No temp change after overclock

gravipack · Aug 5, 2016

Hi, I have overclocked my FX 8350 to 4.5ghz and stress tested it and had the same max temp and stable temp when I stress tested it at 4.0ghz. The temps both only hit a max of 58C, then started to settle and lower a bit ~49-58C. Should I see a temp difference or is my overclock not working? Idle I get around 25-35C and sometimes spikes up to 40C then back down quickly. CPU-Z does say I am running @4.5ghz and so does HWmonitor. I put my CPU ratio to 22.5 and CPU vcore to 1.4375.

FX 8350 @4.5
Gigabyte 970a-D3P
Corsair H60
16GB @1333mhz

techgeek · Aug 5, 2016

I have found most temp monitoring software is unreliable for AMD's FX CPU's. Especially at idle. The best way to monitor is use Overdrive and look at the Thermal Margin. This tells you how far away you are from throttling. It works in reverse to absolute temperature, the higher the number, the cooler the CPU is. As the Thermal Margin approaches 0, the closer it is to throttling.

As for whether it's working or not, first determine that you weren't throttling at load under both scenarios. If you weren't / aren't, then benchmark the CPU to see if the performance scales with the overclock. This will tell you if there is any difference in performance. Often it takes a synthetic benchmark to highlight the difference sufficiently for it to be seen / felt.

Lastly temps only change slightly with clockspeed, it's not until you start adding voltage that you see noticeable increases in temperature. So if you didn't need to raise Vcore to go from 4.0 to 4.5GHz, then you might not see much increase in temps.

gravipack · Aug 5, 2016

techgeek :

Hi, thanks for responding so quickly. I have looked in AMD overdrive and my thermal margins seem to be where they were before.(~35-45C) I have benchmarked in Unigine Valley before and after i overclocked and it did have a noticeable difference. These were the results (if you dont mind clicking a link)
https://gyazo.com/48af169c2e5764fc268b883a9b313033
https://gyazo.com/721481c7bbcc329698773d3e2652afca
https://gyazo.com/247573c258c16e9cf1dadd00bbb4baa2

I feel comfortable at this speed and temp so would I be able to overclock more? If so do you have a recommendation on what a good vcore would be at what ghz? Or should i settle here?

maxwellmelon · Aug 5, 2016

your temps also might not change much depending on how aggressive your fan speed profile is.

(while overclocking I always set my fans to run at full speed the whole time that way I know how much room temp wise i have to increase the clock, because your cooling is one of the biggest factors in limiting overclocking, yes i know its not the only but its usually the main liminiting factor) and in case the system crashes its allows it to cool the system off good before i reset the power.

techgeek · Aug 5, 2016

I am actually quite surprised that FX CPU bottlenecked your graphics card that much. When I clicked the first link and seen it was Unigine, I thought to myself, that's not going to be any good, the graphics card will be the bottleneck. However I was surprised that there was a definite difference. I was thinking something like Cinebench CPU benchmark would be better.

As for what voltage is good, that is really dependent on the CPU. A general rule of thumb is the lower the stock voltage is at stock speeds, the better it will overclock as ultimately it will take less voltage to overclock, This isn't a hard and fast rule, but it's generally applicable. I think the general consensus on the FX 8350 is that 1.5V is the max safe voltage, some push that to 1.55V. Of course safe voltage is relative or tied to the temp. You may hit a thermal wall before you get to 1.5V depending on how efficient your cooler is and how good your CPU is.

When overclocking, never add voltage until you become unstable. So raise your multiplier until you lose stability, then raise your Vcore until you regain stability. You never want to use more voltage than you need for any given core clock frequency.

techgeek · Aug 5, 2016

Looking at your benchmarks again, there appears to be more going on there. Normally a given percentage overclock results in a smaller percentage increase in benchmark scores. Particularly in this case since the benchmark in question is more of a graphics benchmark rather than a CPU benchmark. In your case going from 4.0GHz to 4.5GHz is an overclock of 12.5% whereas your Unigine benches went from 2501 to 2917 which is a 16.66% improvement. This is backwards from what I expect. In most cases you are lucky to get close to a 1 to 1 improvement, though this is very rare. Often one or more of the systems other subsystems will drag the score down from a 1 to 1. So in your case I would have expected something like a 10% increase in Unigine with a 16.66% overclock.

Did you overclock your graphics card in here as well?

gravipack · Aug 5, 2016

techgeek :

My GPU is overclocked from 1279mhz to 1350mhz in all three of the tests. Also, how would i determine when I become unstable to know I need to increase the vcore? Would I stress test to see if I crash or something goes wrong?

techgeek · Aug 5, 2016

In all overclocking, when you make a change of any sort be it core clock speed, voltage or both, you need to follow it up with a stress test to verify stability. So for a CPU overclock, you'd raise your multiplier and then boot the OS and stress test with something like Prime 95 Small FFT's. This will push the CPU beyond any normal usage and if it remains stable under those conditions then it "should" be stable under normal usage. Stress testing also gives you a worse case scenario for temperatures.

A similar approach is performed with a GPU, after any increase in core clock, mem clock or both, you should run a video stress test like Unigine. If it's stable in that, then it "should" be stable in games.

Instability would be anything that results in an unpredictable behavior. For CPU's it could be BSOD's, reboots, applications crash to the desktop, an error in the application, rounding errors in Prime 95 specifically. For GPU's it could be a driver crash / recover, games crash to desktop, artifacts, game errors, reboots, BSODs.

gravipack · Aug 5, 2016

techgeek :

In all overclocking, when you make a change of any sort be it core clock speed, voltage or both, you need to follow it up with a stress test to verify stability. So for a CPU overclock, you'd raise your multiplier and then boot the OS and stress test with something like Prime 95 Small FFT's. This will push the CPU beyond any normal usage and if it remains stable under those conditions then it "should" be stable under normal usage. Stress testing also gives you a worse case scenario for temperatures.

A similar approach is performed with a GPU, after any increase in core clock, mem clock or both, you should run a video stress test like Unigine. If it's stable in that, then it "should" be stable in games.

Instability would be anything that results in an unpredictable behavior. For CPU's it could be BSOD's, applications crash to the desktop, an error in the application, rounding errors in Prime 95 specifically. For GPU's it could be a driver crash / recover, games crash to desktop, artifacts, game errors, reboots, BSODs.

Thanks for all this help and info! Learned a couple of new things!

maxwellmelon · Aug 5, 2016

I just ran the benchmark got a very close score to you (2601). nearly same fps. I have a r290x card which is considered very close to the gtx980. running stock freq 8350.

my 8350 ran about 50% load on four cores the whole test, other four cores stayed at idle(yes I understand its due to them not being full cores). while gpu was maxed at 100% the whole test but that's not my point.

my point is I don't think that benchmark is anywhere near fully loading the cpu. that is why your temp is not changing much (or even going down as your cpu is getting done its end faster and is able to drop into lower power state for longer ms of time).

id be amazed if your gtx 980 is that much faster then the amd r290x that it is able to fully load the cpu when the r290x cannot.

I think you need to run like prime 95 and then see what temps you get before you push your cpu more.

techgeek · Aug 5, 2016

gravipack :

No problem. Just a word of caution, when running Prime95 keep a close eye on your temps until they plateau. You'll want to stop Prime95 if the temps get out of hand. If you are aircooled, you should probably plateau in a couple of minutes. After that it should stay pretty much the same temp.

Now for a quick dirty benchmark to see if it's making a difference run Cinebench on the CPU. There is a GPU portion, but this isn't what you'll be interested in when overclocking your CPU.

techgeek · Aug 6, 2016

Let us know how you make out. Just remember with overclocking, go in small steps, don't rush it. Make sure to stability test between each change. If you hit a thermal wall, then stop, back the overclock down some. The only way to break past a thermal barrier is to improve cooling performance. Even if you have the thermal headroom, I wouldn't use more than 1.5V for Vcore for a 24/7 overclock. Voltage is the CPU killer, temperature is the secondary affect of voltage.

Search

No temp change after overclock

gravipack

Commendable

techgeek

techgeek

Splendid

gravipack

Commendable

maxwellmelon

Splendid

techgeek

Splendid

techgeek

Splendid

gravipack

Commendable

techgeek

Splendid

gravipack

Commendable

maxwellmelon

Splendid

techgeek

Splendid

techgeek

Splendid

TRENDING THREADS

Latest posts

Moderators online

Share this page