Overclocking my FX-8320 on a Hyper 212 Evo

[TheRongGuy]

So I've been a bit confused about a couple of aspects of overclocking. I have my FX-8320 running at 4.1 GHz and 1.27 V, and load temps in the mid-high 50's after 1 hour of Blend on Prime95. I haven't been able to push 4.2 GHz, because that's not stable at even 1.3 V, and with those volts my temps go up to 70. For me these temperatures seem a bit high, because other people seem to be able to pump more volts through the same CPU and get similar temps with the same cooler... Have I botched the installation? 

 

Also, what is the best way to test stability and max temperature? As of now I've been using Prime95 with the "blend" test, because I observed that I could run small FFT's for hours on end but get errors in the blend test after 10 minutes. Is Prime95 good for measuring the max temps? I've heard people recommend Intel Burn Test, OCCT, LinX, AIDA64 etc.

 

Oh and I'm talking about core temps, not socket temps. Just to clear that out.

[Moredeath]

It might be possible that you just didn't get a perfect overclocker. It's luck of the draw.

[Lord Pantaloons]

The silicon lottery is likely to blame though you could try remounting the cooler, CM coolers have lower mounting pressure so more TIM is required to get good contact than other coolers.

[orangecat]

Welcome to the silicon lottery.

[TheRongGuy]

Interesting, I thought the silicon lottery meant that CPU's differ in that some overclock better than others, but it also means that CPU's can produce different amounts of heat at the same volts? 

[Bagmup]

So I've been a bit confused about a couple of aspects of overclocking. I have my FX-8320 running at 4.1 GHz and 1.27 V, and load temps in the mid-high 50's after 1 hour of Blend on Prime95. I haven't been able to push 4.2 GHz, because that's not stable at even 1.3 V, and with those volts my temps go up to 70. For me these temperatures seem a bit high, because other people seem to be able to pump more volts through the same CPU and get similar temps with the same cooler... Have I botched the installation? 

 

Also, what is the best way to test stability and max temperature? As of now I've been using Prime95 with the "blend" test, because I observed that I could run small FFT's for hours on end but get errors in the blend test after 10 minutes. Is Prime95 good for measuring the max temps? I've heard people recommend Intel Burn Test, OCCT, LinX, AIDA64 etc.

 

Oh and I'm talking about core temps, not socket temps. Just to clear that out.

What's your motherboard? I had high temps with my 8350 @ 4.6Ghz in an Asus 99fx pro 2.0 board.

I ended up ghetto mounting a stock cpu cooler fan on the back of the mobo to cool the socket which also lowered the core temps.

[Khiol]

I ended up ghetto mounting a stock cpu cooler fan on the back of the mobo to cool the socket which also lowered the core temps.

 

Dude you have pics? or can you explain your process? I am interested in doing this myself.

[TheRongGuy]

What's your motherboard? I had high temps with my 8350 @ 4.6Ghz in an Asus 99fx pro 2.0 board.

I ended up ghetto mounting a stock cpu cooler fan on the back of the mobo to cool the socket which also lowered the core temps.

I have a M5A97 EVO R2.0. So it might be caused by shitty case airflow? Maybe I should try running some tests with the sidepanel off.

[Bagmup]

Dude you have pics? or can you explain your process? I am interested in doing this myself.

I thought I had pics on my phone, must of deleted them..

Basically just find the strongest double sided tape you can and stick it onto the cooler bracket or directly to the mobo. My case had a bulge built into the side panel, so it went back on as well. I think there's a thread on overclockers.net showing it.

[Priller]

Remount CPU cooler. Give it plenty of air flow.

 

You will most likely need 1.35-1.37 volts to get to 4.3.

 

Also check your LLC is set correctly. Too much Vdroop will cause instability. You may set 1.4 volts but will only get 1.30 under load as an example.

[Bagmup]

I have a M5A97 EVO R2.0. So it might be caused by shitty case airflow? Maybe I should try running some tests with the sidepanel off.

I think it's more of an Asus motherboard problem tbh, what program are you using to get your temps? Amd is notorious for being inaccurate when it comes to reporting correct temps.

[TheRongGuy]

I think it's more of an Asus motherboard problem tbh, what program are you using to get your temps? Amd is notorious for being inaccurate when it comes to reporting correct temps.

I'm using HWMonitor, from what I've heard it's the most reliable one.

[Priller]

I'm using HWMonitor, from what I've heard it's the most reliable one.

 

In my experience the motherboard manufacturer's software and oddly enough OCCT are the most accurate for me.

[Faa]

Also, what is the best way to test stability and max temperature? As of now I've been using Prime95 with the "blend" test, because I observed that I could run small FFT's for hours on end but get errors in the blend test after 10 minutes. Is Prime95 good for measuring the max temps? I've heard people recommend Intel Burn Test, OCCT, LinX, AIDA64 etc.

If the vrm overheats, the safety feature just kicks in and causes the cpu to downclock just to keep the vrm temps below the max so you're not going to crash when it overheats. When the vrm doesnt deliver enough power; the PC just shuts complety off just like when the power supply' OCP kicks in.

That you can pass small fft's without any problems and crashing in blend pretty much means up the vcore, small fft's just singles the cpu out doesnt use any ram so neither the memory controller where as blend does. Also it differs from cpu to cpu, some cpu's crash much faster in Intelburntest like 2500K's for example but for 3930K's its complety the other way prime causing them much faster to crash.

 

[i_build_nanosuits]

Man i run a 1.435V all day long on my FX-8320 and hyper 212 i can reach 4.7ghz stable with that and my temps while playing BF4 mutliplayer are like 47c...don't worry push that voltage and ramp up the speed you can go as high as 1.5v on the core

if you want to and remember no games or applications will reach temps as high as prime 95 does...also don't run prime for hours, 15 minutes and if no core fails you're good to go...

[TheRongGuy]

Man i run a 1.435V all day long on my FX-8320 and hyper 212 i can reach 4.7ghz stable with that and my temps while playing BF4 mutliplayer are like 47c...don't worry push that voltage and ramp up the speed you can go as high as 1.5v on the core

if you want to and remember no games or applications will reach temps as high as prime 95 does...also don't run prime for hours, 15 minutes and if no core fails you're good to go...

I'm beginning to lean towards me botching the installation. Just tried to bump both LLC to high and volts to 1.3, had to stop it after 2 min on Prime95 blend because it was going to hit TJ max. Or is it possible that a chip after a certain voltage just goes crazy and pumps out insane amounts of heat?

[TechFan@ic]

I have had the exact same experience as you with Prime95, Blend testing spits out more errors primarily due to RAM instability.
Make sure your RAM is installed securely. Most of the time it's very deceptive you think you had the RAM crammed all the way in but then give it a push and goes in even further.
Second thing you want to do is go into the BIOS and input the latencies for your RAM manually.

Each chip is different. My particular 8320 is happy to run at 4.4Ghz with 1.38v. and 4.6Ghz with 1.425v.

You will never run into any workload that stresses your CPU as much as Prime95 does. So if you just want to make sure that you're applying enough voltage for each particular clock speed you can run prime95 and then only designate 6 threads to the torture test rather than 8.

Will keep your temperatures under manageable levels for a sufficient amount of time to tell if you have enough voltage for the particular clock speed you're testing.
As a general rule of thumb you want to keep your core temps below 61c on the FX chips, once the temperature crosses that threshold the CPU will begin to lose stability & spit out errors in Prime95.

[harrynowl]

Interesting, I thought the silicon lottery meant that CPU's differ in that some overclock better than others, but it also means that CPU's can produce different amounts of heat at the same volts? 

That's called leakage

 

Some chips get really hot with volts, other don't.

 

Professional overclockers go for the low leaking (hot with volts) as apparently they're better under water/exotic cooling for stability.

 

At least this is what I understand.

[AmusedGoat]

:blink:

 

 

Reseat your cooler.

 

70c on 1.3v using a 212 Evo is abysmal.

 

I have a Fx 8320 and a 92mm zalman cns5px that is weaker than a 212 and I am getting 51c max under load IBBT @ 1.34v. Running at 4.4ghz stable.

[Priller]

That's called leakage

 

Some chips get really hot with volts, other don't.

 

Professional overclockers go for the low leaking (hot with volts) as apparently they're better under water/exotic cooling for stability.

 

At least this is what I understand.

 

Actually they go for high leakage. Low leakage clock very well down low but as the heat and voltage increases it actually forms a wall in CPU core speed that is hard to overcome. High leakage cpus will usually need more voltage at lower core speeds but less voltage at higher core speeds. 9370s and 9590s are high leakage 8350s.

[TheRongGuy]

:blink:

 

 

Reseat your cooler.

 

70c on 1.3v using a 212 Evo is abysmal.

 

I have a Fx 8320 and a 92mm zalman cns5px that is weaker than a 212 and I am getting 51c max under load IBBT @ 1.34v. Running at 4.4ghz stable.

I re-installed it, still not any better. For thermal paste, I just put the small pea sized blob in the middle as recommended by most people. The backplate is completely secure, and I tightened the screws as much as I could. I guess I just got really unlucky with the silicon lottery.

[Game Time With Nick]

This guide was helpful! Thanks!