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

[freeagent]

i run my 10700K 24/7@5 GHz. why use it at stock, lol. 0,3 GHz free boost, but still!

I was referring to the 5800X with an all core OC @ 5GHz. Modern Intel CPU’s should be able to do 5GHz without batting an eye with near stock volts lol. I just had my doubts on stability with the 5800.

 

[authorized_to_ill]

EDIT: Hoping someone call help me identify where I went wrong with my attempt to cool this cpu. After 2 weeks of seemingly being stable I started getting WHEA 18 (APIC ID: 0) restarts. These consistently happen when I attempt to download through the Microsoft Store.

My specs:

Ryzen 7 5800x
EK AIO 240
Asus Crosshair VIII Dark Hero
G. Skill Triden Z Neo 16GBx2 3600 CL16-16-16-36
Asus ROG Strix 3070
Samsung 970 Evo Plus 1TB
Seasonic Focus GX-850

First I set DOCP to Standard (timings were automatically set to 16-16-16-36) and set FCLK ot 1800.
In an attempt to tame temps, I messed with CO and did -5 on all cores (didn't realize even that could be too much for one of them at the time).
Later I learned about PBO and ultimately settled on the same settings I found from the mentioned Reddit discussion (PPT=120, TDC=75, EDC=110).

Everything seemed stable. However, the restarts started happening when I attempted to download Sunset Overdrive from the MS store.
Disabling CO (or set to Auto?) didn't help. Then I changed PBO to settings I thought were previously stable. Nothing. So finally I hit F5 in bios and defaulted to "Optimized settings", then changed DOCP to 3200mhz (same timings) and FCLK ot 1600.

Issue seems to have disappeared. Was able to download again and no more WHEA errors. However, I am back to high temps and my RAM is now below the advertised 3600mhz.

I hope someone can help me better understand where I went wrong and how I might still be able to help my cpu temps. I've seen some of the success posts like those from @outpt, but since I don't fully understand it I don't want to just copy and paste settings. Especially since ram and FCLK seemed to have played a major part in my particular instance. Any insight or advice would be incredibly appreciated.

 

[Cheese_On_tsaot]

Sitting at a comfy 4750mhz max boost and only hitting 66c on my cooler.

 

[freeagent]

I am using 200 ppt 140 tdc and 180 edc -30 +200 and she boosts pretty good.

Edit:

Whoops mistook the 8 for a 9.. but. I run that with my 5600X too.

 

[Space Lynx]

Sitting at a comfy 4750mhz max boost and only hitting 66c on my cooler.

this is the best way to be, find the perfect balance between great temps, and enough cpu speed -99.99% of games don't even know the difference between a few hundred mhz

 

[natr0n]

I got a 5800x coming in a day or 2. Very excite !!!

 

[Mussels]

this is the best way to be, find the perfect balance between great temps, and enough cpu speed -99.99% of games don't even know the difference between a few hundred mhz

yep, this is my results after 24 hours of gaming and mining

70C max (GPU pre-heating the loop, with gaming and mining at the same time... cpu alone is around 50C) and seeing the true max boost of a 5800x, at 5.05GHz

 

[outpt]

1.525V is normal for you in this instance. I’ve seen 1.5V on mine but never higher

 

[authorized_to_ill]

I've edited my previous post to contain the most relevant information. Hoping someone can help me out or point me in the right direction.

 

[freeagent]

1.525V is normal for you in this instance. I’ve seen 1.5V on mine but never higher

Pretty sure I’ve seen 1.55 a couple of times

Usually between 1.5-1.535 at the tippy top for single core stuff.

 

[Cheese_On_tsaot]

Wow is 1.5v not high as fck for ZEN 3?

If not then I have loads of headroom, my chip only boosts on 1.41v of voltage.

 

[freeagent]

Wow is 1.5v not high as fck for ZEN 3?

If not then I have loads of headroom, my chip only boosts on 1.41v of voltage.

Yup its out of my hands

Edit:

I saw similar behavior with my old 3600XT and my 5600X

 

[outpt]

I've edited my previous post to contain the most relevant information. Hoping someone can help me out or point me in the right direction.

reset to docp settings and now test. leave off any CO settings. my chip dose not any combination of CO and pbo overclocking. you might be in the same boat. i have ran core cycler over night without any errors. my main testing platform is any tomb raider(SoTTR or RoTTR). core cycler dosen't equal stability. your problem is a down load. how high are our temps.?
what kind of whea errors are you getting?

 

[authorized_to_ill]

reset to docp settings and now test. leave off any CO settings. my chip dose not any combination of CO and pbo overclocking. you might be in the same boat. i have ran core cycler over night without any errors. my main testing platform is any tomb raider(SoTTR or RoTTR). core cycler dosen't equal stability. your problem is a down load. how high are our temps.?
what kind of whea errors are you getting?

According to HWinfo, when I use Cinebench R23 my max temp on CPU (Tctl/Tdie) = 86, CPU Die (average) = 85.6, and CPU CCD1 (Tdie) = 87.5. When I adjusted PBO (when I thought I was stable) my max temps were in the low 70s.

For whatever reason attempting to download a game through the MS store consistently made my PC restart when I had the PBO and CO settings adjusted (and DOCP at 3600 FCLK at 1800).
Event viewer showed the following:

"A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Bus/Interconnect Error
Processor APIC ID: 0"

Resetting the bios and bring the DOCP down to 3200mhz and FCLK to 1600 made the issue go away.

 

[outpt]

That’s the memory controller, infinity fabric error. Memory problems are beyond my knowledge.

 

[authorized_to_ill]

That’s the memory controller, infinity fabric error. Memory problems are beyond my knowledge.

I kept assuming that the FCLK power was revealing an issue with the CPU. I guess I need to research more into the potential issue with the memory itself. Either way, thank you for trying to help.

 

[natr0n]

My chip is hitting 5050 and i even reduced volts -1.14

I can go to 1900 mclk fclk uclock too, Last chips would do 1833

have more testing to do. also extreme excite

 

[cobhc]

Can anyone explain to me why setting the same power limits results in different core clocks on different motherboards, as well as different voltages? I'm genuinely interested.

My 5800x on an Asus B550 Strix-E using 95/60/90 results in the chip drawing about 1.1V and only boosting to 4.1ghz or so in Cinebench R23 compared to the 1.2v and 4.5ghz clock that @Mussels gets in the OP.

Any knowledge on this would be appreciated, especially if I can change any settings to do with this?

 

[freeagent]

Can anyone explain to me why setting the same power limits results in different core clocks on different motherboards, as well as different voltages? I'm genuinely interested.

My 5800x on an Asus B500 Strix-E using 95/60/90 results in the chip drawing about 1.1V and only boosting to 4.1ghz or so in Cinebench R23 compared to the 1.2v and 4.5ghz clock that @Mussels gets in the OP.

Any knowledge on this would be appreciated, especially if I can change any settings to do with this?

Are you using the curve optimizer? if not you should be

 

[cobhc]

Are you using the curve optimizer? if not you should be

I have, but I feel like I'd need to be at -40 to get the clocks up to be near stock with the power limits set as I mentioned.

 

[mtcn77]

Can anyone explain to me why setting the same power limits results in different core clocks on different motherboards, as well as different voltages? I'm genuinely interested.

My 5800x on an Asus B500 Strix-E using 95/60/90 results in the chip drawing about 1.1V and only boosting to 4.1ghz or so in Cinebench R23 compared to the 1.2v and 4.5ghz clock that @Mussels gets in the OP.

Any knowledge on this would be appreciated, especially if I can change any settings to do with this?

It is because pbo encounters this as a three pronged solution and temperature holds the higher priority. All things being equal, just by plugging the chip on a "faster", but hotter motherboard will lead you to a lower pbo setting.

 

[cyberloner]

My chip is hitting 5050 and i even reduced volts -1.14

I can go to 1900 mclk fclk uclock too, Last chips would do 1833

have more testing to do. also extreme excite

cut voltage and higher boost mhz doesn't means you a re running the cpu top performance...... benchmark vs the default cpu you will know

 

[Mussels]

Wow is 1.5v not high as fck for ZEN 3?

If not then I have loads of headroom, my chip only boosts on 1.41v of voltage.

For single core boost its fine, as its the total amperage that is a danger to zen 3 - not peak voltage

all core at 1.5V would be a nice roasted CPU

Can anyone explain to me why setting the same power limits results in different core clocks on different motherboards, as well as different voltages? I'm genuinely interested.

My 5800x on an Asus B550 Strix-E using 95/60/90 results in the chip drawing about 1.1V and only boosting to 4.1ghz or so in Cinebench R23 compared to the 1.2v and 4.5ghz clock that @Mussels gets in the OP.

Any knowledge on this would be appreciated, especially if I can change any settings to do with this?

Slight variance between CPU's is normal, and then boards have different sensors that give different readings=
If you're getting lower clocks, temps and voltages - raise the limits another 10%

core offset also changes things, and may result in lower clocks (its magic and confusing, doesnt always work as expected)

 

[outpt]

2 questions
how do i test memory controller/infinity fabric for errors?
what should be the voltage on the 12v rail? i am seeing 11.976v under extreme stress in occt.
my cpu is at stock settings and i am still getting whea 41 errors.

nvm I goggled it

 

[jayjr1105]

Any suggestions for a 5600X on a MSI B450 Gaming Plus (same VRM as Tomahawk with smaller heatsink)

Mostly interested in good overall performance. Productivity isn't a big deal but game boosting would be nice.

RN I have it set to 95/80/120. My cooler is a ID Cooling SE 225 XT with good case airflow.