CPU Dropping MHz and voltage... WHYYYYY

[ExplodingTaters]

So I have an AMD FX 8350, and whilst playing Battlefield 4, Planetside 2, and sometimes in CS:GO, (Any Semi-Intensive or Intense game) I will drop frames, and having hardware monitor up on my second monitor, I see that I drop "CPU VCORE" Voltage from 1.325V to 0.888V and my clocks go down from 4.0 GHz to about 1.4 GHz, and sometimes even 233 MHz or higher for brief miliseconds, 233 just being hardware monitor's recorded lowest value ever. 90% of the time it stays at 4.0 GHz.

 

Anyways, I am not sure if this is some dumb power saving option, or if I need a new PSU or Motherboard. I am no BIOS genius, but I've tried to turn off every possible "Power Saving" option in UEFI BIOS, and set all "auto" clocks to manual standards instead of automatic.

 

Thanks in advance!

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

My Specs

 

CPU- AMD FX-8350, 8 Cores, 4.0GHz

Cooler - Corsair H60

MotherBoard- ASUS M5A97 LE R2.0 AM3+ Socket

RAM - 16GB (4x4GB) G-Skill Sniper Series 1600Mhz DDR3

GPU- MSI GTX 760

Storage- 2 WD Blue HDDs WD10EZEX 1TB 7200RPM 64MB Cache

PSU- Corsair CX600 600W Power Supply Semi-Modular

OS- Windows 10 Home Edition (64-Bit)

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

@ionbasa

[Oberon.Smite]

What are your temps? Use HW Monitor while stress testing or playing a game.

[Clyne]

Just disable the pos Turbocore program in the bios that sets voltage and clock speed based on usage and temps.

[QueenDemetria]

OP

What temp is your CPU when it does this? If the temp is fine(65c and below), then you might have to force your CPU to run at full speed(open CCC and go to the CPU power options, and drag the slider to 4Ghz).

[ExplodingTaters]

The highest temperature HW Monitor recorded Today was 51 C, in my CPU, not the socket Temp, nothing higher. @QueenDemetria

 

@Bubblewhale , AMD Turnbocore is disabled.

Edited September 3, 2015 by Godlygamer23
Corrected username tag.

[ionbasa]

So I have an AMD FX 8350, and whilst playing Battlefield 4, Planetside 2, and sometimes in CS:GO, (Any Semi-Intensive or Intense game) I will drop frames, and having hardware monitor up on my second monitor, I see that I drop "CPU VCORE" Voltage from 1.325V to 0.888V and my clocks go down from 4.0 GHz to about 1.4 GHz, and sometimes even 233 MHz or higher for brief miliseconds, 233 just being hardware monitor's recorded lowest value ever. 90% of the time it stays at 4.0 GHz.

 

Anyways, I am not sure if this is some dumb power saving option, or if I need a new PSU or Motherboard. I am no BIOS genius, but I've tried to turn off every possible "Power Saving" option in UEFI BIOS, and set all "auto" clocks to manual standards instead of automatic.

 

Thanks in advance!

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

My Specs

 

CPU- AMD FX-8350, 8 Cores, 4.0GHz

Cooler - Corsair H60

MotherBoard- ASUS M5A97 LE R2.0 AM3+ Socket

RAM - 16GB (4x4GB) G-Skill Sniper Series 1600Mhz DDR3

GPU- MSI GTX 760

Storage- 2 WD Blue HDDs WD10EZEX 1TB 7200RPM 64MB Cache

PSU- Corsair CX600 600W Power Supply Semi-Modular

OS- Windows 10 Home Edition (64-Bit)

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

@ionbasa

I'm not an expert on AMD platforms, but make sure that Turbo core and AMD quiet and cool is disabled. 

 

Maybe try a more aggressive Vdroop setting?

[Clyne]

The highest temperature HW Monitor recorded Today was 51 C, in my CPU, not the socket Temp, nothing higher. @QueemDemetria

 

@Bubblewhale , AMD Turnbocore is disabled.

Just set everything to normal, if everything is like that then just remove the CMOS battery and put it back it to reset to factory defaults. Make sure to disable AMD turbocore, leave voltage at manual, and leave everything on standard on the triangle thing.

I own the almost the same board and with those settings, everything stays at the same voltage and clocks.

[GM_Peka]

Could try going to to windows power plan settings and set min and max processor states to 100%, just to see if that makes any difference.

[SkilledRebuilds]

What are your VRM temps like... (The heatsink under the CPU socket usually is the part I'm talking about)

Don't know what the sensor may be called...

 

Is the VRM heatsink bleeding hot? (dont outright touch it, but does the air around it feel hot)

Even with COLD CPU temps, hot VRM temps can induce throttling under load (4.0Ghz>1.4Ghz!!!)

If hotter than you'd like, aim a fan directly at the heatsink and watch your CPU clocks under load.

 

/Could be something else, but this is not unheard of before, even on similar boards (ie: friend has the same, its fine), some boards can be just weird.

[ExplodingTaters]

@SkilledRebuilds It's very hot, and has been for a year now, and never ran into any problems, and I am not sure how to monitor that temp...

@Pvt. 8Ball Minimum processor state is now at 100%

@Bubblewhale I will also set voltages to manual In BIOS

@ionbasa No prob thanks though

 

Thank you all! Ill run some more intensive programs, and see the results.

[ExplodingTaters]

Okay every voltage is on manual

[ExplodingTaters]

Nope after everything while playing intensive games she still drops down to 0.888 V and ~1.4 GHz

 

The CPU Socket temp hasn't gone above 57 C, the CPU itself hasn't gone above 44 C, and the "Mainboard" Temp hasn't gone above 38 C. Temps aren't a problem here, except for maybe the VRM Temps or something @SkilledRebuilds, I just don't know how to monitor that...

[ionbasa]

 

Nope after everything while playing intensive games she still drops down to 0.888 V and ~1.4 GHz

 

The CPU Socket temp hasn't gone above 57 C, the CPU itself hasn't gone above 44 C, and the "Mainboard" Temp hasn't gone above 38 C. Temps aren't a problem here, except for maybe the VRM Temps or something @SkilledRebuilds, I just don't know how to monitor that...

 

That motherboard rates the maximum CPU wattage as 140 Watts. The 8350 has a stock TDP of 125W. I know TDP == electrical Power. But it's close enough for estimation purposes (And ASUS uses the same metric on their support page for the motherboard CPU list).

 

Running the CPU at 100% brings you pretty close to maxing out the VRMs on the motherboard , either that or you're cutting it close to the total power the motherboard can deliver to the CPU.

 

Also looking at Asus' CPU support documentation here: https://www.asus.com/us/Motherboards/M5A97_LE_R20/HelpDesk_CPU/

The 8350 was officially supported with BIOS 1006. Make sure you have that BIOS or a newer one installed. Maybe try updating to the latest BIOS and re-testing.

[ExplodingTaters]

Here is what system info tells me, @ionbasa

 

BIOS Version/Date               American Megatrends Inc. 1503, 1/11/2013

 

I think its just time to a buy a new mobo... You think?

[GzeroD]

Disable APM in the bios and the chip will no longer downclock to save power. it will obviously fun a noticeably warmer from doing this. (It is under advanced bios features in your BIOS)

[QueenDemetria]

Disable APM in the bios and the chip will no longer downclock to save power. it will obviously fun a noticeably warmer from doing this. (It is under advanced bios features in your BIOS)

Yes, do this ^^^ I'm not sure what it stands for (I do but jokes time) but I'm sure it stands for "Automatic Problem Mechanism". I've had it cause random problems for me, including random throttling. @ExplodingTaters You can try forcing the CPU frequency to the max using the CPU control in Catalyst Control Center(and if you don't have CCC then this might be your problem). Disable APM first though, it needs to be disabled.

 

And thanks @Godlygamer23 for the tag

[SkilledRebuilds]

-SNIP-

 

I think its just time to a buy a new mobo... You think?

Could "for a short while, to test throttling differences" , put 1-2 more fans in the machine, directing airflow towards the VRM heatsink (aim between GPU/CPU) onto the heatsink there. (Strap em in anyway you can, zipties are easy, make sure of their airflow direction, and mount aimed towards Vrm block.)

If it can bring it down some degrees, maybe it wont throttle the CPU down to 1.4Ghz as much or at all hopefully.

(I thought APM means Auto Power Management, like Intels Speedstep, idle = downclock) (Which you can still keep on btw, when Apps/Games need it they DO rise)

[ExplodingTaters]

Okay I had turned off "master mode APM" or something in BIOS and I played some BF4, no luck, my voltages still dropped, and my CPU MHz dropped too... I think @ionbasa was right, my mobo might be struggling...

[ionbasa]

Okay I had turned off "master mode APM" or something in BIOS and I played some BF4, no luck, my voltages still dropped, and my CPU MHz dropped too... I think @ionbasa was right, my mobo might be struggling...

You can always try updating to the latest BIOS and see if Asus ay have patched something. The last few BIOSes for that board mention increasing system stability. Might as well try it and see the outcome.

[GM_Peka]

Try maybe using optimized defaults, then dial all your OC settings in again.

[ExplodingTaters]

I can try updating bios sometime, but for now I'll just downclock from 4 GHz to like 3.6 GHz or something so I'm not killing my mobo in the meantime.

I've also already tried resetting to defaults and all that jazz

[Skander1345]

I can try updating bios sometime, but for now I'll just downclock from 4 GHz to like 3.6 GHz or something so I'm not killing my mobo in the meantime.

I've also already tried resetting to defaults and all that jazz

Try the fan over VRM suggestion and see if that helps. If it does change the mobo.

[ionbasa]

Try the fan over VRM suggestion and see if that helps. If it does change the mobo.

Probably won't help. The VRM on that mobo is rated up to 140W. He's got an 125W chip in there.  Which means the VRMs are only doing about ~90% of their duty cycle. The fact that they aren't even near their specced output W (VA) means their dying due to a defect during manufacturing.

 

Cooling might not help much in this case (if at all) if the issue is due to a faulty VRM implementation. Yes, the VRMs will run cooler, but the issue will most likely still exist.

[GzeroD]

Probably won't help. The VRM on that mobo is rated up to 140W. He's got an 125W chip in there.  Which means the VRMs are only doing about ~90% of their duty cycle. The fact that they aren't even near their specced output W (VA) means their dying due to a defect during manufacturing.

 

Cooling might not help much in this case (if at all) if the issue is due to a faulty VRM implementation. Yes, the VRMs will run cooler, but the issue will most likely still exist.

He is using a AIO cooler, Many boards are designed to have a downdraft cooler installed. with little to no air flow a lot of lower end boards fold pretty quickly.

[ionbasa]

He is using a AIO cooler, Many boards are designed to have a downdraft cooler installed. with little to no air flow a lot of lower end boards fold pretty quickly.

There should still be incidental airflow in the case from the radiator fans. Most VRMs do fine even without cooling. Most of the heat will just be dissipated by natural convection. Most VRMs are rated to about 125C. Most MOSFETs run in the 80-90s and cooling rarely makes a noticeable impact. Even people who have waterblocks on their VRMs seem to get in the low 80s.