RAID0

 03/06/2014 11:34 AM
User is offline View Users Profile Print this message

Author Icon
TeddyBeer
Peon

Posts: 12
Joined: 02/27/2014

Originally posted by: AMDforMe The HWMonitor data does not look right, which is not uncommon.

 

If your CPU is running 75-77C it's WAY OVER-HEATED as 61C is the max for your CPU. That's why I wanted to see what OCCT shows for vcore under load and the true core temps via Core Temp or OCCT. OCCT is useful as it records the data while you run stress tests.

 

As far as the vcore voltage it will change all over the place on "auto". In manual it should stay close to what you manually set it. You may have several issues.

 

LLC is suppose to maintain the vcore voltage closer to what you manually set it or the default vcore max when run in "auto" mode under stress loads. Typically one of the four LLC options will hold the vcore closest to what you manually set it but Gigabyte has had issues with the VRM circuits on their AM3+ mobos being insufficient to power 8-core FX processors so there is no telling what the results will be.

 

The OCCT core temp and vcore graphs would be very helpful in troubleshooting your PC problem. If you haven't cleaned your CPU HSF in a year it could be filled with dust and causing the CPU to overheat. Your CPU fan speed is not particularly high either, which may be a BIOS setting?

 

 

Hello,

I ran OCCT test now for 5 minutes and i see a Vcore drop when i start tests i not large drop 0,2 volt but here is the graph:

Here is the temperature graph:

 

 

 

 

 

Do you think my CPU VCORE setting should be 1.4 ~ 1.5 of something like that? and i get max temp 67 over 5 min do you also think my CPU is getting overheated?

Thanks for help so far.

Greetz, Sander

 03/06/2014 11:44 AM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

The vcore drop is excessive but the problem looks to be the CPU is overheating. You need to keep the core temp at or below 61C under the heaviest load you run your PC. Have you compared Core Temp to see if the CPUTIN from OCCT and Core Temp are close in value - under stress testing?

Have you checked the HSF to see if it might be filled with dust or loose on the CPU? That could be your problem. The fan speed may not be at 100% either which could be a BIOS adjustment for fan speed?

After you get the core temp down where it belongs you should try adjusting LLC to whatever setting keeps the vcore closest to the default 1.376v setting under heavy load like OCCT or P95 stress testing. From your CPU-Z post it appears that the default vcore for your CPU is 1.376v which would be pretty typical. You should not need to increase it unless you are OC'ing, as long as the mobo VRM circuit can maintain the voltage fairly close to 1.376v under heavy load. That is what the LLC adjustment is for so you'll need to test to see what setting works best under load.



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.



Edited: 03/06/2014 at 12:24 PM by AMDforMe
 03/06/2014 03:32 PM
User is offline View Users Profile Print this message

Author Icon
TeddyBeer
Peon

Posts: 12
Joined: 02/27/2014

Just cleaned the whole PC form dust. did some testing again got better temp results now.

TEMP Graph

 

 

 

But had to clear the BIOS Setting after clean up because of software failure on boot but i already fixed it. But anyway the LLC is on auto now so got really unstable Vcore voltage this is Auto LLC setting:

So will do the setting to 1/4 and run test on new and post back with new graphs on right setting.

 

 03/06/2014 09:19 PM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

The cleaning was certainly worth the effort!



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.

 03/07/2014 01:22 AM
User is offline View Users Profile Print this message

Author Icon
QB the Slayer
Master Looter

Posts: 329
Joined: 11/27/2012

Indeed cleaning helped and so does the 1/4 LLC...  and please post the CPU Package temp...  not the CPUTIN (I am more than certain that's not your core temp)...  it's useful, but core temp is more important.

 

And DEFAULT voltage for an FX-8350 is 1.325 V...  CPU-z just shows the applied voltage (AMDforME really should be calling that voltage the "Auto" voltage).  If you want to find the VID (default v-core that the CPU requests) run Core Temp like this:

 

 

Anyway...  let us know how the 1/4 LLC affects your stability.  There may be some other issues to tweak, to get you all the way there if it's needed

 

QB



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

The MONSTER HTPC:

CPU: AMD FX-8350.||.Cooler: Corsair H80i
MB: Gigabyte 990FXA-UD7.||.RAM: 8 GB Mushkin Blackline DDR3 2000MHz (7-10-8-27-1T)
Case: CoolerMaster HAF 932.||.PSU: Corsair HX750
GPU:Asus R9 270X DirectCU II TOP.||.Audio: Creative X-Fi Titanium Fatal1ty Pro w/ Logitech Z-5300e (5.1, 280W-RMS)
System Drive: 2xSamsung 840 Pro 128GB RAID0.||.Working Drive: 2xMushkin Chronos 60GB RAID0

 03/07/2014 10:31 AM
User is offline View Users Profile Print this message

Author Icon
TeddyBeer
Peon

Posts: 12
Joined: 02/27/2014

Hi Guys,

I did a lot of testing today and got this on the last test. put the LLC on 1/4 and i put the voltage to 1.4375 in BIOS. I have the most stable Vcore ive ever had but temperature is too high i think.

 

 

 

 

I have the Coolermaster V8 wich is an 180 Watt cooling solution. So i dont think my CPU cooler is the problem.

 

Greetz,

Sander

 

EDIT:

This is what HW monitor says druing test:



Edited: 03/07/2014 at 10:37 AM by TeddyBeer
 03/07/2014 10:31 AM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

No 1.325v is NOT the definitive default vcore for the FX-83xx models. There is no longer one set vcore for a given model CPU as in the past. The default vcore varies over a range by the CPU as programmed by AMD. His 1.376v CPU-Z value is very typical of the vcore used on FX-83xx CPUs.

I'm trying to get an official explanation from HWMonitor exactly what they are measuing for temp with CPUTIN and Package as the terms they are using are obviously misleading. Even if "Package" is the core temp, 60C is right at the thermal limit of 61C as specified by AMD. The CPUTIN temps before and after HSF cleaning would suggest the package temp was above the AMD specified 61C before the cleaning.

Mr. Beer - Unless you need to run the vcore that high for stability under load , I would suggest lowering it to the default 1.376v indicated for your CPU and then running some stress tests to check for stability. Use whatever LLC setting holds the vcore closest to the preset vcore.

I agree the temp is too high. The increased vcore will drive the core temp and thermister temps high even with a good cooler so you want to use the lowest vcore possible. I'd see if the system is stable with the lower vcore. The "package" temp should decrease also.



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.



Edited: 03/07/2014 at 11:01 AM by AMDforMe
 03/07/2014 10:53 AM
User is offline View Users Profile Print this message

Author Icon
TeddyBeer
Peon

Posts: 12
Joined: 02/27/2014

Originally posted by: AMDforMe No 1.325v is NOT the definitive default vcore for the FX-8350. The default vcore varies over a range by the CPU as programmed by AMD. His 1.376v CPU-Z value is very typical of the vcore used on FX-8350 CPUs.

 

I'm trying to get an official explanation from HWMonitor exactly what they are measuing for temp with CPUTIN and Package as the terms they are using are obviously misleading. Even if "Package" is the core temp, 60C is right at the thermal limit of 61C as specified by AMD. The CPUTIN temps before and after HSF cleaning would suggest the package temp was above the AMD specified 61C before the cleaning.

 

Mr. Beer - Unless you need to run the vcore that high for stability under load , I would suggest lowering it to the default 1.376v indicated for your CPU and then running some stress tests to check for stability. Use whatever LLC setting holds the vcore closest to the preset vcore.

 

I agree the temp is too high. The increased vcore will drive the core temp and thermister temps high even with a good cooler so you want to use the lowest vcore possible. I'd see if the system is stable with the lower vcore. The "package" temp should decrease also.

 

Thanks for fast reply,

I will set the Voltage back to default and test again. ill post results over 15 minutes.

Greetz,

Sander

 03/07/2014 11:02 AM
User is offline View Users Profile Print this message

Author Icon
QB the Slayer
Master Looter

Posts: 329
Joined: 11/27/2012

I am pretty sure the 61C is the AMD recommended limit, but the CPU will go all the way to 71C or higher, before shutting down...  I would not recommend going that high either, and since Mr. Beer has yet to show any screen shots of the package going over 60C, I think his board was doing so wonky stuff when the socket was going over 60.  It may have been the board's attempt to cool...  just guessing at this stage though.

 

QB



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

The MONSTER HTPC:

CPU: AMD FX-8350.||.Cooler: Corsair H80i
MB: Gigabyte 990FXA-UD7.||.RAM: 8 GB Mushkin Blackline DDR3 2000MHz (7-10-8-27-1T)
Case: CoolerMaster HAF 932.||.PSU: Corsair HX750
GPU:Asus R9 270X DirectCU II TOP.||.Audio: Creative X-Fi Titanium Fatal1ty Pro w/ Logitech Z-5300e (5.1, 280W-RMS)
System Drive: 2xSamsung 840 Pro 128GB RAID0.||.Working Drive: 2xMushkin Chronos 60GB RAID0

 03/07/2014 11:20 AM
User is offline View Users Profile Print this message

Author Icon
TeddyBeer
Peon

Posts: 12
Joined: 02/27/2014

Got the new results. I think that the Vcore is a little bit unstable now but the temps are good now.

 

 03/07/2014 01:00 PM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

QB-

Actually the FX-8000 series start throttling the vcore and frequency @ precisely 70C core temp. I've confirmed this in testing. Errors or freezes can occur in the ~65C range IME.

As far as the mobo itself it doesn't do anything based on the CPU temp but if the VRM is running hot it too can throttle the CPU vcore and frequency typically dropping to .875v and 1400 MHz. for a few seconds then back to full vcore then back to .875v and 1400 MHz. constantly as the VRM circuit tries to keep from burning up.

And to make sure that you are clear on CPU-Z and applied voltage... When the system is in "auto" mode and you place a stress load on it like OCCT , P95, Linpack, etc., the BIOS defaults to the P0 power state which by definition is the vcore (max), i.e. default vcore voltage for the CPU, which is internally written for each CPU. You may want to read the AMD white papers on power states if you still don't understand.

In addition in recent years with multiple power planes being used in CPUs, there is no longer one specific vcore voltage for a given model CPU/APU. AMD uses a range of vcore voltages for their CPUs/APUs even of the same model number. I believe Intel has adopted this technique also.

Mr. Beer-

As you can see the CPUTIN and Package temps dropped dramatically when you lowered the vcore. That's a good sign.

I'm not sure how you are judging "system stability" but I'd run P95 or OCCT for awhile longer to see if all works fine. Then I'd try whatever games or other apps caused you problems. We don't know if the Win 7 install had anything to do with the system hanging or not. Hopefully cleaning and adjusting LLC will fix the issue.

Below are some other manual BIOS adjustments that you might try if you actually need them for improved stability. If everything is cool, I'd just use your PC as is and enjoy. Raising the voltages below will slightly increase core temps. If you need to raise the vcore to gain stability, only increase it the least amount necessary and you should be OK on core temp.

CPU-NB = ~1.3v

HT = 1.25v

RAM - increase .5v above default for RAM with a 1.5v or higher default. Do not exceed 1.7v

Make sure fan speed setting in BIOS is set to 100% if you have this option



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.



Edited: 03/07/2014 at 01:14 PM by AMDforMe
 03/07/2014 01:53 PM
User is offline View Users Profile Print this message

Author Icon
TeddyBeer
Peon

Posts: 12
Joined: 02/27/2014

Originally posted by: AMDforMe QB-

 

Actually the FX-8000 series start throttling the vcore and frequency @ precisely 70C core temp. I've confirmed this in testing. Errors or freezes can occur in the ~65C range IME.

 

As far as the mobo itself it doesn't do anything based on the CPU temp but if the VRM is running hot it too can throttle the CPU vcore and frequency typically dropping to .875v and 1400 MHz. for a few seconds then back to full vcore then back to .875v and 1400 MHz. constantly as the VRM circuit tries to keep from burning up.

 

And to make sure that you are clear on CPU-Z and applied voltage... When the system is in "auto" mode and you place a stress load on it like OCCT , P95, Linpack, etc., the BIOS defaults to the P0 power state which by definition is the vcore (max), i.e. default vcore voltage for the CPU, which is internally written for each CPU. You may want to read the AMD white papers on power states if you still don't understand.

 

In addition in recent years with multiple power planes being used in CPUs, there is no longer one specific vcore voltage for a given model CPU/APU. AMD uses a range of vcore voltages for their CPUs/APUs even of the same model number. I believe Intel has adopted this technique also.

 

Mr. Beer-

 

As you can see the CPUTIN and Package temps dropped dramatically when you lowered the vcore. That's a good sign.

 

I'm not sure how you are judging "system stability" but I'd run P95 or OCCT for awhile longer to see if all works fine. Then I'd try whatever games or other apps caused you problems. We don't know if the Win 7 install had anything to do with the system hanging or not. Hopefully cleaning and adjusting LLC will fix the issue.

 

Below are some other manual BIOS adjustments that you might try if you actually need them for improved stability. If everything is cool, I'd just use your PC as is and enjoy. Raising the voltages below will slightly increase core temps. If you need to raise the vcore to gain stability, only increase it the least amount necessary and you should be OK on core temp.

 

CPU-NB = ~1.3v

 

HT = 1.25v

 

RAM - increase .5v above default for RAM with a 1.5v or higher default. Do not exceed 1.7v

 

Make sure fan speed setting in BIOS is set to 100% if you have this option

 

Hello,

I think my system is a lot more stable then it was before. I have learned a lot of you guys about system stability. A few months ago it was the issue that my PC freezes sometimes and got errors on stress testing. That all is fixed now so i think i gonna leave the settings just like this and thank you guys for you time and thinking.

Greetz, Sander (_TeddyBeer_)

 03/07/2014 07:03 PM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

Enjoy and remember to clean the HSF every 3-6 months.



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.

 03/07/2014 11:16 PM
User is offline View Users Profile Print this message

Author Icon
QB the Slayer
Master Looter

Posts: 329
Joined: 11/27/2012

"And to make sure that you are clear on CPU-Z and applied voltage... When the system is in "auto" mode and you place a stress load on it like OCCT , P95, Linpack, etc., the BIOS defaults to the P0 power state which by definition is the vcore (max), i.e. default vcore voltage for the CPU, which is internally written for each CPU. You may want to read the AMD white papers on power states if you still don't understand."

 

I totally understand...  I just have a gripe with your terminology and what CPU-Z reports for voltage.  Using your terminology and logic, my default v-core is 1.424 V...  which I can guarantee you it's not.  And the 1.424V is not even what is set in the BIOS...  so for you to state that's a default voltage is not the case.  On "Auto" the BIOS is supposed to target 1.325 (VID) as the P0 state...  whether or not the motherboard is capable of achieving this goal depends on the board.  CPU-Z does not report the voltage that is set in the bios....  it reports the voltage being applied at any given time...  so to say it's a default voltage couldn't be further from the truth.

 

Mr. Beer....  glad things are better for you....  feel free to come back if other problems crop up or similar problems persist.

 

QB



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

The MONSTER HTPC:

CPU: AMD FX-8350.||.Cooler: Corsair H80i
MB: Gigabyte 990FXA-UD7.||.RAM: 8 GB Mushkin Blackline DDR3 2000MHz (7-10-8-27-1T)
Case: CoolerMaster HAF 932.||.PSU: Corsair HX750
GPU:Asus R9 270X DirectCU II TOP.||.Audio: Creative X-Fi Titanium Fatal1ty Pro w/ Logitech Z-5300e (5.1, 280W-RMS)
System Drive: 2xSamsung 840 Pro 128GB RAID0.||.Working Drive: 2xMushkin Chronos 60GB RAID0

 03/08/2014 12:02 AM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

QB-

As I explained, when the system is in "Auto" mode and you apply a stress load the BIOS attempts to set the vcore to the programmed default P0 voltage. It doesn't pick a vcore out of thin air, it uses the default vcore that is programmed into the CPU and that is what CPU-Z displays. The applied vcore voltage under this specific instance is the default vcore and nothing else. There is nothing wrong with my terminology. 

Some but not all BIOS also show the default vcore when you are in manual mode so you know how far you are deviating from the internally programmed default vcore.



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.

 03/08/2014 11:49 AM
User is offline View Users Profile Print this message

Author Icon
AMDforMe
Overclocker

Posts: 614
Joined: 09/08/2013

QB- UPDATE

I think I understand now why you are confused about CPU-Z and the default vcore.

This is how a properly functioning mobo works:

When you power up a PC in "auto" mode the BIOS reads the embedded CPU information such as model I.D., vcore, frequency, RAM frequency, etc. and the BIOS sets the vcore to whatever is specified in the specific CPU data. Whenever the system is subjected to a high stress load while in "auto" mode the BIOS will again set the vcore to the P0 power state which is precisely the vcore max voltage for that specific CPU.

In your case you are working with a mobo that is known to have a defective VRM design that is not capable of providing sufficient current to the FX-8xxx series CPUs under full load. In fact almost all Gigabyte AM3+ mobos suffer from a VRM design that is insufficient to power the 8-core FX CPUs, other than the very latest revised GA-990FXA-UD7, UD5 and possibly the UD3 models, I believe.

If your system is running in "auto" vcore mode and it's supplying 1.424v vcore then there are a couple possibilities:

1. The mobo is defective and over-volting

2. The vcore sensor is bad and providing bad data

3. Gigabyte has written a non-AMD compliant BIOS to over-volt the CPU to compensate for their deficient VRM circuit design

The point here being that a properly operating mobo/BIOS does exactly what I have been telling you all along - The BIOS sets the vcore voltage to the embedded CPU specified voltage. It doesn't pick some voltage out of thin air. Any mobo that doesn't provide the precise voltage that AMD has programmed into the CPU for the P0 power state, in "auto" mode when under stress test loads is defective in design, engineering, manufacture or operation.

I would not be surprised at all that Gigabyte has provided a hack BIOS mod to over-volt the CPU to compensate for the known deficiency in their VRM designs on the AM3+ mobos. Otherwise they could be responsible for warrantying hundreds of thousands of defective AM3+ mobos.

Over-volting the CPU to compensate for a deficient VRM design is unscrupulous IMO and not healthy for the CPU as it forces the CPU to run hotter and can shorten the CPU's life. Technically AMD could even decline warranty if the CPU fails prematurely as a result of over-volting from an improper mobo or BIOS. 

So now you know the whole story. Your defective Gigabyte mobo may have a non-compliant BIOS that over-volts the CPU and that's why on your mobo CPU-Z doesn't in fact show the proper vcore as it would on a properly functioning mobo. Thus I will qualify my prior comment to compensate for Gigabyte's product defects and potentially non-AMD conforming BIOS.

NEW statement on CPU-Z

If you have a properly functioning mobo and properly written BIOS, when your PC is in "auto" mode and a stress load is applied to the system, the BIOS will set the vcore voltage to the P0 state which is the max vcore voltage for the specific CPU installed. If however you have a defective mobo or a non-AMD compliant BIOS, there is no telling what the BIOS will supply for vcore voltage.

I'd strongly suggest that people return defective mobos and not use non-AMD compliant BIOS that may be hacked to over-volt the CPU to compensate for a VRM deficiency.

BTW, I hope we were able to resolve Teddybeer's PC issues and that he has no further issues. Regular cleaning of the CPU/GPU HSFs and the case and PSU fans can go a long ways towards preventing overheating related issues. Obviously cleaning won't compensate for a defective mobo VRM circuit or bad BIOS.



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

Building a reliable PC involves more than just assembling the parts. You need to be able to configure all of the BIOS settings appropriately. This can be quite involved and frustrating as it can require a lot of trial and error with stress testing. It is however often the only means to get a 100% reliable PC.



Edited: 03/08/2014 at 02:42 PM by AMDforMe
 03/08/2014 05:50 PM
User is offline View Users Profile Print this message

Author Icon
Vegan
Elite

Posts: 1233
Joined: 01/31/2010

+1 for cleaning the the rig.

my carbide 300r has air filters which seem to be material in reducing dust in the CPU fan and GPU coolers

using air purifiers in the gaming room goes far to keeping a rig clean too

 



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

AMD Support and Game » AMD Processors (CPU) » AMD FX-8350 hardware failure

<< 1 2 3 Previous Next Last unread
Topic Tools Topic Tools
Statistics
85513 users are registered to the AMD Support and Game forum.
There are currently 6 users logged in.

FuseTalk Hosting Executive Plan v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.