Sign in with
Sign up | Sign in
Your question

5GHz for AMD FX-8320

Tags:
  • Overclocking
  • Cooler Master
  • AMD
Last response: in Overclocking
Share
30 September 2014 18:09:12

Hi,

This is more of "is it possible" rather than "I need to do it" situation. Specs listed below:

PSU: Cooler Master V850
Mobo: ASRock 990FX Extreme 9
CPU: AMD FX-8320
Cooler: Phanteks PH-TC14PE
RAM: AMD Radeon R9 Gamer Series (2400MHz)
GPU: MSI GTX770 Lightning

Raising the multi to hit 5GHz@1.43V in the UEFI I can boot into Windows fine and dandy. However 3/4 of the way through the "Assessing processor" stage of the Windows Experience Index it froze causing me to reboot my PC. I would just like to point out I know he WEI is naff and I don't use it as a benchmarking software. I was only curious to see if the numbers would change.

During the WEI test the CPU reached as high as 70C before freezing. I've since then lowered the multi and Vcore to 4.7GHz@1.38V and I hit around 67C during the WEI and this time around it completed fine. I know and understand that doesn't mean it's stable, I need to properly test it to be sure. To me though it means it passed the "I'm not going to break at any given moment" test.

But my question being, how far away am I to being stable at 5GHz? I've searched around and found multiple threads on this matter which all state 5GHz on an 8320 is bonkers and highly unattainable, not to mention "should only be tried on extreme water cooling solutions". So I wouldn't be surprised at the responses I am going to get.

More about : 5ghz amd 8320

30 September 2014 18:11:15

You will need some kind of liquid cooling to hit that speed.
m
0
l
30 September 2014 18:12:00

It's the heat that's causing it to crash...
m
0
l
Related resources
30 September 2014 18:12:42

And you can get Air Coolers that can do it, like the Phanteks TC14-pe
m
0
l
30 September 2014 18:15:38

Shain Taylor said:
It's the heat that's causing it to crash...


I'm pretty sure CPU's only shut down from overheating when the temps start getting really dangerous. For example, I had a laptop with an AMD Athlon X2 processor hitting 92C and that never shut down due to heat. I hardly think 70C is dangerous for a CPU

EDIT: I have that cooler..
m
0
l
30 September 2014 18:43:21

You will probably need better cooling, but try this first:

Turn up the voltage to 1.45 or 1.4725 and put your LLC to 50%. Obviously monitor temps under stress test, but I found the LLC turned all the way off DRASTICALLY raised temps on my overclocks, and at 50% I was able to raise the voltage significantly and it made my overclock stable. May not work, but its worth a try.
m
0
l
30 September 2014 18:48:35

CptBarbossa said:
You will probably need better cooling, but try this first:

Turn up the voltage to 1.45 or 1.4725 and put your LLC to 50%. Obviously monitor temps under stress test, but I found the LLC turned all the way off DRASTICALLY raised temps on my overclocks, and at 50% I was able to raise the voltage significantly and it made my overclock stable. May not work, but its worth a try.


Ooo that's impressive and a nice trick, it might work seeing as I have my LLC set to "off" too. I'll give it a try, cheers!
m
0
l
30 September 2014 19:29:09

CptBarbossa said:
You will probably need better cooling, but try this first:

Turn up the voltage to 1.45 or 1.4725 and put your LLC to 50%. Obviously monitor temps under stress test, but I found the LLC turned all the way off DRASTICALLY raised temps on my overclocks, and at 50% I was able to raise the voltage significantly and it made my overclock stable. May not work, but its worth a try.


Ahh stupid me I forgot the LLC settings on ASRock boards were reversed. I set the Vcore to 1.45V and changed the LLC Setting to 50%. This caused a BSOD at the welcome screen and setting to 75% BSOD's after the Loading Windows screen. With 25% and "Disabled" I can get into Windows fine, although 25% still fails the "I'm not going to break at any given moment" (WEI) test.
m
0
l
30 September 2014 19:42:09

Another thing you can try is overclocking you northbridge. I don't know what they are set at by default on the fx chips, but try raising it by 200mhz and reboot.

And yet another thing you can try is loosening up your RAM timing (this also helped me).

Chances are you need better cooling and just more Volts, but these other things can also help with stabilizng your overclock. However, if you are not comfortable with overclocking your NB or messing with RAM timing then you need to stick with core multiplier and voltage adjustments.
m
0
l
30 September 2014 19:56:39

CptBarbossa said:
Another thing you can try is overclocking you northbridge. I don't know what they are set at by default on the fx chips, but try raising it by 200mhz and reboot.

And yet another thing you can try is loosening up your RAM timing (this also helped me).

Chances are you need better cooling and just more Volts, but these other things can also help with stabilizng your overclock. However, if you are not comfortable with overclocking your NB or messing with RAM timing then you need to stick with core multiplier and voltage adjustments.


Increasing my NB Frequency from stock (2200MHz) to 2400MHz causes Windows to boot up without Windows Aero enabled and no matter what I can't enable it. I don't know why that is, but probably down to memory or something like that. I'm quite competent to change the RAM timings, however I set my RAM to load the AMP profile for 2100MHz so I have just left it at that. I did change the RAM frequency to 1866 but it doesn't seem to have made a difference.

EDIT: Increasing CPU-NB voltage from 1.8V to 2.2V has fixed the loading in Windows Basic issue.
m
0
l
30 September 2014 20:02:31

It would seam that either RAM or voltage are still your bottleneck then, and more volts means more heat.

One last thing I would try is putting the LLC back to 50% and crank the voltage up pretty high (1.475-1.5). Obviously you will have to keep a hawks eye on temps and make sure to stop any stress testing if it gets too high, but that is an option.

When I did my overclock setting the LLC to 50% bought me 5-7c, just to give you an idea.
m
0
l
1 October 2014 13:15:09

CptBarbossa said:
It would seam that either RAM or voltage are still your bottleneck then, and more volts means more heat.

One last thing I would try is putting the LLC back to 50% and crank the voltage up pretty high (1.475-1.5). Obviously you will have to keep a hawks eye on temps and make sure to stop any stress testing if it gets too high, but that is an option.

When I did my overclock setting the LLC to 50% bought me 5-7c, just to give you an idea.


I gave up for the night but I'll give it a go when I got home tonight. Not sure I want to go to 1.5V, but I'll give it a go between 1.45-1.47 and see what happens. One thing to note is that the 70C is what HWMonitor and Speedfan are reporting which I have been told is the socket temp.

http://postimg.org/image/mxp811kpb/full/

Which is highest, the socket or core temps? I would assume the core temps is going to be higher than the socket temps, so if 70C is the socket, the core is going to be about 75-80?

I've been advised to use the Thermal Margin reported by AMD Overdrive so I'll give though that a go tonight. Out of sheer determination I want to get stable at 5GHz, but I'm quite happy running stable at 4.5-4.7GHz haha
m
0
l
1 October 2014 15:07:54

I use HWmonitor and core temp when testing my system.

I understand your desire to hit 5ghz. Just keep in mind it may not be possible. I had an a10 6800k that I couldn't get past 4.829 even with water cooling, Even at 1.49v.

The more I think about it the more I am convinced you need more voltage. I think 1.5v is fine (but obviously don't do it if you arent comfortable with it). Just keep in mind you may not reach your goal.
m
0
l
a c 428 K Overclocking
a c 457 À AMD
1 October 2014 15:36:40

The CPU can throttle around 70 degrees and shut down.
m
0
l
1 October 2014 19:13:28

SR-71 Blackbird said:
The CPU can throttle around 70 degrees and shut down.


Yeah, I've just learned this in another thread.

Quote:
It's the heat that's causing it to crash...


I apologise to you Shain Taylor as you were right, I just didn't believe it to be true because of past experiences.

I learned the art of reading the thermal margin in AMD Overdrive and I've finally past a WEI test at 5GHz. Sadly the numbers didn't change though. I find it odd how it's not changed from a 7.8 rating going from 3.5GHz to 5.0GHz, but that's Windows for you haha.

Anyway, during WEI it will only push the CPU as far as 90% during it's "Assessing CPU" phase, but it allows me to get a 'feeler' of what to expect without pushing it too hard.

To pass I used a Vcore of 1.475V and a LLC of 25%. The socket temp reported by Speedfan reached 72C at it's highest and a thermal margin of 2.2C. So you can see, even at 90% usage this isn't stable at all. Still not clear which way around this LLC is supposed to go. I know what LLC does - it either adds Vcore or takes it away depending on the setting. But can't seem to make any sense of it so far.

I noticed the Vcore hit a peak of 1.53V during the WEI test, bearing in mind the I only set this to 1.475 with 25% LLC. Using the same Vcore but changed to 50% it BSOD's upon entering the welcome screen. That suggest to me the higher the percentage the more it's taking away and this is proven when upping the Vcore on 50% LLC it will actually boot into Windows fine. But then that contradicts what I see reported by Speedfan earlier.

I'm still trying to experiment and try different things, so if anyone has any suggestions, please shout up!
m
0
l
a c 428 K Overclocking
a c 457 À AMD
1 October 2014 19:31:57

Thermal instability.
m
0
l
1 October 2014 21:01:09

Ro-Tang Clan said:
SR-71 Blackbird said:
The CPU can throttle around 70 degrees and shut down.


Yeah, I've just learned this in another thread.

Quote:
It's the heat that's causing it to crash...


I apologise to you Shain Taylor as you were right, I just didn't believe it to be true because of past experiences.

I learned the art of reading the thermal margin in AMD Overdrive and I've finally past a WEI test at 5GHz. Sadly the numbers didn't change though. I find it odd how it's not changed from a 7.8 rating going from 3.5GHz to 5.0GHz, but that's Windows for you haha.

Anyway, during WEI it will only push the CPU as far as 90% during it's "Assessing CPU" phase, but it allows me to get a 'feeler' of what to expect without pushing it too hard.

To pass I used a Vcore of 1.475V and a LLC of 25%. The socket temp reported by Speedfan reached 72C at it's highest and a thermal margin of 2.2C. So you can see, even at 90% usage this isn't stable at all. Still not clear which way around this LLC is supposed to go. I know what LLC does - it either adds Vcore or takes it away depending on the setting. But can't seem to make any sense of it so far.

I noticed the Vcore hit a peak of 1.53V during the WEI test, bearing in mind the I only set this to 1.475 with 25% LLC. Using the same Vcore but changed to 50% it BSOD's upon entering the welcome screen. That suggest to me the higher the percentage the more it's taking away and this is proven when upping the Vcore on 50% LLC it will actually boot into Windows fine. But then that contradicts what I see reported by Speedfan earlier.

I'm still trying to experiment and try different things, so if anyone has any suggestions, please shout up!


The WEI only goes up to 7.8. Also, it is partly true. 8320's are made to be 8320's because they aren't as capable as 8350's... So to get the same result as an 8350 on an 8320, it will need more volts to complete it, if it can; and at the same time increasing heat... However on 8350's, 4.7Ghz is the sweet spot for performance as any increased clocks before that, the gains will become smalller and smaller... Personally I would not reccommend going any higher than 65'c under 100% load to pro-long the life of your CPU. But it's a personal preference...
m
0
l
1 October 2014 22:19:03

I think 1.55v under load is fine as long as your temps stay in good range. At this point I think you can reach 5ghz stable if you can get sufficient cooling and you up the voltage to 1.5.

I never set my chips above that because when I put them under load HWMonitor indicates voltage increases up to .o5v above what your bios set them to. I don't know if HWMonitor is being accurate or not, but I don't want to risk it any higher than 1.55v. I know some people will say 1.55v is too high, but I have also had people say you can go 1.65 (I never did). It is boiling down to your lack of cooling. You could try it out with your case open and maybe a small tabletop fan to blow cooler air into the case, but this is obviously not a long term solution.
m
0
l
4 October 2014 11:58:25

CptBarbossa said:
I think 1.55v under load is fine as long as your temps stay in good range. At this point I think you can reach 5ghz stable if you can get sufficient cooling and you up the voltage to 1.5.

I never set my chips above that because when I put them under load HWMonitor indicates voltage increases up to .o5v above what your bios set them to. I don't know if HWMonitor is being accurate or not, but I don't want to risk it any higher than 1.55v. I know some people will say 1.55v is too high, but I have also had people say you can go 1.65 (I never did). It is boiling down to your lack of cooling. You could try it out with your case open and maybe a small tabletop fan to blow cooler air into the case, but this is obviously not a long term solution.


Right, first things first, many thanks for helping me through this, you've been a massive help in trying to get me to achieve this. I was honestly expecting people to be like "you can't achieve this, don't even bother". I thought now I've got the Phanteks PH-TC14 cooler I'll give it a go seeing as this cooler is capable of performing on par with some AIO liquid coolers. To be honest, I think 5GHz is capable on this chip, but you really do need high end cooling to pull it off, which I don't have.

I finally figured out the LLC on ASRock boards by looking in the BIOS (why didn't I do that the first time, duh?) and I noticed with a Vcore of 1.38V and the LLC set to "disabled" my actual Vcore was hovering around 1.43V at idle and at load going to 1.48V.

Setting LLC to 50% gave me an idle Vcore pretty much on par with what I set it to in the BIOS. In this instance it was 1.400V and at load it only peaked at 1.44V. So higher percentage= lower Vcore for LLC on this board. With the same settings, I'm able to be stable at 4.7GHz and gaming in Farcry 3 for an hour has only seen temps rise to a max of 56C.

Oh and I should mention that during the time of my last post and this post, I repasted my CPU and removed the middle HDD cage on my case to allow better airflow from the front two 120mm fans

To sum it up then, it seems like I can comfortably hit 4.7GHz, but to stretch for an extra 300MHz you have to put a hell of a lot more volts into it causing the temps to rise significantly and in my case the temps are the bottleneck here. Unless you really really need the extra 300MHz, in my experience I would advise not to push for 5GHz on an 8320. Many thanks to you all for helping me!
m
0
l