Overclocking AMD FX-6300

Jul 13, 2018
8
0
10
Hello All, I am new to overclocking and new to this forum.

I was recently able to successfully overclock this processor to 4.3GHz. I think I did it correctly; incrementally increasing the CPU multiplier & voltage to get a successful PRIME95 stress test.

My concern is that I had to push the CPU voltage all the way up to 1.4925V in order to get a stable 4.3GHz...

I have read that it should be OK as long as voltage does not exceed 1.5V; but I'm right on the limit.

Are there any other "adjustments" in the BIOS that I could try in order to maybe maintain the 4.3GHz, but at a lower voltage?

Thanks in advance...Paul
 
Solution
As far as reading, 1.55V seems max voltage for FX-6300, but that includes safe temps for VRM's, and LLC is load line calibration, that means if you put 1.5V it will and have to manage stay close to the voltage that you set, that is recommended.
The temps for the FX is around 70*C, usually recommended staying under 70*C.
Sources:
https://www.overclock.net/forum/10-amd-cpus/1440068-fx-6300-max-core-temp.html
http://www.tomshardware.co.uk/answers/id-1958405/max-safe-voltage-6300.html

There is always risk when you have 4 VRM's (not sure how technically its called), its a bit risky cause you can burn them down if the chipset cooling isnt enough, the color will start to degrade around VRM's.
Well i also recently bought gift for my friend, an fx-4100 and m5a78 lx3, from what I understand, voltage should not be above 1.5V.
From what I've read for my chip, I managed 4.5GHz without changing voltage at all with AMD overdrive software (1.42V) and stays cool under 60*C with stock copper cooler.

Always what I do, load default, lower voltage until it starts to crash, then overclock without touching voltage till it doesnt book anymore, then lower clock by 100MHz from last unstable overclock, test if its stable, if yes add up 100mV, then repeat pushing until doesnt boot, if not stable add 100mV and test again. Now its repeat process until you got close to 1.5V and under.
I didnt had time for that to do, but i hope this would be helpfull to you.
 


1.4925 at 4.3G with a 6300 does sound like a lot. You may try using LLC using an offset voltage instead of a manual voltage. Try to use as low of an LLC setting as you can to limit overshoot as a load is removed.

A lot depends on how robust the VRM is on your mobo. For reference: I had a 6300 at 1.46V, 4.4G on a mobo with a sketchy VRM, an M5a88M.

Are you sure your stability issues aren't due to processor over heating? A hot processor requires more voltage to return to stability, so better cooling could allow a lower voltage.

Don't place to much emphasis on P95 stability as it's totally not a normal load pattern your system will see in use. It's not called a 'power virus' for nothing. You should feel free to lower the voltage once you've proven stability at the higher one in P95. Then run a test with a real-world load: like a long video encode using Handbrake or 6 or 7 quick passes in CineBench. If it passes with that then you should be good to go so long as temperatures are well controlled too.
 
Jul 13, 2018
8
0
10
Thanks Robert, I did my tuning via the BIOS, adjusting only the CPU multiplier and CPU voltage.

I tried to use AMD Easy Tune 6 tuning software and could not get anything to work using that.

open


 
Jul 13, 2018
8
0
10


DD

I was trying to post a graph of my temps but can't get the image to post... ??

Anyway, my CPU temp is 55c while under the P95 load test @ 4.3GHz/1.4925v. MB temp is 44c. I have a liquid cooler...

For LLC, it is currently set on "AUTO". My only other options in BIOS are "Extreme" and "Regular"...which should I do?

Yeah, I kinda suspected that P95 load test was overkill, but being new to this OC stuff everything I've read on-line says to use it...

I will try to back down the voltage and run something that is more real world and see what happens...

 


I'd try lowering voltage to 1.4 then set LLC to 'REGULAR'. On reboot monitor how VCore responds in HWMonitor or HWInfo64. It should pop up as you start P95, or ideally simply hold solid, but if it pops up then watch how far and adjust as necessary. It's a reboot-iterative process so be ready to wear out the delete key LOL

Extreme should do as it says: it should pop up a lot. But then, on lesser boards LLC acts weird and sometimes counter-intuitive so you have to do some experimenting to find out how it works on yours. Also, pay attention to offset voltage adjustment vs. manual voltage adjustment if you have a choice.

55C under P95 certainly sounds good, the MB temp means nothing since I have no idea what it's monitoring. It could just as easily be an unterminated sensor line on the monitoring chip and reporting nonsense. Ideally, it should be the VRM section but 44C under load there is simply to great to be real.

 
Jul 13, 2018
8
0
10


Yeah, not really sure either what the MB temp sensor is actually measuring ...when I installed the liquid cooling system I took the old CPU fan and placed it right over the AMD 760G heatsink on the MB though...figured it can't hurt.
 


I'd locate it to blow on the VRM section since that's where I've always had the biggest problems. Since you're liquid cooling there's very little air movement back there now.

I did that on my old AM3 (an M5a88m with 6300) and I'm doing it on my Ryzen board (B350M Mortar with 1700.)
 
One thing I wanted to add about P95....

I do think people put too much stress on P95 stable when all they do is gaming but there are use cases where it is probably as important as anything else. It's great to help you determined if you've achieved best DE-RATED performance. That is, not only is it performance optimized, i.e. overclocked, but you've also derated it so that you can be confident it has very high long-term stability too.

This is important if you want to do unattended processing that takes a long time to complete: like very large video encoding queues or long scene renderings. These are jobs that can last one (or more) days of continuous, high-level processing. If you're P95 stable for over two hours you'll have a high degree of confidence your machine won't crash out half way through a two day job of that kind of intensity. I may not crash out even if it's NOT P95 stable, but where time is money you want some assurances.
 
Jul 13, 2018
8
0
10
I set LLC to "regular" and dialed back the voltage to 1.4125...it booted fine...I could run Xplane11 (usage about 50% max CPU) fine for an hour+. CPU temp stayed <50
P95 stress test crashed immediately.

I tried to push the GHZ to 4.4 @ 1.4125v and would not boot. Going to continue to play around and see what I can get.

Understand completely your P95 logic, and will also try to move the extra fan over the VRM area
 
Jul 13, 2018
8
0
10
Just a quick update here...I've played around with some settings and here is where I have settled:

LLC=Extreme
OC=4.5GHz
CPU Volts=1.3925


Runs AIDA64 stress test 30 mins w/o any issues. I haven't tried P95, but likely that would fail.

CPU Temp is at 40C max during A64 stress test; (35C max if I remove the front cover of PC case...this is where my radiator is mounted)

 


AIDA64 Stress test was as hard as P95...at least if running with co-processor test and cache test but memory test off. That's pretty much running small FFT's that nest within processor cache, the same as P95 small FFT testing. Of course, Aida64 has to be FX aware to know how to load up the Bulldozer core caches. It was certainly Ryzen aware when I still had a trial license LOL.

At 1.392 volts you've certainly margin to bump it up some if you wanted: I've read where bit-miners would run vishera cores at 1.55 volts for months on end at 4.7-5.0Ghz for months on end no problem.
 
Jul 13, 2018
8
0
10


Wow...do you know what temps these guys were at with 1.55V/5.0GHz ?? I haven't tried anything higher than 4.5GHz yet, but maybe this weekend I'll do some more toying...
 


I didn't get that bit, but i'm certain all they would worry about is limit for stability. Hardware life wasn't a concern; just as many hashes as feasible in the shortest time

 
As far as reading, 1.55V seems max voltage for FX-6300, but that includes safe temps for VRM's, and LLC is load line calibration, that means if you put 1.5V it will and have to manage stay close to the voltage that you set, that is recommended.
The temps for the FX is around 70*C, usually recommended staying under 70*C.
Sources:
https://www.overclock.net/forum/10-amd-cpus/1440068-fx-6300-max-core-temp.html
http://www.tomshardware.co.uk/answers/id-1958405/max-safe-voltage-6300.html

There is always risk when you have 4 VRM's (not sure how technically its called), its a bit risky cause you can burn them down if the chipset cooling isnt enough, the color will start to degrade around VRM's.
 
Solution
Jul 13, 2018
8
0
10
So playing around with OC a little bit more this morning...

Got up to 4.8GHz but had to increase voltage to 1.4925 to get a stable run at that speed...

Using the AMD Overdrive stability test, CPU Thermal Margin readings were at 22C after 15 minutes. CPU temps in AIDA64 & HWinfo64 were both around 48C.

I guess doing the math using these above readings, max temp for the FX-6300 confirms the 70C number.

Maybe later I will try to see if I can get a temp reading off of my VRM's. I have heat sinks on them with fans mounted on top, so I'm thinking these temps are OK.

Edit: Temp on back of MOBO behind VRM is 45C after 15 minutes AMD Overdrive stability test. (Same as CPU temp with pc case back removed)