Ryzen 3000 degradation with static voltage and static frequency

[mushmx]

It is more than that, they specifically designed these chips to throttle. Sense M.I is a big piece of the puzzle. When manually set, the SV2(did I recall the name correctly?) circuit is turned off. You cannot have the best of both worlds in that regard, either your chip floats like a butterfly or it falls flat like a brick.
It fails me how people want their justified purchase not to work as intended.
The EDC parameter is specifically designed for such single/full duty workload balancing. It is a soft pttl setting all in one, since throttling also determines the max temperature threshold.
When 'decreased', single threads do continue to boost, but multiple threads run at a lower bin. This has the largest impact on the chip since available bins are tied to the voltage bin(prime determinant of power curve).

talkning about this, keep all static, malfunction or degrade the chips?
even if system looks more stable and cooler?
i mean at stock temps and voltages going crazy (normal on ryzen) but at static everything goes fine, stable, and cool so i cant understand when somone say its not really ok

 

[oobymach]

correct me if im wrong, but im not overclocking or i doing?
because as i explain with stock defaults stuff, i get higher and peaking temps (even iddle) , peaking fps ... and after set all on static i get lower and stable temps, and better performance and higher stable fps.
and i think im not going over factory settings, on GHz or Voltages, thats why i ask to you if im overclocking.

I set mine static to 1.25v and 4.2ghz but I have a decent cooler, I don't believe you will hurt your chip as long as you stay below 1.35v, in my system also I get lower temps and better gaming performance with a static clock. If your chip runs without issues at 4.3ghz on 1.25v I say run it like that.

The only thing you gain from auto is higher single thread performance, while the static multi-thread boost is usually better than auto. On auto it drops under load (while maintaining a higher voltage) whereas on static you're reducing the voltage and increasing the frequency to gain performance and reduce heat. It's a no brainer imo but many people will argue that you should use auto, even though auto runs the chip hotter and slower.

 

[mushmx]

I set mine static to 1.25v and 4.2ghz but I have a decent cooler, I don't believe you will hurt your chip as long as you stay below 1.35v, in my system also I get lower temps and better gaming performance with a static clock. If your chip runs without issues at 4.3ghz on 1.25v I say run it like that.

The only thing you gain from auto is a higher single thread performance, while the static multi-thread boost is usually better than auto. On auto it drops under load (while maintaining a higher voltage) whereas on static you're reducing the voltage and increasing the frequency to gain performance and reduce heat. It's a no brainer imo but many people will argue that you should use auto, even though auto runs the chip hotter and slower.

Yes but its no brainer but, it came from factory in that way for a reason, should be a tech reason.

Anyway as you say i think static run cooler and better so... i just wanna resereach about bad effects of doing it meaning degrade or early malfunction.

 

[mtcn77]

It's a no brainer imo but many people will argue that you should use auto, even though auto runs the chip hotter and slower.

Thanks, although I happen to know it is because cc6 that keeps the cores cool when idling.
It really should be a no brainer that a manual oc chip with 'cc6' turned on will be cooler than one that is on auto, but cc6 tampered to be off. We shouldn't be discussing pointlessly, either you run your chip cool, or it burns due to electromigration end of story.
And I don't need to endorse stupid amd practices just to point out, it can work out nicely. You just have to do the procedure which some do not like, I do not argue against that.
All I argue is, pbo is for throttling and people use it like it is an overclocking utility - it came up solely to keep threadripper motherboards from ripping a whole in its earnings reports.

 

[oobymach]

Yes but its no brainer but, it came from factory in that way for a reason, should be a tech reason.

Anyway as you say i think static run cooler and better so... i just wanna resereach about bad effects of doing it meaning degrade or early malfunction.

I've run stress tests at 1.4v and higher before I learned about degradation, not long tests most were just a few seconds before a p95 worker failed so I stop and go back to bios and change things and re-test. AMD themselves say if you set a static voltage higher than 1.35v it will cause degradation. Also you might notice performance loss anything over 1.3v static (if you test with aida64 there is a performance decrease in ram read speed, at least in my tests if I go beyond 1.3v).

 

[mushmx]

I've run stress tests at 1.4v and higher before I learned about degradation, not long tests most were just a few seconds before a p95 worker failed so I stop and go back to bios and change things and re-test. AMD themselves say if you set a static voltage higher than 1.35v it will cause degradation. Also you might notice performance loss anything over 1.3v static (if you test with aida64 there is a performance decrease in ram read speed, at least in my tests if I go beyond 1.3v).

Wow really interesing! can you provide the amd post or source (no sarcasm or something like that) i wanna read and check that i really enjoy reading about this and this topic!
so 1.29 could be usfeful, like a balance?
as GNexus say go far from1.25 its not tangible and dont really worth it. I like steve content since hes very methodic and idk even if dont speak my language is very clear.
so thats why i wanna read amd information about voltages

 

[mtcn77]

i wanna read amd information about voltages

ClockTuner for Ryzen Simplifies "Zen 2" Overclocking, Squeezes Out Double-digit Percent Performance

I tried it with my early 3600 bin. Rated bronze by the software. It's running in an ITX-case with a single 120mm AIO so I wanted to optimize for thermals. Worked pretty well: Voltage down from 1.367 to 1.225V CCX1/CCX2 down from 4100/4100Mhz to 4000/4000Mhz CB20 multi score down from 3680 to...

www.techpowerup.com

This is from a website that is now offline(ferra.ru). Be careful about the tables, one is charting power scaling, the other the temperature.
I really question you would require any more voltage, though I admit: faster cores go to sleep(4.3Ghz) state faster than an idling(1.4GHz) core. AMD has gone all tape measure with frequency scaling on that end.
Still, the fact that you gain headroom by running a lower voltage should be evident to anybody: in fact, according to AMD, CC6 saves up ~92% of idling power consumption, so think of all the benefits you will get when you max out the cores, but run them cool in the mean time. Your cpu will be early to bed, early to rise.

 

[mushmx]

ClockTuner for Ryzen Simplifies "Zen 2" Overclocking, Squeezes Out Double-digit Percent Performance

I tried it with my early 3600 bin. Rated bronze by the software. It's running in an ITX-case with a single 120mm AIO so I wanted to optimize for thermals. Worked pretty well: Voltage down from 1.367 to 1.225V CCX1/CCX2 down from 4100/4100Mhz to 4000/4000Mhz CB20 multi score down from 3680 to...

www.techpowerup.com

This is from a website that is now offline(ferra.ru). Be careful about the tables, one is charting power scaling, the other the temperature.
I really question you would require any more voltage, though I admit: faster cores go to sleep(4.3Ghz) state faster than an idling(1.4GHz) core. AMD has gone all tape measure with frequency scaling on that end.
Still, the fact that you gain headroom by running a lower voltage should be evident to anybody: in fact, according to AMD, CC6 saves up ~92% of idling power consumption, so think of all the benefits you will get when you max out the cores, but run them cool in the mean time. Your cpu will be early to bed, early to rise.

Sorry, unfortunately i run out my english knowings. I cant fully understand your message.

So u saying im doing good? or im wrong? setting all static with c6 enabled?

 

[mtcn77]

Sorry, unfortunately i run out my english knowings. I cant fully understand your message.

So u saying im doing good? or im wrong? setting all static with c6 enabled?

Follow the link and you will witness the power cost, basically the power is not the deciding factor here. Your frequency can be met even at the opposite ends of the efficiency scale. You can decide for yourself.

 

[mushmx]

Follow the link and you will witness the power cost, basically the power is not the deciding factor here. Your frequency can be met even at the opposite ends of the efficiency scale. You can decide for yourself.

Ive looked at that and sorry man i didnt understand at all.
according table 1.25 and 4300 will kill my cpu early? because temps arent close to that table, i cant talk about watts neither

im not very experienced on this stuff

 

[mtcn77]

according table 1.25 and 4300 will kill my cpu early?

That is a 1800X. Essentially, performance depends on the 4300 and not the 1.25, so max out one while minimising the other.
I told you CC6 works faster at a higher base frequency, although that is just part of it. You will electromigrate the cores if you try to run it manually on full threaded workloads. One even did it at 1.305v. I don't think testing the waters is a good idea, what you have at the beginning is all you will end up with, let's try to make it easier to keep it along the way.
People are accustomed to wrong pbo at onset, so they don't think much about it when in fact you can do better: AMD proved that with 5000. You undervolt to gain more score. It is basic stuff, you save on power to spare more boost power.