Definitive guide to configuring the Ryzen 3900X/3950X and all other 3000 Series CPUs

[BrainMuncher]

Use Load line calibration. I set mine on 5 which reduces the voltage the most.

Heh I didn't think of that, get as much droop as possible. Might be worth a try although it will still request 1.475V. Some other day though I've had enough reboots for one day.

 

[moproblems99]

Heh I didn't think of that, get as much droop as possible. Might be worth a try although it will still request 1.475V. Some other day though I've had enough reboots for one day.

Yeah, I know the feeling. I have some A-Die ram I want to try out but just too lazy right now. I was also surprised by undervolting immediately dropping clocks. I have gotten my 3900x to about 4.15 all-core by changing LLC and limiting EDC.

 

[petedread]

Micheal Nager. I'm at the top of page 3 (just skipped to the end of the thread to post this) and I am going to have to come back and Finnish reading this thread tomorrow because the arguing has drained my enthusiasm. It looks like you have some very interesting information and ideas here, and very little in the way counter-arguments from those that disagree with you. I'm not saying you are right and them wrong because I just don't know. But I certainly don't see any harm in your method and your work and effort here should be appreciated even by those who think this is the wrong way to go about things. Looking forward to reading more and trying this.

Bios defaults. Water will be warm after 12 hours of gaming lol.

crome is the only app open.

 

[TheoneandonlyMrK]

Micheal Nager. I'm at the top of page 3 (just skipped to the end of the thread to post this) and I am going to have to come back and Finnish reading this thread tomorrow because the arguing has drained my enthusiasm. It looks like you have some very interesting information and ideas here, and very little in the way counter-arguments from those that disagree with you. I'm not saying you are right and them wrong because I just don't know. But I certainly don't see any harm in your method and your work and effort here should be appreciated even by those who think this is the wrong way to go about things. Looking forward to reading more and trying this.

Bios defaults. Water will be warm after 12 hours of gaming lol.View attachment 149819
crome is the only app open.

Those who disagree largely gave up due to wall of text replys, wherein legitimate points were dodged, but it's true to say his way is a legitimate way to imitate his efforts, which some have legitimate issues with still.

 

[Michael Nager]

Micheal Nager. I'm at the top of page 3 (just skipped to the end of the thread to post this) and I am going to have to come back and Finnish reading this thread tomorrow because the arguing has drained my enthusiasm. It looks like you have some very interesting information and ideas here, and very little in the way counter-arguments from those that disagree with you. I'm not saying you are right and them wrong because I just don't know. But I certainly don't see any harm in your method and your work and effort here should be appreciated even by those who think this is the wrong way to go about things. Looking forward to reading more and trying this.

Bios defaults. Water will be warm after 12 hours of gaming lol.View attachment 149819
crome is the only app open.

I started off with three videos from Buildzoid where he discussed overclocking (which I don't do) his 3950X.

In the first video Buildzoid is using the same motherboard as the one I have (GigaByte AURUS XTREME) the same CPU (Ryzen 9 3950X) and a kit of RAM with the same specs as mine.

His ambient temperature was a lot colder than mine (from the description of the room his ambient temperature was at least 10 °C colder than mine) and yet, even at that, he was running into thermal issues (which I do not have) in the likes of CineBench R20. Just looking at his BIOS screen and the CPU temp, his temperature is 29 °C and mine is 37 °C so that should tell you all you need to know about the difference in ambient temperature with regard to which Buildzoid and I conducted our respective experiments.

Under those circumstances, and Buildzoid taking his best shot at overclocking his essentially identically specified system he managed to get a high score of 9,554 in CineBench R20.

Without overclocking and just using the steps I outlined in my guide, I achieved a CineBench R20 score of 10,170, at lower temps and lower voltage.

That's the bottom line.

I have been experimenting with 3rd Gen Ryzen CPUs for eight months now, and with my particular 3950X in conjunction with my GigaByte AURUS XTREME motherboard for four months.

One thing about all the replies, is that NOT ONE SINGLE PERSON has demonstrated a better way of doing things to get a better result.

Surely if, as has been suggested by some, I don't know what the eff I am doing, then they would be able to come up with something to show me up for the incompetent dabbler they obviously consider me to be.

But ... nothing, zip, nada, bupkis, zero, zilch, has been forthcoming. For all the verbiage, the sound of crickets has been overwhelming with regard to any tangible refutation.

 

[lorry]

I started off with three videos from Buildzoid where he discussed overclocking (which I don't do) his 3950X.

That could be that your cpu is better than his maybe?

 

[Michael Nager]

Those who disagree largely gave up due to wall of text replys, wherein legitimate points were dodged, but it's true to say his way is a legitimate way to imitate his efforts, which some have legitimate issues with still.

If you don't read the replies, then how do you know what was, or was not dodged.

It's not my fault if my keyboard buys ink by the barrel.

If a point has been raised, I have tried my best to address it.

Something like the following from you:

So we should blindly follow you, over the hundreds of engineers and scientists, interesting.

I do not consider that to be legitimate considering that when I asked you to point me to one, just ONE single, solitary, individual scientist/engineer you could cite there was no reply.

You employed a rhetorical device and as Herber Wehner once said in the German Bundestag:

There is, and all in this house have experienced it, an objective power of the factual; there is however no objective fact replacing power of the polemical

I don't consider the lying baldy headed little shit-stain of a Marketdroid Robert Hallock to be an authority on anything; but if you do, then you are welcome to him.

Show me one person, including yourself, who has disagreed, who can come up with a better solution and I will adopt that in a heartbeat.

That could be that your cpu is better than his maybe?

Over 600 CineBench R20 points better than his best?

I don't think so, not even in my wildest dreams.

Remember I am conducting my experiments at a far higher ambient temperature, with far lower voltage than Buildzoid. There just isn't that amount of variance in silicon which could account for the difference.

Do I think that my methodology is better than his with regard to getting results?

Yes, without a doubt.

 

[TheoneandonlyMrK]

If you don't read the replies, then how do you know what was, or was not dodged.

It's not my fault if my keyboard buys ink by the barrel.

If a point has been raised, I have tried my best to address it.

Something like the following from you:



I do not consider that to be legitimate considering that when I asked you to point me to one, just ONE single, solitary, individual scientist/engineer you could cite there was no reply.

You employed a rhetorical device and as Herber Wehner once said in the German Bundestag:


I don't consider the lying baldy headed little shit-stain of a Marketdroid Robert Hallock to be an authority on anything; but if you do, then you are welcome to him.

Show me one person, including yourself, who has disagreed, who can come up with a better solution and I will adopt that in a heartbeat.



Over 600 CineBench R20 points better than his best?

I don't think so, not even in my wildest dreams.

Remember I am conducting my experiments at a far higher ambient temperature, with far lower voltage than Buildzoid. There just isn't that amount of variance in silicon which could account for the difference.

Do I think that my methodology is better than his with regard to getting results?

Yes, without a doubt.

Fine example of a wall of text where you answer what you want, do you want The question again ok answer this, oh and I read all replys.


What about software crashes corrupting your OS, software clocking amplifies the errors possible.
And over time corrupts the OS.

Go on dodge it then.

 

[Michael Nager]

Fine example of a wall of text where you answer what you want, do you want The question again ok answer this, oh and I read all replys.


What about software crashes corrupting your OS, software clocking amplifies the errors possible.
And over time corrupts the OS.

Go on dodge it then.

It's Windows 10, what you describe is what everyone else calls "Tuesday".

You don't need to employ my methodology to achieve something that Micro$haft can do all by itself without the aid of Ryzen Master.

I literally just considered that to be another rhetorical question, that you could not possibly be seriously putting up for consideration.

The ONLY time (and I am pretty sure I discussed this in one of my posts in this thread) I had serious problems was when I tried configuring the system in the BIOS with regard to the parameters THAT WORK FLAWLESSLY when applied via Ryzen Master.

The result was garbage being displayed on a screen with the same background colour as a BSOD and a reboot (after reverting the settings) ending up in Windows going into repair mode (which restored the functionality). For good measure I decided to repeat that experiment, with the same result. So I gave that up as a bad idea; stupidity of course being, "Doing the same thing over and over and expecting a different result".

That having been said one of my prior corporate positions was Senior German Engineer for Enterprise Disaster-Recovery Tech-Support and my direct supervisor was the Vice-President for Tech-Support at a corporation that was, at that time, in the Top Ten of software companies worldwide.

So backups are a thing. You might want to try it sometime, it's amazing what kind of peace of mind that brings.

In the entire time I have been applying and refining my methodology over the past six months or so (I have been experimenting in total for about eight months), I have suffered no corruption of the OS.

I regularly check the health of the OS and I also know how to configure it so that it is a lot more robust than just leaving it the way M$ thinks it should be.

I have also deployed Windows 10 Enterprise, which prevents many of the shenanigans that the "Raptors from Redmond" are free to inflict on those who use other versions of the Win10.

In other words, I know what the fuck I am doing.

 

[TheoneandonlyMrK]

It's Windows 10, what you describe is what everyone else calls "Tuesday".

You don't need to employ my methodology to achieve something that Micro$haft can do all by itself without the aid of Ryzen Master.

The ONLY time (and I am pretty sure I discussed this in one of my posts in this thread) I had serious problems was when I tried configuring the system in the BIOS with regard to the parameters THAT WORK FLAWLESSLY when applied via Ryzen Master.

The result was garbage being displayed on a screen with the same background colour as a BSOD and a reboot (after reverting the settings) ending up in Windows going into repair mode (which restored the functionality). For good measure I decided to repeat that experiment, with the same result. So I gave that up as a bad idea, stupidity of course being, "Doing the same thing over and over and expecting a different result".

That having been said one of my prior corporate positions was Senior German Engineer for Enterprise Disaster-Recovery Tech-Support and my direct supervisor was the Vice-President for Tech-Support at a corporation that was, at that time, in the Top Ten of software companies worldwide.

So backups are a thing. You might want to try it sometime, it's amazing what kind of peace of mind that brings.

That being said, in the entire time I have been applying and refining my methodology over the past six months or so (I have been experimenting in total for about eight months), I have suffered no corruption of the OS.

I regularly check the health of the OS and I also know how to configure it so that it is a lot more robust than just leaving it the way M$ thinks it should be.

I have also deployed Windows 10 Enterprise, which prevents many of the shenanigans that the "Raptors from Redmond" are free to inflict on those who use other versions of the Win10.

In other words, I know what the fuck I am doing.

You may know what the f your doing, might have backed up adequately.

But you're fuckin telling others to copy you , without knowing or being able to know their abilities.

Because you have had few issues doesn't mean others mileage wouldn't differ.

As I said , I have tried your method back at Ryzen gen one and before that with a fx8350 , it's not new.

Both times after a time gave odd issues due to corrupting of system software due to crashes, such as the new Cod game produces ATM.

Your limited by your hardwares bios, others are not.

Perspective, there's more than one.

I can easily roll better benchmarks then yours but that proves nought with regards to efficacy or stability of the system.

It's far easier to just blame Ms though eh, defence posture A.

 

[Michael Nager]

You may know what the f your doing, might have backed up adequately.

But you're fuckin telling others to copy you , without knowing or being able to know their abilities.

Because you have had few issues doesn't mean others mileage wouldn't differ.

As I said , I have tried your method back at Ryzen gen one and before that with a fx8350 , it's not new.

Both times after a time gave odd issues due to corrupting of system software due to crashes, such as the new Cod game produces ATM.

Your limited by your hardwares bios, others are not.

Perspective, there's more than one.

I can easily roll better benchmarks then yours but that proves nought with regards to efficacy or stability of the system.

It's far easier to just blame Ms though eh, defence posture A.

Oh so now we are resorting to goalpost shifting?

Please point out to me ONE SINGLE TIME in all the posts I have made on this thread where I have been guilty of, as you say, "Telling others to copy you".

What I have done, is document my methodology and show the results and then answer questions which have arisen.

You are not seriously trying to suggest that configuring a 1st Gen Ryzen and a 3rd Gen Ryzen are the same thing are you? Please tell me you are not.

I just helped someone build and configure a Ryzen 5 1600 (the AF variant) system two days ago on and MSI board, and configuring it was piss easy.

Which part of the word "Experimenting" or the phrase "For eight months" do you not understand?

I have also stated in one of the posts I made on this thread that I have had the experience of helping someone configure his Ryzen 3950X on an ASUS ROG Crosshair VIII Hero WiFi, also that I have directly experimented with a 3600X and a 3950X which I own and also a 3900X which my friend from the US loaned to me (the same guy who has the ASUS board).

I also experimented with the 3600X on a GigaByte X470 Gaming 7 WiFi Rev 1.1 mobo I have for about four months and will be putting the current system away and will be bringing out that motherboard to experiment with the 3900X before sending it back to my friend.

So no, as far as hardware is concerned I am not a one trick pony.

If you can achieve better benchmark results than I, with the same or higher clockspeeds with less voltage at lower temps (taking ambient temperature into account) then I would be happy for you to enlighten me. More than that, I would be genuinely grateful. I am not happy and content with my methodology, it is simply the best one that I have been able to come up with to go with 24/7 when I finally go Gold with the system.

Benchmarks are not an end in themselves, they are simply validation tools to test the robustness and efficacy of a configuration.

You should know that, if you have had any serious experience in configuring systems.

When I wrote:

It's Windows 10, what you describe is what everyone else calls "Tuesday".

Which part of the smiley at the end was ambiguous?

Have you lived under a rock, and not heard of the serious chaos Micro$haft as created with its various major updates to Win10?

Did I in any way, shape or form, state, imply or even suggest, that Win10 was giving me personally any problems in my endeavours to optimally configure my system? If so, could you point that out to me?

You have obviously not been doing the same thing I have, given that you document a spectacular lack of success which I have not experienced.

 

[TheoneandonlyMrK]

See a page, bored hours and posts ago.

Are you seriously suggesting that the act of tuning has changed significantly and not just the silicon's operating parameters?.

I am aware we own different kit so, I am not tuning or advising you to tune yours because I wouldn't tune with yours, and because I get paid for that , Even advice is not always free.

8 months , well done, if it is time relative I have that beat Easily.

Cautioning fool's of an actuality they have not experienced yet in life is just what happens in life, deal with it.

And I would much rather leave my pc at 4.125 all cores loaded 24/7 and the GPU doing what it can on helping to find simulation solutions for vaccines then to massage my ego , trying to prove you wrong or help you achieve better clocks on a system doing f all 99% of the time.

I hope that soothes your opinion, mine won't be changing, and you can imply as much as you want I did something wrong/bad to cause corruption, I fucking told you I did something wrong, by doing what You are now doing.

 

[RoyZ]

Well you guys were right about the offset voltage being useless, what a shame, I thought that was such a good way of doing things. You could just reclaim back any extra headroom of your individual chip while keeping all other behaviors intact and without any risk. And if in any scenario you were power limited it would net you some extra performance from higher boosts.

So I tried out the whole PBO thing and that was also a bit pointless for me, seems like a feature for extreme overclockers. All it does is let the chip use more power but since the stock voltages are so astronomically high it doesn't help you because you end up thermally limited instead. And the chip just doesn't boost much higher that normal, I saw 4.225 for like one microsecond instead of the normal 4.2, not very useful, just extra heat, the opposite of what I want.

My mobo had two extra methods for setting the voltage under "AMD overclocking" - a second override, and a "max voltage offset".

View attachment 149484

Not exactly sure how this works but it does result in reduced performance. It seems to behave differently to the regular offset voltage setting, but I didn't bother to study it in depth since it doesn't do what I want. I'd also like to know what the difference between the regular voltage override and the "AMD overclocking" voltage override is. I can't really find any explanation.

What I would really want is some way to adjust the built in voltage/frequency curve, basically a working offset voltage setting. But it just refuses to deviate from the stock curve.

Anyway after a little experimentation I settled on a manual setting of 4.1GHz @ 1.25V. 1.225V was not stable. My goal was stock performance with lower volts/power/heat. Single core is actually the same as stock because even though it boosts to 4.2 at stock (and uses 1.475V to get there), I guess it doesn't stay there long enough to make a difference. All-core performance is slightly better than stock, at stock in sane all-core workloads like x264 it would sit at around 3950MHz at about 1.35V. But the temperatures are much lower, especially at idle and in light workloads, but all core temps are 7-10C lower as well. I was able to set everything in the BIOS without issue, although I did use Ryzen master to arrive at the values.

People say that the stock settings are pushing the silicon to it's limits out of the box, but it is clearly not. There is a huge difference between 1.475V and 1.25V. I'm guessing some of this stuff makes more sense with higher end models because you might actually need such voltages to get to 4.6GHz, but for the 3600 it's really over the top.

Since I'm using the stock cooler on my 3600 I'm getting very high temps in some games, and an average of 3900ghz of core clocks in these hot games.

So couple nights ago I tried something similar to what you tried. Got ryzen master and set all cores to 4.0ghz and voltage to 1.25.

Same game ran like 10 to 15° C colder than on stock, and with great performance.

Only tried that for one night though... cause being a noob on this stuff, got scared after reading people talk about cpu degradation by messing with this stuff.

I really liked what I saw that one night though lol

 

[moproblems99]

Only tried that for one night though... cause being a noob on this stuff, got scared after reading people talk about cpu degradation by messing with this stuff.

There is no need to be any more scared using this method than any other. If you don't know what you are doing and why you are doing it, then you'll likely run into problems with any method.

The key difference to this method is that you'll need to apply your profile on startup instead of the uefi doing it for you during load. There is always the possibility of a software issue causing something to go haywire but I doubt it is too much more likely than any other issue.

 

[RoyZ]

There is no need to be any more scared using this method than any other. If you don't know what you are doing and why you are doing it, then you'll likely run into problems with any method.

The key difference to this method is that you'll need to apply your profile on startup instead of the uefi doing it for you during load. There is always the possibility of a software issue causing something to go haywire but I doubt it is too much more likely than any other issue.

Ah yeah, I get how the ryzen master works. I was talking about still not being secure enough on messing with manual voltages and static core clocks, thats why I only tried that one time.

I'm reading a lot about it though and gonna keep reading.

I'm seeing that the fact these ryzens are new, even the experienced people are still finding out what are the safe voltages and clocks for them to be left on manual.

 

[mtcn77]

TL; DR: We are only discussing this because ryzen does not yet sport high voltage monitoring - scalar setting is still a manual setting operating behind the scenes. That is the only problem we have with the 'stock setting' cult following having a bash with the stated problem with ryzen: what is automatic, is itself quite not.
You can put your vapid concern to rest though, ryzen won't always remain stagnant. There are at least two more generations with which AMD can introduce iterations of the necessary correction to its cpu fivr voltage monitoring.

 

[Michael Nager]

See a page, bored hours and posts ago.

Are you seriously suggesting that the act of tuning has changed significantly and not just the silicons operating perameters?.

I am aware we own different kit , I am not tuning or advising you to tune yours because I wouldn't tune with yours and because I get paid for that , Even advice is not always free.

8 months , well done, if it is time relative I have that beat Easily.

Cautioning fool's of an actuality they have not experienced yet in life is just what happens in life, deal with it.

And I would much rather leave my pc at 4.125 all cores loaded 24/7 and the GPU doing what it can on helping to find simulation solutions for vaccines then to massage my ego , trying to prove you wrong or help you achieve better clocks on a system doing f all 99% of the time.

I hope that soothes your opinion, mine won't be changing, and you can imply as much as you want I did something wrong/bad to cause corruption, I fucking told you I did by doing what You are now doing.

That sounds like Chingrish at its finest. I have no idea what much of that is supposed to mean. It's as if Google translate has taken a wobbly. But I don't want to be unkind to you if English is not your native language.

Taking a wild guess at it, I would say that there is a big difference between dealing with a CPU that is just one chip, like a 1st or 2nd Gen Ryzen, as opposed to two or three interconnected chips on the PCB (as is the case with 3rd Gen Ryzen) and how they have to be addressed and configured.

You sound like you are trying to be magnanimous but where you say "wouldn't" what your really mean is that you couldn't and cannot advise me on how to configure my system better than it presently is.

It is eight months out of the 38 years I have been a computer techie and yes, to my mind those eight months were well spent.

So let's take a look at what this fool can achieve on your own territory shall we?

To this end I castrated my 3950X to run on only one CCD making it a virtual 3800X, and I have configured it my way, let's see what happens with FAH when I run it:

4.125 GHz running FAH you say? You can run it all day you say?

Makes my 4.425 GHz look really shitty in comparison. Obviously I am doing something horribly wrong.

And those temps! My system is pretty close to catching fire!

Or not.

Way to go with calling me out there as a fool. You certainly gave me what-for sonny.

 

[TheoneandonlyMrK]

That sounds like Chingrish at its finest. I have no idea what much of that is supposed to mean. It's as if Google translate has taken a wobbly. But I don't want to be unkind to you if English is not your native language.

Taking a wild guess at it, I would say that there is a big difference between dealing with a CPU that is just one chip, like a 1st or 2nd Gen Ryzen, as opposed to two or three interconnected chips on the PCB (as is the case with 3rd Gen Ryzen) and how they have to be addressed and configured.

You sound like you are trying to be magnanimous but where you say "wouldn't" what your really mean is that you couldn't and cannot advise me on how to configure my system better than it presently is.

It is eight months out of the 38 years I have been a computer techie and yes, to my mind those eight months were well spent.

So let's take a look at what this fool can achieve on your own territory shall we?

To this end I castrated my 3950X to run on only one CCD making it a virtual 3800X, and I have configured it my way, let's see what happens with FAH when I run it:

View attachment 149913

4.125 GHz running FAH you say? You can run it all day you say?

Makes my 4.425 GHz look really shitty in comparison. Obviously I am doing something horribly wrong.

And those temps! My system is pretty close to catching fire!

Or not.

Way to go with calling me out there as a fool. You certainly gave me what-for sonny.

It's like talking to a minor ,look at mine , no ,That's a waste of power,, crunch, use it for something ffs, it's not even your main pc because you think your i7 4790k is on par ,,,,odd epeen boy, do you want to see mine clocked higher than that, is that it , would that shut you up ,with walls of text decrying to me you are not asking people to copy you with walls of fucking text on a thread labeled the definitive guide to overclocking Ryzen , go all out with your cock out while you're at it ,oddball.
As for you Foolish 3870x demo it's not the same binned silicon you prove naught , wanna see mine at 4.5.
Your signposting yourself as some kind of genius at this shit because 8 months with a bad back indoors on a mess about pc, your talking to people for which this has always just been life, but do go on, ##ck, we know you will.

 

[mtcn77]

I think the issue with these folk is a lack of comprehension that clock stretching is a defining feature of ryzen and that they think vdroop is bound for instability on this generation of hardware. Otherwise, there is no outlining logic to mentioning bulldozer in the same mindframe.
PS: how much was it that took buildzoid to put it to the test, 1.385v was it? Right on scalar 10 'manual' which is why it is not safe at stock.

 

[moproblems99]

Ah yeah, I get how the ryzen master works. I was talking about still not being secure enough on messing with manual voltages and static core clocks, thats why I only tried that one time.

I'm reading a lot about it though and gonna keep reading.

I'm seeing that the fact these ryzens are new, even the experienced people are still finding out what are the safe voltages and clocks for them to be left on manual.

Also, don't worry about degradation. There is no evidence to support that running a Ryzen at stock settings will cause degradation. @Michael Nager 's method will similarly not harm your stuff. If I was going to go this route, and I might just to play around, would be to run an all-core load like Cinebench in a loop and see what your CPU voltage is. If you stay under that voltage, I would think you would be all right. I am still experimenting with what I can get letting Ryzen do it's thing but after that, if curiosity still hasn't killed that cat, I'll probably read through this thread again once all the BS has been cleaned out and give it a go.

 

[sneekypeet]

One and only warning to all in this thread.
This thread is not for arguments, it is not to go at each other, it is not to bait and troll.
Anyone else wants to be funny or otherwise difficult, you will be met with permanent thread bans, and possibly points, depending on the post.

 

[Michael Nager]

Also, don't worry about degradation. There is no evidence to support that running a Ryzen at stock settings will cause degradation. @Michael Nager 's method will similarly not harm your stuff. If I was going to go this route, and I might just to play around, would be to run an all-core load like Cinebench in a loop and see what your CPU voltage is. If you stay under that voltage, I would think you would be all right. I am still experimenting with what I can get letting Ryzen do it's thing but after that, if curiosity still hasn't killed that cat, I'll probably read through this thread again once all the BS has been cleaned out and give it a go.

After forking out £750 for a CPU I'm going to be pretty careful about how I treat it.

If I had a GigaByte board from a few generations ago, then LLC would have been a thing. They did get called out on it and they made the necessary corrections.

The thing is though, that even with Vdroop it was still granting itself 1.37 Volts or more under load when left to itself, and I just don't trust that for the long term.

Never mind that it irks me to give away over 200 MHz of clockspeed for a higher temperature when I leave it to itself to "boost".

 

[moproblems99]

After forking out £750 for a CPU I'm going to be pretty careful about how I treat it.

If I had a GigaByte board from a few generations ago, then LLC would have been a thing. They did get called out on it and they made the necessary corrections.

The thing is though, that even with Vdroop it was still granting itself 1.37 Volts or more under load when left to itself, and I just don't trust that for the long term.

Never mind that it irks me to give away over 200 MHz of clockspeed for a higher temperature when I leave it to itself to "boost".

I understand that you don't like it. But that doesn't constitute a general problem for everyone else.

 

[FleischmannTV]

TBH I'd rather deactivate CBP altogether than leave it at default. Activated I get 3975 MHz with 1,362 V on CB20 Multicore. With static OC, I can achieve 4000 MHz with 1.2 V or >= 4200 with 1.3 V. I don't get why 4000@1.2V should degrade the CPU faster than stock settings. Just look at the difference. Hence my current conservative supposedly safe setup is CBP off, V-Core auto on the desktop and for gaming I just activate my 4000@1.2V Profile in Ryzen Master.

From left to right: CBP enabled + V-Core Auto, 4000@1.2V, 4200@1.3V.



I mean. I love these processors, because they can be really powerful and efficient at the same time, yet the factory settings seem like the worst kind of Auto-OC ever.

 

[Michael Nager]

I understand that you don't like it. But that doesn't constitute a general problem for everyone else.

It is where I would beg to differ, because as I said in an earlier post AMD has a track record of pushing their processors well past their optimal efficiency, one glaring example would be GCN, or another would be Vega, or the FX series of CPUs.

If the Marketdroids demand that points be put on the scoresheet with regard to bragging rights of "Boost Speed", which is a stupid concept with regard to 3rd Gen Ryzen, then I don't trust AMD to have my best interests at heart over the demands of said SalesCritters/Marketdroids.

The other thing to consider is that with 1st Gen and 2nd Gen Ryzen consumers were getting prime silicon because AMD had to do this to be competitive with Intel. They also couldn't really compete with Intel in either the Server or the HEDT segment in those generations (seems like a lifetime ago, but it was less than one year).

You will notice that in the 1st and 2nd Gen Ryzen CPUs performance could be achieved at far lower voltage. Just three days ago I was configuring someone's 1600 (AF version) and could comfortably hit 3.85 GHz at 1.137 Volts and under load it was running at 1 Volt because of the Vdroop.

With 3rd Gen Ryzen we, as consumers, are getting the shit that is not good enough to be put into the Server grade CPUs and AMD has racked up impressive sales in the Server market.

So we are getting shitty silicon, and the only way to get that silicon to perform the way the SalesCritters/Marketdroids (or as I called them at my company, "Computer User, Non-Technical" - I'm sure there is an acronym in there somewhere) is to punt in a lot more voltage.

What does AMD care if the CPUs degrade? They are only on the hook if the performance of a CPU falls below the minimum clockspeed, because the "Boost" is of course wrapped in the weasel words "Up to ...", or if the CPU actually fries.

So as opposed to first and second Gen Ryzen where the incentives for AMD were stacked in favour of the consumer, they are now, with the third Gen stacked against those interests.