Buildzoid's 3700x Static Overclock Degrades Processor

[kapone32]

He reminds me of that one time some dud tried to drill through a gpu board only to discover buried resistors underneath. Oh the times...

To be honest bullzoid is very good when it comes to detailing components for the VRM and subjects like this one. I don't think he made this video as what the 3700X could do but not what to do with a 3700X.

 

[Mr McC]

This brings me to something I have noticed in my own experiences. I had the 1700 and 1.3 volts was great for 3.9 GHZ, the 2600 at 1.27 volts would do 4.2 GHZ all day long the 1900X would do 4.1 GHZ @ 1.3, the 1920X would do 4.1 GHZ @ 1.25 volts and now the 2920X runs at 4.2 GHZ @ 1.215 volts. Running a 3700X @ 1.375 volts may be actually dangerous as the 65W TDP vs the 105W for the 3800X should be observed.

Forgive my ignorance, but I was under the impression that the TDP was merely "defined" by the type of stock cooler AMD ships with each processor.

 

[mtcn77]

To be honest bullzoid is very good when it comes to detailing components for the VRM and subjects like this one. I don't think he made this video as what the 3700X could do but not what to do with a 3700X.

I know. It is a rhetorical dissertation on his part. Still, the youth never cease to grant us with free entertainment.

Forgive my ignorance, but I was under the impression that the TDP was merely "defined" by the type of stock cooler AMD ships with each processor.

It takes a standard thermal resistance in series formula and sorts out from the temperature gradient its cooling capability.

 

[R0H1T]

Well TDP doesn't have a standard industry definition, & if you ask Intel they'll give a different answer from AMD. Talk with forum dwellers like us, & that's generally not gonna be pretty. Not defending Intel (or AMD) here but in today's world TDP definition is kinda malleable, especially depending on boost clocks as well as OCed speeds. You generally get more cooling headroom than what the manufacturer states as their "TDP" & you're good to go, just like with PSU & (total) system power draw.

 

[lexluthermiester]

Please understand that by posting this, I was not attempting to promote controversy in any form.

Such is unavoidable in posts like this. Not because of the subject but because of certain users.

I assumed that Buildzoid's insight into the degradation would prove of interest

I agree, it is interesting and gives insight to the limits of the Ryzen 3000 range of CPU's.

Agreed, but if his intention was to determine the precise limits, surely we have something to learn?

Ah, but therein we find the crux of this: It is impossible to precisely determine the upper limits of a CPU line. This is because of variations in the manufacturing of materials, the manufacturing of the dies with those materials, and the imperfections found within each. This is where the phrase "Silicon Lottery" came from.

 

[GlacierNine]

Forgive my ignorance, but I was under the impression that the TDP was merely "defined" by the type of stock cooler AMD ships with each processor.

TDP is meaningless because the way the number is derived is in no way standard. You can safely ignore what mtcn77 is saying - he simply doesn't have a clue what he's talking about and even if he did, he's utterly incapable of putting together a sequence of words that makes enough sense to explain it.

Intel and AMD both define TDP completely differently, to start with, so even if there were no variable clockspeeds or boost clocks, the TDP numbers couldn't be compared. Anyone claiming there is an "industry standard" for TDP is delusional - AMD and Intel are the industry, and they don't use the same formulas to calculate TDP. How standard can anything be if the two largest players in the market don't use it?

Also again, that would presume that "TDP" was based on one fixed clockspeed and voltage. This isn't how AMD or Intel quote TDP - for both companies, they quote TDP is based on a base or average clockspeed. In Intel's case they claim that max all core boost is a 25% increase on the base TDP, but they also specify it's only supposed to be maintained for a short period of time, and that it can be done because of, essentially, "spare heat capacity" in a heatsink, not indefinitely, so that claim is dubious.

AMD do something very similar, since their chips effectively overclock themselves based on ambient conditions, via Precision Boost Overdrive - how can you quote a TDP for a product that will change it's operating frequency to a number you as a manufacturer don't actually set in stone?

You can't, and AMD don't, making both manufacturer's quoted TDP ratings completely meaningless.

And even if they weren't, if AMD and Intel can't quote TDP consistently, what makes you think cooler manufacturers can? (Protip: They can't and don't)

 

[Zach_01]

Yes, and the 1.328V that my 6700K has run on for the last 3 years is also higher than the 1.2V it is intended to use.

And the 1.36V that my i7 920 ran on for 5 years is also higher than the 1.26 it was supposed to use.

Neither of those chips degraded over time because the manufacturing processes they were built with were quite capable of dealing with the extra voltage as long as cooling was available.

We've been able to run chips well out of spec with no ill effect, for decades. If 7nm can't do that, that's *new* and doesn't deserve you being so damn condescending about it. Especially considering you're trying to educate someone who can easily break down the VRM and Phase layout of a motherboard to the level of individual ICs and calculate the current capability on the fly while he does so.

Buildzoid is no idiot, don't act like he is.

Don't have much personal Experience with Zen2 apart from a a day testing , comparing it with my 1700. But in all the Years I've been running AMD I've never really suffered from the phenomena, expect perhaps Mildly with an Oppy 165 on Socket 939. I should read the thread in youtube to see if he comments on what boards have been used to "degrade" and if this CPU does this work on another manufacturers board.

This is the most common mistaken thoughts of users. The use their previous knowledge and experience on these new chips. What we know from previous CPUs including Ryzen1000/2000 series is irrelevant and do not apply to Ryzen3000. Users must forget everything they know and start from scratch. ZEN2 is not for static OC. And if you do you must do it by loosing performance setting it in very low speeds and voltage. Auto boost clocking/voltage is complex procedure that we cannot understand. First of all its doing it in bursts that no software can catch.

 

[mtcn77]

Gj hater. This isn't even my final form!

TDP (Watts) = (tCase°C - tAmbient°C)/(HSF θca)

AMD measures HSF θca from a static formula. Normally, you would get lower resistance(better conductivity) if you modulate the fan curve. They set a single level to linearise the polynomial. So you can bet it could be improved if the resistance in the denominator was accounted for. I said all that and more here:95°C is necessary

 

[GlacierNine]

Gj hater. This isn't even my final form!

AMD measures HSF θca from a static formula. Normally, you would get lower resistance(better conductivity) if you modulate the fan curve. They set a single level to linearise the polynomial. So you can bet it could be improved if the resistance in the denominator was accounted for. I said all that and more here:95°C is necessary

Congratulations, you've managed to find one of a number of arbitrary formulas, none of which return the same results as each other, and all of which are or have been used to calculate "TDP". That formula is not the same one that Intel use, and the results are not comparable.

Here, have the Anandtech article on this: https://www.anandtech.com/show/13544/why-intel-processors-draw-more-power-than-expected-tdp-turbo

And here, also have the Intel 7th Gen Core CPU datasheet (You can also look up the 9th and 10th gen datasheets from intel ark, but they actually contain less data than this one does since Intel have gotten more reticent to share the info of how they do TDP)

https://cdrdv2.intel.com/v1/dl/getcontent/335195

 

[mtcn77]

That formula is not the same one that Intel use

Gj you played yourself: cause, it is the one AMD uses!
Which goes to say, if we sideline that and use thermal transmittance formula(same as above, only accounting for area as well) we can approximate much better.

 

[GlacierNine]

Gj you played yourself: cause, it is the one AMD uses!
Which goes to say, if we sideline that and use thermal transmittance formula(same as above, only accounting for area as well) we can approximate much better.

Read what you said, and then read what I said again and try and actually understand it this time around.

 

[mtcn77]

Read what you said, and then read what I said again and try and actually understand it this time around.

Well, shifting the magic blamestick is not in your leisure.
Here is what to do: we can drop all bars when we disable cpb. We set the vid, we set the clock multiplier and work our way manually. At least this way, there is more manual overclock provisioning when we expand the thermal headroom. The resistance doesn't stay the same when heat loads are headed for an increase.

PS: I've stopped making logical attempts for one upmanship instead - the same with your one liners.

 

[A.Stables]

This is the most common mistaken thoughts of users. The use their previous knowledge and experience on these new chips. What we know from previous CPUs including Ryzen1000/2000 series is irrelevant and do not apply to Ryzen3000. Users must forget everything they know and start from scratch. ZEN2 is not for static OC. And if you do you must do it by loosing performance setting it in very low speeds and voltage. Auto boost clocking/voltage is complex procedure that we cannot understand. First of all its doing it in bursts that no software can catch.

I'm not mistaken so please don't try to correct me, I was merely stating in to so many words, that its not Like AMD to put out weak silicon. Thanks

 

[eidairaman1]

Such is unavoidable in posts like this. Not because of the subject but because of certain users.

I agree, it is interesting and gives insight to the limits of the Ryzen 3000 range of CPU's.

Ah, but therein we find the crux of this: It is impossible to precisely determine the upper limits of a CPU line. This is because of variations in the manufacturing of materials, the manufacturing of the dies with those materials, and the imperfections found within each. This is where the phrase "Silicon Lottery" came from.

When you overclock it is 100% Prayer, 80% Skill, 20% Will.

Also box o chocolates, doesn't matter what electronic is, mtbf can be from 1 zeta second to 1000+ years...

 

[rtwjunkie]

He reminds me of that one time some dude tried to drill through a gpu board only to discover buried resistors underneath. It was quite an expensive gpu as well. Oh the times...

You obviously don’t know anything about Buildzoid. He’s not some gullible kid.

 

[lexluthermiester]

When you overclock it is 100% Prayer, 80% Skill, 20% Will.

That's about right, yes! LOL!

 

[mtcn77]

You obviously don’t know anything about Buildzoid.

It is a rhetorical dissertation on his part.

Don't expect unconditional adulation, I won't praise vandalism.

 

[Zach_01]

I'm not mistaken so please don't try to correct me, I was merely stating in to so many words, that its not Like AMD to put out weak silicon. Thanks

Its not my fault that you keep doing the same mistake, and think of it like this. Its not AMD who is putting out weak silicon. Its the nature of this node that cannot be treated like previous ones. Going to 7nm and to 5nm or 3nm... it will keep diminishing headroom. You want overclock and free performance stick to 12nm or 14nm++++. The static OC era is coming to an end whether we like it or not.

 

[TheoneandonlyMrK]

You obviously don’t know anything about Buildzoid. He’s not some gullible kid.

Agreed but everyone is capable of making a mistake, not saying he has just saying.

I saw the video a day or so ago but it's not on his channel and the comments page is gone ,Op your links dead.

Interesting findings and what's come after ie disappearing.

I used to pin clocks and Volt's, now I nearly suggest via settings what to run at but I am interested no doubt my CPU has sat at 75+ since purchase at varying voltages , depending on bios ,ageesa etc.

Used to stick to 1.45 ISH and now is more conservative at 1.25-35 all cores clock higher now though too 4@1.45 ,,,4.15@1.35 all cores crunching.

We'll see if anything pans out as a few crunch with Ryzen and consumer parts were not meant for such, if anything reasonable ages Ryzens that would.

 

[thesmokingman]

Maximum FIT voltage...

 

[Bones]

Been running my 2700x with about 1.25v's since I've had it with only a few times I've ran up the volts on it.
Still doing good.

 

[Zach_01]

Been running my 2700x with about 1.25v's since I've had it with only a few times I've ran up the volts on it.
Still doing good.

But this has nothing to do with 3000. Its night and day, apples and oranges...
What I keep saying here and there over and over again, is that anything we know from other past CPUs is irrelevant with Ryzen 3000.

I really hope users understand this soon. Yesterday if possible...

 

[Deleted member 178884]

Been running my 2700x with about 1.25v's since I've had it with only a few times I've ran up the volts on it.
Still doing good.

And? Of course it's still doing good - it's not rocket science, it's a 12nm chip and has absolutely nothing to do with 7nm or ryzen 3000? Care to elaborate?

 

[mtcn77]

What I keep saying here and there over and over again, is that anything we know from other past CPUs is irrelevant with Ryzen 3000.

Well, I know I won't win any favors, but your manner to me was the same when I elected another benchmark to measure tdp in the zen 2 generation. We can, of course, trust in AMD to do the right thing, but someone has to address this ryzen 3000 uneven cpu heating issue.

Care to elaborate?

We discussed this previously, he wouldn't acknowledge it, though. I must have struck a chord.

 

[moproblems99]

Agreed, but if his intention was to determine the precise limits, surely we have something to learn?

All valid questions. If I came off as condescending, forgive me. It was not intentional. When ever you push things over the limits, expect them to degrade. What ever happens on other chips and other processes is irrelevant.

AMD, for better or for worse, appears to be clocking and volting these to the limits because they have to to keep the single thread and some thread as high as possible. Volting and clocking them more appears to basically be extreme oc now. There is not much left to get out of them. I expect degradation to occur when any changes to Zen2 or above occurs.