FX-8150: Very high VCore for OC?

[Parsleybravo]

My overclocking experience is not extensive, and most of that is with Intel processors. So, please bear with me if I sound like a complete newbie.

Here's the basic gear:

Proc: FX-8150
Mobo: GA-990FXA-UD5 (v1)
RAM: 2x4GB Corsair Vengeance

Cooling: Custom water, 360mm + 120mm (dual GPU between the proc and the 120)

The BIOS is up to date. C1E, C6, K8, AMD APM and TurboBoost are disabled. RAM is booting @ 1333MHz and has passed extensive Memtest86 trials (12+ hours without error).

As I started in on overclocking my 8150, I initially set my multiplier to 22.5 for 4.5GHz and the VCore to 1.45. My intent was to get a ballpark estimate for a bootable OC as a starting point, per this thread.

The first problem appeared just beyond the Windows login, where I experienced a hard-freeze or BSOD 101. After trying a few combinations of lowering the multiplier and raising the VCore, I began running initial stability tests using Prime95 (blend and small FFT). I continued to get hard freezes, BSOD 101s and worker errors (rounding and illegal sumout). Temperatures have been fine during the tests, not peaking above 42C in HWMonitor.

The really perplexing part to me is that I can't even get to a stable 4.3GHz with a VCore of 1.55, the maximum recommended by AMD for a water-cooled system. The most stable OC I've been able to achieve is 4.1GHz @ 1.50 VCore, and even that doesn't pass long-term Prime95 tests; it has been stable for daily usage, to include video editing/encoding and gaming.

I know every chip and motherboard are different, but these results strike me as odd considering the number of forum threads I've found where people are achieving 4.5+ with substantially lower VCore, and even those with air-cooled systems are achieving stable OCs past 4.1GHz.

Am I missing an essential piece of the puzzle, or did I just get some bum hardware?

Any advice would be greatly appreciated.

[The Stilt]

From where exactly have you found the information about 1.55V Core VDD being the 'maximum recommended' voltage for watercooling?
Anything above 1.475V on conventional cooling methods is unsafe in long term use.

[Parsleybravo]

From where exactly have you found the information about 1.55V Core VDD being the 'maximum recommended' voltage for watercooling?
Anything above 1.475V on conventional cooling methods is unsafe in long term use.

Allow me to rephrase:

1.55V is the highest VCore for which the chip is rated, according to AMD. Personally, I think you'd be hard-pressed to find anyone running 1.55V on air. I combined AMD's specification and a reasonable assumption into a single thought in a moment of sheer editorializing.

I apologize if that was confusing.

[The Stilt]

Allow me to rephrase:

1.55V is the highest VCore for which the chip is rated, according to AMD. Personally, I think you'd be hard-pressed to find anyone running 1.55V on air. I combined AMD's specification and a reasonable assumption into a single thought in a moment of sheer editorializing.

I apologize if that was confusing.

"The Maximum CPU VID value is 1.550V"

That has nothing to do with the voltage being 'safe'.
It means the can be adjusted up to 1.55V with VID configuration.
VID = Voltage Identifier.

[Parsleybravo]

"The Maximum CPU VID value is 1.550V"

That has nothing to do with the voltage being 'safe'.
It means the can be adjusted up to 1.55V with VID configuration.
VID = Voltage Identifier.

Thank you for clarifying. I'll try to keep in mind that 1.55V is not considered safe.

And yet, at a voltage so high as to be considered unsafe, I cannot pass a Prime95 stability test @ 4.3GHz without throwing a BSOD that, to my knowledge, indicates that a higher VCore would be required to make the requested CPU clock stable.

Why is that?

[Fr3ak]

First lesson in Overclocking: Almost nothing obeys any logical behavior.

So, play around with different setting, different memory dividers, settings that dont seem to have anything to do with your max CPU overclocking. I am runnning 4.75GHz on 1.4V stable for any games and or video encoding on a FX-8300, which is a different CPU, but until I finally switched to the Asus Crosshair V Formula-Z, I struggled a lot to get even close to that.

[BeepBeep2]

First lesson in Overclocking: Almost nothing obeys any logical behavior.

So, play around with different setting, different memory dividers, settings that dont seem to have anything to do with your max CPU overclocking. I am runnning 4.75GHz on 1.4V stable for any games and or video encoding on a FX-8300, which is a different CPU, but until I finally switched to the Asus Crosshair V Formula-Z, I struggled a lot to get even close to that.

Sure it does, it's just a different kind of logic

[AMDforME]

"The Maximum CPU VID value is 1.550V"

That has nothing to do with the voltage being 'safe'.
It means the can be adjusted up to 1.55V with VID configuration.
VID = Voltage Identifier.

Actually AMD has stated that you can use up to 1.5V on the FX processors if you desire, assuming proper cooling. They even listed it in the chart with 1.5V Vcore being in the "Air-Cooling" category. In the "Extreme" category they list up to 2.0V assuming water or better cooling. The chart at the end of the AMD OC'ing guide indicates what AMD feels is possible and acceptable based on different cooling options as shown in the chart. These recommendations have nothing to do with what the CPU or BIOS will allow.

Parsleybravo-

IME too much Vcore is as bad as too little Vcore on the FX processors. I haven't seen any other CPU that becomes unstable with too much Vcore but the FX processors do in my limited experience with them. YMMV.

There are a variety of OC'ing guides and there is good info. in most of them. In addition to the disabled items already set in the BIOS, start with 4.1 GHz. and try the following:

Vcore 1.425v
CPU-NB - 1.325V
HT - 1.25V
RAM +.5V over default

It's worth noting that some mobo VRM circuits can't handle an OC'd/Over-volted FX 8-core processor well under stress testing and if it hits the over-temp safety it will drop the CPU voltage and frequency which can cause P95 errors IME, even though it should not.

[BeepBeep2]

lol...
Please try 2v on 20 CPUs running phase/dice/cascade even and come back to me on how many live more than 20 seconds...

Furthermore what "AMD feels" in that guide are from people exactly like The Stilt, Macci, chew*, others like myself. People with experience under many different conditions...

Those are not engineering specification, or even official "safe limits" really. That guide was written by overclockers for noobs.

Educating the educators sure is fun isn't it...just because you read something on the internet doesn't make it true nor make the reader an expert...just because someone may assume certain things happen by some made-up logic doesn't mean its true either...

[AMDforME]

lol...
Please try 2v on 20 CPUs running phase/dice/cascade even and come back to me on how many live more than 20 seconds...

Furthermore what "AMD feels" in that guide are from people exactly like The Stilt, Macci, chew*, others like myself. People with experience under many different conditions...

Those are not engineering specification, or even official "safe limits" really. That guide was written by overclockers for noobs.

Educating the educators sure is fun isn't it...just because you read something on the internet doesn't make it true nor make the reader an expert...just because someone may assume certain things happen by some made-up logic doesn't mean its true either...

As with your past comments, you are once again incorrect. You should learn to read more and post less because you continue to confuse people looking for accurate technical information with your meritless comments.

AMD probably has a pretty damn good idea what their products can and can't do and what is safe. They don't make a habit of suggesting how to burn up your CPU and they certainly don't need you to tell them how to OC their CPUs.

You made absolutely no contribution to this thread at all and your comments do not help the OP one bit.

[The Stilt]

Actually AMD has stated that you can use up to 1.5V on the FX processors if you desire, assuming proper cooling. They even listed it in the chart with 1.5V Vcore being in the "Air-Cooling" category. In the "Extreme" category they list up to 2.0V assuming water or better cooling. The chart at the end of the AMD OC'ing guide indicates what AMD feels is possible and acceptable based on different cooling options as shown in the chart. These recommendations have nothing to do with what the CPU or BIOS will allow.

Thanks for the information!

For some odd reason this information doesn't quite line up with the official AMD specifications, thou

[AMDforME]

Thanks for the information!

For some odd reason this information doesn't quite line up with the official AMD specifications, thou

Actually when AMD sent out all of the FX-8350 Vishera CPUs to reviewers, they specifically told them to overclock them as high as they could and that it was safe to use up to 1.55V Vcore. Most reviewers however found no gains beyond 1.500-1.525V. They also needed excellent cooling to use the higher Vcores.

[BeepBeep2]

I'm pretty sure that the answer to this thread is that the Rev 1.0 UD5 does not include any kind of Load Line Calibration therefore the OP must set higher voltage because of the vdroop under load.

The droop is pretty dang bad, 1.5v set probably gives optimistically something like 1.36v at load which he needs for 4.1 stable...


All of the information AMDforME keeps regurgitating that was included in the little .pdf they put together after the Zambezi 8 GHz run is not really helpful either. Ironically chew* is seen pouring LN2 on the same page, he did a lot of testing that went into the guide anyway...

Truth of the matter, that guide is not god or endorsed as accurate in any way by AMD and references to what it says have been blown out of proportion and twisted as it passed through more eyes and ears from reviews to forums to eventual armchair experts.

Note that these results are strictly estimates based on overclocking data from a limited sample size. The actual
results will vary depending on the CPU, motherboard, cooling solution and memory modules used in the tests.

[AMDforME]

The AMD Overdrive guide is just that a guide based on AMD's expertise on their product. With all OC'ing there are no guarantees but AMD isn't foolish enough to suggest Vcore voltages certain to kill their CPUs. As long as you have proper cooling it's common to run the Vcore up to the guideline voltages that AMD suggest without concern about burning the CPU out electrically. In many cases the CPU's maximum overclock however does not require nor benefit from the maximum Vcore voltage possible.

[BeepBeep2]

But it was only 30% done by AMD employees/engineers, do you not understand that?

The opinions in that guide reflect the opinions of about two people, one of which is/was a technical marketing director who ironically was a hardware enthusiast and well known overclocker and the other being an overclocker from this forum.

[AMDforME]

^^^ Complete nonsense as usual from you...

[BeepBeep2]

Idiots can believe what they want, I know the people who were involved.

Brian tested literally 150+ chips by himself in his house on air, water, phase, dice, and LN2 to get most of that data.