Buildzoid's 3700x Static Overclock Degrades Processor

[mtcn77]

You cannot get a consensus on cpu ihs dishing, or asymmetrically displaced cpu and its hideous consequences, or the voltage requirements, but boy doesn't everyone like naming names instead?
I like forums. At least everyone is true to character in a sick flick kind of way.

 

[moproblems99]

Never made any claim what I said was "Bullet Proof" or gospel as you put it, results can and will vary.

Don't forget I'm at the bot too and "Out of spec" is the standing order of the day with chips frying and dying for me.

I didn't mean you as in you but as the proverbial 'you', the in general 'you'. I almost edited my post.

 

[gupsterg]

While on auto these chips can hit 1.475~1.5V

You are reffering to 1xxx and 2xxx chips ? I ask because my 2700X defaults to 1.45V in BIOS.

UEFI is unlike legacy BIOS. UEFI is like a mini OS, Ryzen will boost when sat in UEFI. This has been the case with 1000/2000/3000 series AM4 and Threadripper. Get yourself a wall socket power meter and you'll also see quite high power usage when sat in UEFI.

I’m sorry but you don’t know all of it. 4.2GHz on the 3600 is for single thread only and not all core. And it’s doing this opportunistically and not sustained. This means that it should not work or loaded to 4.2GHz on all core loads. One way or another if you do that you will exceed silicon limits. You know anything about this and the silicon FITness controller?

I've got a launch purchase R5 3600 (besides having had R7 3700X/3x R9 3900X), at stock if CPU power usage is below ~60W I can have ~4.2GHz all cores boost on R5 3600.

Spoiler

PBO set to 105W CPU limits and +150MHz clock over-ride can result in ~4.35GHz all cores boost if ~60W or less is being pulled through CPU, link.

 

[Ferrum Master]

Anyone wouldn't come to idea he has a dud CPU?

I have a old 3960X that bsods on stock voltage also... so what? It has been like that since day zero and is still alive in my LP box, I also noticed that way later, as the bsods were rare. It didn't affect his overclock ability at all, it just need more voltage at certain p states.

We are spamming 7 pages making philosophy etc, but to be fair, there should be a great amount CPU's ran through torture test like in a laboratory and then make such claims... this is nothing like mere gossip talk about cheating housewives.

 

[Bones]

I didn't mean you as in you but as the proverbial 'you', the in general 'you'. I almost edited my post.

Thank you for clarifying that.

 

[gupsterg]

I'm sure not finding out anything about this "silicon FITness controller " via search engine. Got a link?

Zen in reviewers guide this is what it states.

As a general guideline: a CPU voltage of 1.35V is acceptable for driving everyday overclocks of the AMD Ryzen processor. Core voltages up to 1.45V are also sustainable, but our models suggest that processor longevity may be affected. Regardlessof your voltage, make sure you’re using capable cooling to keep temperatures as low as possible.

See page 34 in this PDF.

For Zen+ I saw no official information, the only thing we have is The Stilt stating his observations concerning FIT under heading "Where is the limit?" in this post.

For Zen2 again I have seen no official information. Robert Hallock from AMD does not state one, his post on r/AMD. Now if you wanna see what FIT decides for your 3000 series follow what The Stilt posted here.

I posted a screenie and video link concerning my R5 3600 at stock and PBO+150MHz. Averaged voltage can be ~1.375V if ~60W or less is being pulled through CPU by loading application, once load and or temperature increases the clocks and voltages drop. The R7 3700X and 3x R9 3900X are also the same for this aspect.

 

[moproblems99]

Thank you for clarifying that.

I almost edited my post as I was typing but all that crap sucks on mobile.

 

[Zach_01]

What I wanted to state in the end by saying these CPUs are not like any other we ever know is of course the 7nm process. What @Bones has stated with the light bulb pretty much.

But except the level of voltage (and mostly in conjunction with current) they can sustain, I strongly believe that the process it self and maybe the implementation of it it’s not for static speeds and voltages. All other past CPUs are more sustained in a speed and voltage during a light or a serious load. If you study the range of speed and voltage Ryzen 3000 is fluctuating at a given load, anyone will see that it’s unique. There is a very specific reason it behaves like this.
And this for me is the function of FIT that preserves the silicon.
That’s why I’m always sayIng that not only the level of voltage, but the static state of speed and voltage also contributes to degradation of the silicon. It’s not meant for this kind of operation, in the long term, more than any other CPU.

 

[eidairaman1]

What I wanted to state in the end by saying these CPUs are not like any other we ever know is of course the 7nm process. What @Bones has stated with the light bulb pretty much.

But except the level of voltage (and mostly in conjunction with current) they can sustain, I strongly believe that the process it self and maybe the implementation of it it’s not for static speeds and voltages. All other past CPUs are more sustained in a speed and voltage during a light or a serious load. If you study the range of speed and voltage Ryzen 3000 is fluctuating at a given load, anyone will see that it’s unique. There is a very specific reason it behaves like this.
And this for me is the function of FIT that preserves the silicon.
That’s why I’m always sayIng that not only the level of voltage, but the static state of speed and voltage also contributes to degradation of the silicon. It’s not meant for this kind of operation, in the long term, more than any other CPU.

Apparently I am just shy of 1.55V on my rig (1.524)

 

[Zach_01]

Apparently I am just shy of 1.55V on my rig (1.524)

Oh I had an FX8370 and I know what abuse they can take... I can imagine the level of power draw you are on this voltage...

 

[eidairaman1]

Oh I had an FX8370 and I know what abuse they can take... I can imagine the level of power draw you are on this voltage...

Considering 5.0 lol, yeah, i just found a guide from 2013 for Asus motherboards, i learned the hardway of overclocking this platform because i had no guide then, be sweet if I could do a fsb oc and get the ram above DOCP/XMP specs and all extended dram tweaks set.

 

[Bones]

What I wanted to state in the end by saying these CPUs are not like any other we ever know is of course the 7nm process. What @Bones has stated with the light bulb pretty much.

But except the level of voltage (and mostly in conjunction with current) they can sustain, I strongly believe that the process it self and maybe the implementation of it it’s not for static speeds and voltages. All other past CPUs are more sustained in a speed and voltage during a light or a serious load. If you study the range of speed and voltage Ryzen 3000 is fluctuating at a given load, anyone will see that it’s unique. There is a very specific reason it behaves like this.
And this for me is the function of FIT that preserves the silicon.
That’s why I’m always sayIng that not only the level of voltage, but the static state of speed and voltage also contributes to degradation of the silicon. It’s not meant for this kind of operation, in the long term, more than any other CPU.

I'm glad to hear you finally understood what I was getting at with this.
Again, I never run mine with no more than it needs to operate for that very reason based on the chip itself.

Each nm downsize equals a reduction in voltage required for basic operation and that's why past chips are indeed relevant because this effect from downsizing in nm has been happening all along as a proven, historical trend with each reduction in the nm process.
From day one up until now it's still silicon in use for making them and until they change to something else like Graphine to make them from for example, it's gonna be like that.

And here's what my FX-9590 can take - At subzero of course and yes, it's capable of even more this way.
I just gotta figure it out to get it.

Bones`s 10 min 13 sec 266 ms score: SuperPi - 32M with aFX-9590

The FX-9590 @ 7741MHzscores getScoreFormatted in the SuperPi - 32M benchmark. Bonesranks #null worldwide and #null in the hardware class. Find out more at HWBOT.

hwbot.org

 

[eidairaman1]

I'm glad to hear you finally understood what I was getting at with this.
Again, I never run mine with no more than it needs to operate for that very reason based on the chip itself.

Each nm downsize equals a reduction in voltage required for basic operation and that's why past chips are indeed relevant because this effect from downsizing in nm has been happening all along as a proven, historical trend with each reduction in the nm process.
From day one up until now it's still silicon in use for making them and until they change to something else like Graphine to make them from for example, it's gonna be like that.

And here's what my FX-9590 can take - At subzero of course and yes, it's capable of even more this way.
I just gotta figure it out to get it.

Bones`s 10 min 13 sec 266 ms score: SuperPi - 32M with aFX-9590

The FX-9590 @ 7741MHzscores getScoreFormatted in the SuperPi - 32M benchmark. Bonesranks #null worldwide and #null in the hardware class. Find out more at HWBOT.

hwbot.org

I am fortunate to be pushing mine 2GHz lower than yours on a lower profile tower cooler

 

[Bones]

If this chip holds up vs degredation I believe it can top 8GHz - In fact it's been there but coudn't get a validation of it at the time.
Needs a different board to make it happen.

 

[eidairaman1]

If this chip holds up vs degredation I believe it can top 8GHz - In fact it's been there but coudn't get a validation of it at the time.
Needs a different board to make it happen.

Are you saying you need a Crosshair? Or a high end AsRock/GA board?

 

[Zach_01]

Well I couldn’t go past 4.6GHz with mine. Didn’t have the right cooling solution. When I finally got the AIO I currently have, I was to lazy to try for more...
Even at 5.0GHz it’s like 250+W or what?

 

[eidairaman1]

Well I couldn’t go past 4.6GHz with mine. Didn’t have the right cooling solution. When I finally got the AIO I currently have, I was to lazy to try for more...
Even at 5.0GHz it’s like 250+W or what?

Never worried about that as I have a 1250Watt PSU (was gonna do crossfire till it got abandoned)

I disabled a lot of power settings on the motherboard.

 

[Zach_01]

Yeah never concerned about power usage and PSU wise I always had headroom also.
It’s for the cooling issue I commend for the 250+W

 

[eidairaman1]

Yeah never concerned about power usage and PSU wise I always had headroom also.
It’s for the cooling issue I commend for the 250+W

I need to look at my cooler and see if it is even rated for that.

Well There is no wattage info i could find, just know under gaming conditions it stays around 55°C, under Ryzen Blender 75°C.

Moral of the story is CPUs can only handle less and less vcore without burning up. Buildzoid had a oops moment and that just proves we cant run 1.375 through a 3700X and expect it to survive. And Both Intel and AMD are taking overclocking away from us which sucks

 

[Zach_01]

Well, what else can they do... things are going like that... Not that they doing it on purpose.

 

[R-T-B]

I don't say so needlessly. It is that he is using msi and that entails its own conditions on his supposed 1.375v which is 1.475v in fact.

Ok, prove you know more than him. I'm sure your resume is extensive.

Or, alternatively, stop.

 

[eidairaman1]

Ok, prove you know more than him. I'm sure your resume is extensive.

Or, alternatively, stop.

He needs to take a hike

 

[TheoneandonlyMrK]

Well, what else can they do... things are going like that... Not that they doing it on purpose.

They are doing it on purpose, to optimize performance , really quite uniformly at least for all users , the only outlier to this would be laptops wherein year on year power budgets drop yet meaning full performance is expected to rise.
And in slimmer packages.

Afaik all dies are now ground down to be considerably thinner sometimes flexible even, that can't help with radiation transmission weva.

 

[moproblems99]

From day one up until now it's still silicon in use for making them and until they change to something else like Graphine to make them from for example, it's gonna be like that.

Yes, it is silicon but the individual components are 40% smaller than the last one. For all we know, the properties (functional, for us) start to change and historical trends don't matter anymore. Similar to how Ryzen's, especially 3s, are bucking some of the trends that we are used to operationally. Again, similar to how intel is having problems clocking on 10.

 

[Bones]

Are you saying you need a Crosshair? Or a high end AsRock/GA board?

I've got one, just need to use it with the chip.
The board I was using itself can throw 2.0v's+ to it.