3600(X) What PBO Settings Are You Using ?

[Durvelle27]

Hey guys i have a 3600 that i have had manually OC'd but now i want to see how well PBO would do as I've never tried it and when i looked into it it as straight forward as one presume. So what settings are you guys using and how has it performed.

 

[mtcn77]

I've seen a strange setup which I will share. Though this was for 3800x, not much different;

• offset: -0.102v,
• PPT - 200,
• TDC - 160,
• EDC - 150,
• scalar at 10x,
• 50 MHz auto oc.

 

[Durvelle27]

I've seen a strange setup which I will share. Though this was for 3800x, not much different;

• offset: -0.102v,
• PPT - 200,
• TDC - 160,
• EDC - 150,
• scalar at 10x,
• 50 MHz auto oc.

I would think that would be overkill on the 3600(X)

 

[mtcn77]

Yeah, the gist is offset bound undervolt is well-compensated by high tuned voltage scalar, imo. Basically creating more tdp windows for turbo events.

There's also this(I'm a shameless settings forger)

PPT: 88W -->99W
TDC: 60A -->60.5A
EDC: 90A -->114A
with 75~80C

 

[Zach_01]

Yeah, the gist is offset bound undervolt is well-compensated by high tuned voltage scalar, imo. Basically creating more tdp windows for turbo events.

There's also this(I'm a shameless settings forger)

PPT: 88W -->99W
TDC: 60A -->60.5A
EDC: 90A -->114A
with 75~80C

No no... this is not for setting PBO...

This my writing and its from another thread.
First numbers are the default PBO limits of 3600 nonX and the second ones (99, 60.5, 114, with 75~80C) is the values from an OCed CPU (4.2GHz, 1,3V) that was hitting due to all core workload.
And that was exceeding designed silicon limits... Not because it was set like that but because it was OCed with not sufficient cooling.

Now, you can set the PPT/TDC/EDC to 1000 each if you like... you will not accomplish much. I've done it... The silicon FITness manager will never allow more than 1 value to cross the default limit and this will occur only with very low temp (way below 60) when at 100% load... Not even with water can do that.
TDC is out of the equation because its valueble only when throttling (>95C)
What you can do is to cap one of them (EDC) so the other can go up (PPT) if the temperature is allowing this. And forget the 70~80s for something like that... Never going to happen...
Silicon preservation is the first and most important task and function of the internal FIT controller, and second is the performance.

 

[mtcn77]

I wasn't advocating for manual overclocking. Anyway, thanks. I cannot have enough of those user profiles. Got to fan the flames! Got more chips to excite.
I'm personally looking forward to cracking more distinct pbo guidelines. Sooner or later we'll get them.

 

[Durvelle27]

No no... this is not for setting PBO...

This my writing and its from another thread.
First numbers are the default PBO limits of 3600 nonX and the second ones (99, 60.5, 114, with 75~80C) is the values from an OCed CPU (4.2GHz, 1,3V) that was hitting due to all core workload.
And that was exceeding designed silicon limits... Not because it was set like that but because it was OCed with not sufficient cooling.

Now, you can set the PPT/TDC/EDC to 1000 each if you like... you will not accomplish much. I've done it... The silicon FITness manager will never allow more than 1 value to cross the default limit and this will occur only with very low temp (way below 60) when at 100% load... Not even with water can do that.
TDC is out of the equation because its valueble only when throttling (>95C)
What you can do is to cap one of them (EDC) so the other can go up (PPT) if the temperature is allowing this. And forget the 70~80s for something like that... Never going to happen...
Silicon preservation is the first and most important task and function of the internal FIT controller, and second is the performance.

With my cooling temps pretty much stay under 65C with load

 

[Zach_01]

I wasn't advocating for manual overclocking. Anyway, thanks. I cannot have enough of those user profiles. Got to fan the flames! Got more chips to excite.
I'm personally looking forward to cracking more distinct pbo guidelines. Sooner or later we'll get them.

Never said that you were advocating for OC...
I just pointed out that those numbers was straight from a thread and had different meaning from what you thought so.
All I'm saying is that we have to be carefull when someone is asking advice. We cant and must not copy random numbers that we saw somewhere and that we dont even know what do mean.
ZEN2 is very new, not only as a CPU line but as technology. Its still, and will be for sometime more, work in progress. And whoever is saying right now that he/she knows all of its aspects, take it with a load of salt...

With my cooling temps pretty much stay under 65C with load

Oced to 4.2GHz? voltage? 65C at all core load?

 

[mtcn77]

Never said that you were advocating for OC...
I just pointed out that those numbers was straight from a thread and had different meaning from what you thought so.
All I'm saying is that we have to be carefull when someone is asking advice. We cant and must not copy random numbers that we saw somewhere and that we dont even know what do mean.

Well, I'm all for it. PBO is for safe overclocking, don't you see?
I could misquote you, however it is best to conduct one step at a time...

 

[Durvelle27]

Never said that you were advocating for OC...
I just pointed out that those numbers was straight from a thread and had different meaning from what you thought so.
All I'm saying is that we have to be carefull when someone is asking advice. We cant and must not copy random numbers that we saw somewhere and that we dont even know what do mean.
ZEN2 is very new, not only as a CPU line but as technology. Its still, and will be for sometime more, work in progress. And whoever is saying right now that he/she knows all of its aspects, take it with a load of salt...


Oced to 4.2GHz? voltage? 65C at all core load?

All core 4.2GHz at 1.4V. I must add my ambient is below 10C so its really cold

But I'm trying to OC with PBO vs manual but PBO lost me honestly

 

[mtcn77]

All core 4.2GHz at 1.4V. I must add my ambient is below 10C so its really cold

But I'm trying to OC with PBO vs manual but PBO lost me honestly

PBO is amazing. It has different steps for low-load and high-load voltage ramping. You cannot copy that in user mode.

 

[Zach_01]

Well, I'm all for it. PBO is for safe overclocking, don't you see?
I could misquote you, however it is best to conduct one step at a time...

Telling me, if I see that PBO is safe?
I'm not sure you are following my thoughts here... and you didnt read any other commend I made on other threads. I'm trying to convince users to drop their manual OC and use PBO. But not because they will gain more performance, but because its more safe. ZEN2 CPUs have very small headroom for gains. Temp is critical to that headroom. AMD had squized the most out of them.

All core 4.2GHz at 1.4V. I must add my ambient is below 10C so its really cold

But I'm trying to OC with PBO vs manual but PBO lost me honestly

10C ambient ehh? Very nice! (not for you tho)
Are you OCed right now? What score (single and all core) are you getting in CB-R20 with 4.2GHz 1.4V? And how is the HWiNFO page during those runs?

Single core


All core

 

[mtcn77]

and you didnt read any other commend I made on other threads. I'm trying to convince users to drop their manual OC and use PBO. But not because they will gain more performance, but because its more safe. ZEN2 CPUs have very small headroom for gains. Temp is critical to that headroom. AMD had squized the most out of them.

Certainly, I have and even based my judgement on that basis, already. It seems we are the only guys not using enough expletives and needing further recitation consequentially.

 

[Durvelle27]

Telling me, if I see that PBO is safe?
I'm not sure you are following my thoughts here... and you didnt read any other commend I made on other threads. I'm trying to convince users to drop their manual OC and use PBO. But not because they will gain more performance, but because its more safe. ZEN2 CPUs have very small headroom for gains. Temp is critical to that headroom. AMD had squized the most out of them.


10C ambient ehh? Very nice! (not for you tho)
Are you OCed right now? What score (single and all core) are you getting in CB-R20 with 4.2GHz 1.4V? And how is the HWiNFO page during those runs?

Single core
View attachment 142724

All core
View attachment 142727

I haven't ran R20 but this is R11

Max temp is from Prime95 FFT

What does autoscaler do

 

[mtcn77]

I want to know why 3000 isn't, by default, offering the widest performance per watt window. Changing a few sliders offer the sickest p/w efficiency ratios.
It all works by cutting power to the chip momentary. It is like an artificial stopgap to an architectural workflow efficiency problem. Maybe, the race to sleep algorithm is working better while ram has a chance to fill into the cache. It would be expedient to let edc buffer these stall moments rather than leave it running on immediate mode at continuum.
Over the last decade, the best advances on cpu architecture have not come from brute force extensions, but from mobile advancements for a while now - huge on die caches in Pentium M leading up to Core architecture, buffered frontend program time when decoding in Sandy Bridge all extend the cycles to ram operation runtime.

Sandy Bridge introduced the 1.5K µop cache that caches decoded micro-ops. When future instruction fetch requests are made, if the instructions are contained within the µop cache everything north of the cache is powered down and the instructions are serviced from the µop cache. The decode stages are very power hungry so being able to skip them is a boon to power efficiency. There are also performance benefits as well. A hit in the µop cache reduces the effective integer pipeline to 14 stages, the same length as it was in Conroe in 2006.

Pentium M, Dothan, is the precursor to Conroe, the first of 'Core' Architecture.
Perhaps we should monitor this wildcard by extending ppt throttle limits and limiting edc to find out about ryzen's mobile readiness. It is much too efficient in its low peak power mode to not consider it an option to further overclocking down that path.

 

[Durvelle27]

Heres CB20

 

[Zach_01]

Pretty much the same as mine. My best auto boost score is 482 single and 3570 multi.

 

[Durvelle27]

Pretty much the same as mine. My best auto boost score is 482 single and 3570 multi.

That's with PBO enabled

I think i used the defaults of the 3800X

 

[Zach_01]

That's with PBO enabled

I think i used the defaults of the 3800X

You mean the PPT/TDC/EDC limits of 3800X? As I said elsewhere you can set them to 1000 each. Wont do anything.
Low temp, reduced EDC and setting PBO scalar to X2~3 can increase the clocks for middle/high loads (if temp allow it) and for single core load it will hit the same rated speed but a little more often and sustained

 

[Deleted member 193792]

All core 4.2GHz at 1.4V. I must add my ambient is below 10C so its really cold

But I'm trying to OC with PBO vs manual but PBO lost me honestly

That's way too high for 7nm. It will possibly degrade.

 

[Durvelle27]

You mean the PPT/TDC/EDC limits of 3800X? As I said elsewhere you can set them to 1000 each. Wont do anything.
Low temp, reduced EDC and setting PBO scalar to X2~3 can increase the clocks for middle/high loads (if temp allow it) and for single core load it will hit the same rated speed but a little more often and sustained

Yes same limits as the 3800X

Temps are great like mentioned i hardly ever see 65C

What settings would you suggest to be able to hit 4.15 through 4.2GHx


My RAM also running at 3733MHz 18-22-22-22-42

 

[Zach_01]

Yes same limits as the 3800X

Temps are great like mentioned i hardly ever see 65C

What settings would you suggest to be able to hit 4.15 through 4.2GHx


My RAM also running at 3733MHz 18-22-22-22-42

You mean to hit 4.15+GHz with PBO in auto boosting for all core?
Unlikely to happen...

Mine hits 4050MHz all core boost with 59~60C max and PBO settings:
PPT: 95
TDC: 60
EDC: 63
PBO Scalar X2

...and 4050~4075MHz all core boost with 61~62C max and PBO settings:
PPT: 95
TDC: 60
EDC: 63
PBO Scalar X3

If we could cool it down to 50ies it may clock close to 4100~4150MHz, but that can't happen without sub-ambient cooling (like chiller)

FYI
PBO Scalar is for keeping voltages and clocks a little higher and a little longer, but requires more cooling to have the best effect.
ZEN2 drops clocks as temp is rising. You cooling, it will raise clocks.
Scalar while increases auto-clock and auto-voltage it raises temps too so the final gain in clock is small.
The more Scalar you use the more cooling is required to sustain clocks.

 

[Durvelle27]

Am i using the AMD overclock menu and CBS XFR menu