Question AMD FX 6300 CPU voltage ?

davexnet

Distinguished
Jan 14, 2008
15
0
18,510
Hello -
I have an FX-6300 on an Asrock 960gc-gs. I'm trying to use a manual voltage,
but what seems strange to me is that, according to HWmonitor, the voltage drawn is less when the CPU
is under load and higher when at rest. I thought it should be the other way round - for example, this article
describes the lower power P-states -
https://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-6300.html

Not sure if it's a limitation of my budget motherboard, or I'm not understanding something.

I have AMD Cool and Quiet and C-states active in the BIOS (default settings)

At the moment (based on my manual voltage setting of 1.2v in the BIOS,
I see 1.19v at rest at 1.10v under load (encoding video @ 95% cpu utilization)

Thanks for any info
 
according to HWmonitor,
What does HWInfo tell you?

Not sure if it's a limitation of my budget motherboard
That would be my understanding as well, due to the lack of a proper VRM design...coupled with the fact that your platform isn't exactly new hence the VRM's might be heating up and inducing lower power output by the VRM.

This would be a worthwhile read;

To add, even if your motherboard states support for 95W processor parts, that VRM area would state otherwise.
 
Hello -
I have an FX-6300 on an Asrock 960gc-gs. I'm trying to use a manual voltage,
but what seems strange to me is that, according to HWmonitor, the voltage drawn is less when the CPU
is under load and higher when at rest. I thought it should be the other way round - for example, this article
describes the lower power P-states -
https://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-6300.html

Not sure if it's a limitation of my budget motherboard, or I'm not understanding something.

I have AMD Cool and Quiet and C-states active in the BIOS (default settings)

At the moment (based on my manual voltage setting of 1.2v in the BIOS,
I see 1.19v at rest at 1.10v under load (encoding video @ 95% cpu utilization)

Thanks for any info
That is usually by design on processors.
For a detailed explanation look up "V Droop"
https://www.google.com/search?client=firefox-b-1-d&q="V+Droop"
Now upper mid and higher end motherboards have an option for LLC or line load calibration to compensate for overclocking and keeping voltages low and stable.

You do not start with a higher voltage knowing it will droop under heavy load. You can lower voltage and LLC will keep it from drooping much depending on level you set.
 
Hello -
I have an FX-6300 on an Asrock 960gc-gs. I'm trying to use a manual voltage,
but what seems strange to me is that, according to HWmonitor, the voltage drawn is less when the CPU
is under load and higher when at rest. I thought it should be the other way round - for example, this article
describes the lower power P-states -
https://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-6300.html

Not sure if it's a limitation of my budget motherboard, or I'm not understanding something.

I have AMD Cool and Quiet and C-states active in the BIOS (default settings)

At the moment (based on my manual voltage setting of 1.2v in the BIOS,
I see 1.19v at rest at 1.10v under load (encoding video @ 95% cpu utilization)

Thanks for any info

id like to point out that the asrock boards from am3 beyond were always a bit flakey due to not having great vrms or heatsinks on they used to always recomend using a blow down cooler design for better performance.
 

Lutfij,​

You're right, HWinfo gives completely different info, see these two images that show approx.
0.9v at rest and 1.5v under load.

Unolocogringo,​

thanks for the info I didn't know about that.

IN general, the CPU heats up too quickly and the PC shuts down (5 minutes encoding video),
with the CPU voltage/mulitiplier set to auto.
But setting the voltage manually (same as I did with an earlier FX 4300 in this same motherboard)
allow the cooler to keep things steady, and within the "safe" range.
I have another motherboard (similar vintage) a Gigabyte GA-78LMT-USB3 (rev 6.0)
sitting in a box, I've never tried it, but I believe it may be better.



 
Last edited:
Hello -
I have an FX-6300 on an Asrock 960gc-gs. I'm trying to use a manual voltage,
but what seems strange to me is that, according to HWmonitor, the voltage drawn is less when the CPU
is under load and higher when at rest. I thought it should be the other way round - for example, this article
describes the lower power P-states -
https://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-6300.html

Not sure if it's a limitation of my budget motherboard, or I'm not understanding something.

I have AMD Cool and Quiet and C-states active in the BIOS (default settings)

At the moment (based on my manual voltage setting of 1.2v in the BIOS,
I see 1.19v at rest at 1.10v under load (encoding video @ 95% cpu utilization)

Thanks for any info
It is most likely as you suspect: the limitation of your budget board. When put under load an inexpensive voltage regulator can't do what it's main task is: regulate the voltage to the CPU. So it's sagging, or going down. The more power-hungry the CPU (6 core 6300's were very power hungry back in the day) the more it goes down.

Better boards implement something called LLC or Load-Line Calibration which would pump the voltage up a bit as load increases in a bid to keep it level. But that can create it's own set of problems like extremely short voltage spikes during load-off transients which might feed the CPU an extremely high voltage and kill it over time. This was typically done by overclockers who knew what they were about and used it with discretion.
 
IN general, the CPU heats up too quickly and the PC shuts down (5 minutes encoding video),
That implies you've applied too much manual voltage, but see my next comment.

I am indeed using the stock cooler with the "blow down" fan
I tend to fit bigger coolers to most of my CPUs and leave stock coolers in the box. Sometimes it's complete overkill, but with a 95W or higher TDP, I feel happier with bigger cooling.

I have an old AM3 3.4GHz AMD 965 Phenom II X4 and 3.2GHz 955, three years older than your FX6300. The Phenoms crash at just over +60°C. As a result, they're difficult to overclock even with big heatsinks. My 3800X (NH-D14) and 7950X (NH-D15) reach thermal throttling at +95°C, but don't crash.

I see a figure of +70.5°C mentioned for the 6300 in CPU World, but I'm not sure if this is max CPU die temperature or max socket temperature. Either way, you might get a few more MHz out of the 6300 with a bigger heatsink.
https://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-6300.html

N.B. There seems to be some concern about inaccuracies of socket temperatures in discussions. I'm not sure if the 6300 has a built in sensor on the Silicon die, or relies on an external temperature diode under the socket.
https://www.techpowerup.com/forums/threads/fx-6300-oc-temps.190250/

It's worth noting that many motherboards apply far too many volts to CPUs (both Intel and AMD), especially when overclocking. They're catering for especially bad CPUs which need a bit more voltage to remain stable.

From your screen shots, I see you had 1402.9MHz at 0.825V when idle and 3507.2MHz at 1.55V, presumably with all 4 cores running near 100%. This sort of voltage range is quite normal. The BIOS dynamically boosts and reduces the voltage, as the CPU freqency rises and falls.

The 6300 should automatically boost 2 cores to 3800MHz or 1 core to 4100MHz, but at significantly lower voltages than the 1.55V shown in your sceen shot, namely 1.425V and 1.4125V. BIOS differences may account for this.

Frequency ? 3.5 GHz / 3500 MHz
Maximum turbo frequency
4.1 GHz / 4100 MHz
Boosted P states^M#1: 4100 MHz, 1.425V
#2: 3800 MHz, 1.4125V

I'm trying to work out where your 1.55V ties in with your manual 1.20V setting. Perhaps your screen shots were taken with the BIOS set to Auto?

Do take extra care when applying manual voltage. Exceed the safe level by 0.3V and you may get magic smoke and a dead CPU.

From this discussion on Tom's I see someone was applying circa 1.5V whilst trying for 4.8GHz. This is what I term pushing the limit, unless you're cooling the CPU with Liquid Nitrogen or Helium. You stand the very real risk of destroying your CPU with too high a manual overvolt.

1.20V seems a bit low for a stable all-core overclock at 4.00GHz, but if your CPU is overheating after only 5 minutes of video coding, something needs to change. Either lower the voltage or increase the cooling.

I suggest a big Socket AM3 air cooler, or a mid-range AIO, if you can find one to fit such an old processor.

I have no idea how good this really cheap 6-heatpipe cooler is, but it's the sort of thing I fit when trying to extract the last watt of heat using big air. Overkill for the 6300 maybe, but it comes with that all important AM3 fitting hardware. Might not fit in your case though. Check the clearance (height). Make sure you get the AM3 version.
https://www.amazon.com/upHere-Cooler-Pipes-1600rmp-compatible/dp/B0CRDX9F1V

At the other end of the scale is this twin heatpipe cooler, rated at only 95W, so not good enough for an overclocked 6300.
https://www.amazon.com/SilverStone-Technology-Profile-Cooling-RL-KR01/dp/B07GN1C3G5

I suspect your standard AM3 cooler is failing to cope with the overclocked 6300. Buy the biggest cooler you can fit in the case. My AM3 Phenoms are using obsolete (single tower, single fan) Noctua air coolers, but they're not cheap.

I'd suggest a bare minimum of 4 heatpipes, preferably 6. It doesn't always guarantee better cooling, but more heatpipes, two big towers and two 120mm/140mm fans are often better than one small tower + 90mm fan.
 
Miscar - thanks for bring up some interesting points. I wanted to clear something up -
when I said the PC crashes (CPU too hot) after 5 minutes of video encoding, it happens when the
voltage in the BIOS is set to AUTO

HWinfo shows the Vcore in realtime, with recorded min and max.
With this CPU set to auto voltage (no overclocking) I get a rang of 1.32v (load) - 1.25 v(rest)

if I set the voltage to manual, the "suggested" voltage is 1.15v. Using this gives a range of
approx 1.10 - 1.15 (load vs. rest). This is a big diference to the vcore when AUTO is used

Those other voltages mention earlier seem to be fixed p-state voltages,
as can be seen in the report from CPU-Z:
P-State FID 0x19 - VID 0x11 - IDD 17 (20.50x - 1.337 V)
P-State FID 0x16 - VID 0x18 - IDD 17 (19.00x - 1.250 V)
P-State FID 0x13 - VID 0x20 - IDD 12 (17.50x - 1.150 V)
P-State FID 0xE - VID 0x26 - IDD 9 (15.00x - 1.075 V)
P-State FID 0x9 - VID 0x2C - IDD 7 (12.50x - 1.000 V)
P-State FID 0x4 - VID 0x32 - IDD 6 (10.00x - 0.925 V)
P-State FID 0x10C - VID 0x3A - IDD 4 (7.00x - 0.825 V)

I've got no idea how it ties into changing the vcore in the BIOS

EDIT according to the HWinfo, Core VIDS shows the cpu's requested voltage,
(corresponds to the P-states) not voltage really supplied by the voltage regulator

For now I set the voltage manully to 1.15v, the suggested value in the BIOS, and I'll
see how it goes. I noticed with my video encoding test it seems stable and temps are
well below 60C

drea.drechsler - thanks for your extra info

 
Last edited:
I just picked up the CPU from eBay for $15
With any second hand device, you've no idea of its condition and with such old hardware, electromigration may have set in, due to overclocking or just general use. I overclocked an Athlon II X4 620 from 2.8MHz to 3.25GHz for 6 years. After that it became unstable so I disabled the overclock.

If you decide to continue using the FX-6300. see if you can find a large heatsink/fan on eBay, but postage on heavy items can bring the price up to that of a new cooler on Amazon. Make sure it fits AM3.

When my video editing rig crashed randomly (after several hours transcoding) I "fixed" it by limiting GPU power to 95% in MSI Afterburner. The CPU is running with stock BIOS settings (no PBO). The GPU seems perfectly stable in Furmark, but crashes in Topaz Video AI. Might just be an Nvidia quirk.

If your FX-6300 voltage "fix" works and you're happy with the performance, there's not much more to do.