5800x (and other Zen 3 chips) PBO settings/Temperature fix

[ntdouglas]

Semi-automated undervolting, less volts means less heat - so higher boost

Do it too far and it's unstable at idle, usually

Mussels whats man. ts been awhile. Yeah I've been reading about it. Might try might not. I'm not quite the tinkerer I used to be lol. Last night I put a -5 all core and didnt see anything.

I've gotten up 4,915mhz on my 5600x with a slight Bus speed bump. You have a golden sample like mine in CTR
Might does that too usually I read HWinfo for the real.

Hope so. The game was just loading thats why the temps are so low. It maxes out around 63c. But still getting 4800-4825 all core boost. What did you bump the bus up tp?

 

[taakefyrsten]

In case someone with the same questions as I do come and read here. These are some data for reference about thermals and performance.

5800X
3080TI Air cooled
Gigabyte Aorus elite ax v2 rev 1.0
Meshify C - 2x NF-A14 Front intake, NF-A14 top rear exhaust, NF-A12x25 Rear exhaust. Fans running at 40-50% until 65c, then slowly ramp up to 100% at 85c

Tested at ~22c ambient

PPT 111 EDC 74 TDC 117
Curve offset: -21 -15 -30 ... -30

CPUZ
SC: 660-670
MC: 6850

R23
SC: ~1600
MC: 153XX-low 15400 | NH-U12A - 75-76c | SilentiumPC Fera 5 DualFan 80-81c

R20
SC: 624 | NH-U12A - 64c
MC 6020 | NH-U12A - 75c







PPT 130w EDC xx TDC xx
Cinebench R23 multicore
NH-U12A - 89.8c - 15300



Cyberpunk 10 minute idle
NH-U12A - 74-75c
SilentiumPC Fera 5 DualFan 77-82c

Metro Exodus numbers of the top of my head from playing the game
NH-U12A - 63-70c
SilentiumPC Fera 5 DualFan 65-75c

Jedi Fallen Order
NH-U12A - 64-73c
SilentiumPC Fera 5 DualFan N/A

Note that the Fera5 was a fantastic cooler for the price, but the single tower cooler saturates very quickly and the fan has to ramp for minor stuff such as opening a program which the U12A doesn't have to ramp for.

 

[ntdouglas]

So I messed with this all weekend on and off with no positive results so I got fed up and put pco back to default settings and disabled it. Now my 5600 is boosting 200 mhz slower. Rebooted and went into bios and reset to optimum settings and I'm still down 200 mhz. Kinda stumped. But I was thinking with over 2 dozen bios changes would cmos reset be in order? Also I seen that mussels recommends using the pbo section in Ai tweaker. Everything I've read says to use pbo in the advanced menu so what gives? What is everybody here using? Any input would on both questions would be appreciated.

 

[de.das.dude]

So I messed with this all weekend on and off with no positive results so I got fed up and put pco back to default settings and disabled it. Now my 5600 is boosting 200 mhz slower. Rebooted and went into bios and reset to optimum settings and I'm still down 200 mhz. Kinda stumped. But I was thinking with over 2 dozen bios changes would cmos reset be in order? Also I seen that mussels recommends using the pbo section in Ai tweaker. Everything I've read says to use pbo in the advanced menu so what gives? What is everybody here using? Any input would on both questions would be appreciated.

maybe i can help? what are you boosting to?

 

[ntdouglas]

I was 4.85ghz. Now 4.65-4.7.

 

[de.das.dude]

I was 4.85ghz. Now 4.65-4.7.

well mine does 4.65 with default settings on all cores.

i have never seen it go to 4.85.

At 4.65 it is already at 110W consumption. which is almost double the designed power.

are you sure it was on all cores? because few cores goes that high

 

[ntdouglas]

Look at my screeshot on page 13. Do you have autooc enabled?

 

[Ibizadr]

Do

I was 4.85ghz. Now 4.65-4.7.

Do you update your bios recently?

 

[ntdouglas]

Yep on the latest. 4021

 

[Mussels]

I was 4.85ghz. Now 4.65-4.7.

You have to add the +200Mhz in PBO to get the higher boost

If you're on the new agesa, it boosts lower than previous BIOS did

 

[ntdouglas]

What's up mussels. Yeah 200mhz is enabled. I was gonna clear cmos but then I thought maybe it's the ambient Temps up here in Chicago. It was in the 50's for 2 days. Colder than hell tonight though. I'll check it out tonight.

I'm on bios 4021 stable. Are you talking about the beta you flashed?

What about ai tweaker verses advance menu for pbo Mussels? I think most use advanced pbo menu for overclocking. Your findings?

 

[de.das.dude]

Look at my screeshot on page 13. Do you have autooc enabled?

that game screenshot one?
its not even loading all the cores. so that screenshot really has no value. PBO works by optimizing how much cooling you get under load, while taking into account your silicons TDP etc.
since its very hard to reach tdp with a few cores, that limit wont be hit and you will almost always get higher clocks.

run some aida64 stress test or cinebench to load all cores at 100%. that is the real boost clock and use hwbot to monitor.

i have seen it randomly jump to 4.8ghz on few cores while gaming, but that is meaningless.

the real clock is what you end up with after 30mins of full load. for me its 4.65 and i dont have a really good radiator. just a normal 100$ antec one with good fans in push pull as an intec.
Plus its stupid hot here.

 

[Ibizadr]

Yep on the latest. 4021

Maybe it's not a bug maybe it's a "feature"
If you do some research with last agesa bios most of people can't get the max boost you got in the past.

 

[Taraquin]

Maybe it's not a bug maybe it's a "feature"
If you do some research with last agesa bios most of people can't get the max boost you got in the past.

There is a voltage limit in agesa 1.2.0.5 that is a bit lower than previous agesas. If you don't use CO then boost might be limited due to this.

 

[ntdouglas]

that game screenshot one?
its not even loading all the cores. so that screenshot really has no value. PBO works by optimizing how much cooling you get under load, while taking into account your silicons TDP etc.
since its very hard to reach tdp with a few cores, that limit wont be hit and you will almost always get higher clocks.

run some aida64 stress test or cinebench to load all cores at 100%. that is the real boost clock and use hwbot to monitor.

i have seen it randomly jump to 4.8ghz on few cores while gaming, but that is meaningless.

the real clock is what you end up with after 30mins of full load. for me its 4.65 and i dont have a really good radiator. just a normal 100$ antec one with good fans in push pull as an intec.
Plus its stupid hot here.

I was wondering why it was boosting when not all were loaded. But still something has changed because it doesn't do that since I tried undervolting. Weird.

 

[OneMoar]

not sure I have a problem with the voltage limit of ~1.46v

what I really want is a way to tell pbo to stop going into boost clocks when the cpu isn't doing anything
at <20% utilization it should not be boosting all cores

 

[Ibizadr]

not sure I have a problem with the voltage limit of ~1.46v

what I really want is a way to tell pbo to stop going into boost clocks when the cpu isn't doing anything
at <20% utilization it should not be boosting all cores

I think this is normal. On my system it's normal and I think it's because even at<20 utilization the cpu need to handles a lot of things so I think it's normal but I'm not an engineer.

 

[Mussels]

What's up mussels. Yeah 200mhz is enabled. I was gonna clear cmos but then I thought maybe it's the ambient Temps up here in Chicago. It was in the 50's for 2 days. Colder than hell tonight though. I'll check it out tonight.

I'm on bios 4021 stable. Are you talking about the beta you flashed?

What about ai tweaker verses advance menu for pbo Mussels? I think most use advanced pbo menu for overclocking. Your findings?

I mean, i've got an entire two asus boards here: of course yours may behave different. I think its mostly a problem if the two pages have different settings - i think i've seen PBO settings listed upto four times on one board.

The new AGESA does boost less, but in reality the performance loss is tiny. It's going to reduce those temp spikes for a lot of people as well

 

[ntdouglas]

Been tuning each core a little bit. Probably a stupid question but here goes. In ryzen master cores 2 and 4 are preferred and all cores are numbered 1 through 6. In curve they are numbered 0 through 5. Is core 2 from ryzen master really core 1 in curve?

 

[OneMoar]

I mean, i've got an entire two asus boards here: of course yours may behave different. I think its mostly a problem if the two pages have different settings - i think i've seen PBO settings listed upto four times on one board.

The new AGESA does boost less, but in reality the performance loss is tiny. It's going to reduce those temp spikes for a lot of people as well

id like to test it but the VDDG voltage control is still broken I have opened multiple tickets with gagablow and they keep closing them saying they can't reproduce the issue despite multiple other people also opening tickets

 

[pf100]

Been tuning each core a little bit. Probably a stupid question but here goes. In ryzen master cores 2 and 4 are preferred and all cores are numbered 1 through 6. In curve they are numbered 0 through 5. Is core 2 from ryzen master really core 1 in curve?

The way to handle that is to look at whatever you're using, no matter what utility or bios, and see what number the core count starts and ends with, as long as it shows the correct number of cores and compare them that way. Ryzen Master starts at 1 and in your case ends in 6. In the bios the core count starts at 0 and ends with 5. So yes, core 2 in Ryzen Master is core 1 in the bios. In any case where the number starts with 1 compared to something where it starts with 0 it will be that way.

 

[ntdouglas]

The way to handle that is to look at whatever you're using, no matter what utility or bios, and see what number the core count starts and ends with, as long as it shows the correct number of cores and compare them that way. Ryzen Master starts at 1 and in your case ends in 6. In the bios the core count starts at 0 and ends with 5. So yes, core 2 in Ryzen Master is core 1 in the bios. In any case where the number starts with 1 compared to something where it starts with 0 it will be that way.

Thanks man. I thought so but just wanted to make sure.

 

[Athlonite]

Been tuning each core a little bit. Probably a stupid question but here goes. In ryzen master cores 2 and 4 are preferred and all cores are numbered 1 through 6. In curve they are numbered 0 through 5. Is core 2 from ryzen master really core 1 in curve?

in PC world 0 is always the first number so 0 1 2 3 4 5 6 so yes core 1 is actually the second core on your CPU

 

[pf100]

in PC world 0 is always the first number so 0 1 2 3 4 5 6 so yes core 1 is actually the second core on your CPU

Obviously not because Ryzen Master starts at core 1.

 

[outpt]

Did not Intel start the Core 0 naming scheme and some but not all label it as they wished? You can always use cpu-z if there’s any doubt