Overclocking FX8320

[Nomelas]

Hi, I recently upgraded my whole system basically with everything but the PSU and mobo (and case). I installed my new FX8320 with the Hyper 212 evo cooler and my motherboard is an old first gen Asus M5A97 (non anything). I'm trying to overclock it and I keep hitting a brick wall at around 4100 MHz. running prime95 the 8th core always fails (or stops?) while running the test. I've tried bumbing up the voltage and messing around with a ton of different settings, OC'ing the multiplier and the FSB frequency but it seems like no mater what I try, it doesn't want to go past 4100Mhz. My old cpu was an FX4100 and I also could only OC to 4100mhz before it did the same. Could this be an issue with my board?

[helping]

Download CPU-Z, check and calculate vdroop between load voltage and set voltage in BIOS.

 

Change LLC settings or raise voltage by the difference to compensate. 

[TechFan@ic]

Make sure your RAM is stable, input the latencies manually, the SPD (auto) settings for my memory were in fact read incorrectly and caused instability issues and blue screens.
This was on a 990FX Asus Sabertooth R2.0 motherboard.

[Nomelas]

I'm using 4x corsair vengeance 4gb 1600mhz. They are running at 1333 for some reason so I manually changed them to 1600mhz

[TechFan@ic]

I'm using 4x corsair vengeance 4gb 1600mhz. They are running at 1333 for some reason so I manually changed them to 1600mhz

I have 2x 8GB 1600mhz Corsair Vengeance modules, so now I'm fairly certain your memory is causing the instability.

Go to the BIOS and change the timings according to the specs of your ram.

Mine were 9-9-9-27 T2 @ 1600mhz.

 

[Nomelas]

I've manually set it to 8-8-8-24 @ 1600mhz, I have 4x CMZ8GX3M2A1600C8, I noticed those were not the same numbers that my BIOS auto set them to. I'm still running into the same issue though, I  have noticed that at idle my Vcore is ~1.320 to 1.332 even though I manually set it to 1.35, and when I run prime95 it actually lowers to 1.28 and stays there the entire time... I'm thinking this might be the reason? Is the some BIOS setting I'm missing? the board again is M5A97 and it's lacking the detailed LLC settings that it seems other boards have.

[-rascal-]

I've manually set it to 8-8-8-24 @ 1600mhz, I have 4x CMZ8GX3M2A1600C8, I noticed those were not the same numbers that my BIOS auto set them to. I'm still running into the same issue though, I  have noticed that at idle my Vcore is ~1.320 to 1.332 even though I manually set it to 1.35, and when I run prime95 it actually lowers to 1.28 and stays there the entire time... I'm thinking this might be the reason? Is the some BIOS setting I'm missing? the board again is M5A97 and it's lacking the detailed LLC settings that it seems other boards have.

 

All memory modules (I believe it is all...) will be automatically set to run at 1333MHz at default. That is the "Fail Safe" settings, and the timings are according to the industry JEDEC standards.

 

It seems like your motherboard indeed does not have LLC (Load-Line Calibration) feature available. The voltage droop to 1.28V could be your problem. When your CPU is under load, the voltage is dropping too low to remain stable past 4.1GHz. Since this is the case, the only way for you to compensate for the voltage drop is to manually bump the Core Voltage up. Instead of setting the Core Voltage to 1.35V in BIOS, you may need to bring it up to 1.37V, 1.4V or something likewise.

 

The ASUS M5A97 motherboard uses a 4+2 heatsinked VRM power system. It should be perfectly fine for overclocking a 125W FX-8320 chip, but it is always wise to keep an eye on the temperatures.

[Nomelas]

I ran memtest86+ overnight and it seems that my memory had an error on me. I'm not sure of the cause because I've barely used the program before. I realized after I was finished that I kept my cpu clock speed at 4200 mhz (unstable for me so far with prime95) so would that also cause my memory to error? I'm going to test more tonight after I reset everything to defaults.

[Faa]

I ran memtest86+ overnight and it seems that my memory had an error on me. I'm not sure of the cause because I've barely used the program before. I realized after I was finished that I kept my cpu clock speed at 4200 mhz (unstable for me so far with prime95) so would that also cause my memory to error? I'm going to test more tonight after I reset everything to defaults.

Pretty much yes. Also whats the speed of your memory? Lots of AMD cpu's cant even do 1866MHz on ram..

[Nomelas]

I have 4x4gb modules rated for 1600MHz 8-8-8-24, shouldn't be a problem I wouldn't think. piledriver is suppose to be able to go pretty high....

[Faa]

I have 4x4gb modules rated for 1600MHz 8-8-8-24, shouldn't be a problem I wouldn't think. piledriver is suppose to be able to go pretty high....

Does memtest detect any errors at stock cpu clocks? Just quote me so I can know that you responded

[Nomelas]

Does memtest detect any errors at stock cpu clocks? Just quote me so I can know that you responded

I'm going to run it again overnight tonight and see...

[Faa]

I'm going to run it again overnight tonight and see...

You only need 1 run.. Memtest just scans for bad sectors doesnt test stability so its pointless to run it overnight..

[Nomelas]

You only need 1 run.. Memtest just scans for bad sectors doesnt test stability so its pointless to run it overnight..

I ran it overnight last night and there were no errors for the first hour or 2 but when I woke up the next morning everything was red Lol..

[Faa]

I ran it overnight last night and there were no errors for the first hour or 2 but when I woke up the next morning everything was red Lol..

Ah okay. If it still comes with errors you need to RMA your ram..

[Nomelas]

I didn't have a chance to last night, fell asleep watching a movie... However, doing some more digging i found that my ram comes with XMP config that basically tells my mobo what timings to set my ram to. XMP however only works for intel boards but there's an option on my ASUS board for D.O.C.P which supposedly reads the XMP SPD settings. is this accurate and will this work correctly?

[airdeano]

I didn't have a chance to last night, fell asleep watching a movie... However, doing some more digging i found that my ram comes with XMP config that basically tells my mobo what timings to set my ram to. XMP however only works for intel boards but there's an option on my ASUS board for D.O.C.P which supposedly reads the XMP SPD settings. is this accurate and will this work correctly?

 

DOCP works MEH.. since you have already set the system manually, you in a sense

did what DOCP/XMP do. verify DRAM voltage (not AUTO) and retest. if still unstable

bump the CPU vCore voltage up till stable. keep an eye on the temperatures as voltage

is increased so will temperatures. use a reliable temperature app or included suite to

monitor temperatures. not all are correct. some are close and some are in left field

chasing butterflies. 

[TechFan@ic]

I didn't have a chance to last night, fell asleep watching a movie... However, doing some more digging i found that my ram comes with XMP config that basically tells my mobo what timings to set my ram to. XMP however only works for intel boards but there's an option on my ASUS board for D.O.C.P which supposedly reads the XMP SPD settings. is this accurate and will this work correctly?

You're having EXACTLY the same issues I had with my Corsair memory modules.

It's specifically the last generation of Corsair Vengeance & Dominator that people have had stability issues with, including reviewers.

It sounded like you haven't set all the timings, set the command rate to 2T.

D.O.C.P was less stable in my personal experience than inputting the latencies manually.

Set the HT Link and NB manually to 2200.

Leave the voltage at auto, disable all power saving features, like C6 & APM.

Disable turbo, now set the clock speed to 4.2Ghz and test for stability, should be all stable for you.

Now you can try to push the clocks further up, the highest voltage I'd recommend on the Hyper 212 is up to 1.38v.

[Aniallation]

Keep in mind that with 8320, you're much more likely to get a bad overclocker. Parts binning life

[Priller]

You're having EXACTLY the same issues I had with my Corsair memory modules.

It's specifically the last generation of Corsair Vengeance & Dominator that people have had stability issues with, including reviewers.

It sounded like you haven't set all the timings, set the command rate to 2T.

D.O.C.P was less stable in my personal experience than inputting the latencies manually.

Set the HT Link and NB manually to 2200.

Leave the voltage at auto, disable all power saving features, like C6 & APM.

Disable turbo, now set the clock speed to 4.2Ghz and test for stability, should be all stable for you.

Now you can try to push the clocks further up, the highest voltage I'd recommend on the Hyper 212 is up to 1.38v.

 

Never leave voltage at auto. It will apply ridiculous voltage if it wants to and that could cause other issues depending on the motherboard etc.

 

For best performance set HT link and NB multi to x12. 

 

Keep in mind that with 8320, you're much more likely to get a bad overclocker. Parts binning life

 

This. After AMD started binning like mad alot of the 8320s just dont OC.

[harrynowl]

Never leave voltage at auto. It will apply ridiculous voltage if it wants to and that could cause other issues depending on the motherboard etc.

 

For best performance set HT link and NB multi to x12. 

 

 

This. After AMD started binning like mad alot of the 8320s just dont OC.

Unless you're lucky like me

 

My friend got a FX8320 and it couldn't run past 3.7GHz. I got a feeling his craptastic board (N68C-GS FX) was a bigger issue, we had to disable turbo to prevent bluescreenies.

[TechFan@ic]

Never leave voltage at auto. It will apply ridiculous voltage if it wants to and that could cause other issues depending on the motherboard etc.

 

For best performance set HT link and NB multi to x12. 

That's actually not what I've seen in my testing, there is a limit on how high the voltages will go when set to auto, so at higher frequencies instead of doing what auto usually does on intel platforms or older AMD ones, which is apply copious amounts of volts, it actually provides an insufficient amount of voltage so the overclock becomes unstable.

And in my testing with the 8320 that's exactly at 4.3Ghz+, so auto was totally fine for anything at or below 4.3Ghz on my particular 8320 sample.

[Priller]

That's actually not what I've seen in my testing, there is a limit on how high the voltages will go when set to auto, so at higher frequencies instead of doing what auto usually does on intel platforms or older AMD ones, which is apply copious amounts of volts, it actually provides an insufficient amount of voltage so the overclock becomes unstable.

And in my testing with the 8320 that's exactly at 4.3Ghz+, so auto was totally fine for anything at or below 4.3Ghz on my particular 8320 sample.

 

What board was that?

 

I have seen boards put 1.55 volts to OCed CPUs and it resulted in very high temps, throttling, and one board even coil whined worse then any GPU I have ever heard.

[TechFan@ic]

What board was that?

 

I have seen boards put 1.55 volts to OCed CPUs and it resulted in very high temps, throttling, and one board even coil whined worse then any GPU I have ever heard.

Asus Sabertooth R2.0.

Same UEFI as the M5A97.

The automatic OC profiles (OC genie & such) have come a long way, but manual voltage tuning will always be superior.

[Priller]

Asus Sabertooth R2.0.

Same UEFI as the M5A97.

 

The board may have been why. 

 

The high voltage on Auto have been with lower end boards.