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

[Athlonite]

Obviously not because Ryzen Master starts at core 1.

Task Manager starts at 0

 

[pf100]

Task manager starts at 0. Bios starts at 0. Ryzen Master starts at 1. CPU-Z starts at 0.
You have to look at the number that whatever you're using starts at.
Almost everyone is missing that everything doesn't start at 0.
If you're using Ryzen Master or any other utility that starts at 1, and you're thinking "everything starts at 0" you will be wrong and you'll be doing things with the wrong core.

 

[Hoper]

5800x
x570 unify
CB-23 of 16050
5,050 ghz @ 38c idle @ 82c ish max load
WUT?

 

[Mussels]

5800x
x570 unify
CB-23 of 16050
5,050 ghz @ 38c idle @ 82c ish max load
WUT?

That is bang on for what a 5800x should do, with the +200Mhz in the BIOS


Hell, you can go back and find me with screenshots basically the same - even our stsock RAM speeds are similar

 

[freeagent]

Yup should be able to at least kiss 5050

 

[OneMoar]

I am not getting anything close to 1600 in r23 without cranking on the power limits
127/95/140 gets me barely 14k

 

[Mussels]

I am not getting anything close to 1600 in r23 without cranking on the power limits
127/95/140 gets me barely 14k

Checked your IF is 1:1?
Getting WHEA errors?

Yo ram be crazy fast, has me suspishus

 

[OneMoar]

never did update that ram in the specs its crucial micron e die 3600 CL 16
and yes its running 1:1
even with pbo limits on auto I barely clip 15k at 140w
not a single whea error out of this system ever since I exchanged the cpu

Even setting it to board defaults the score is barely 15k

other then cinebench being wierd it games all day at 4750-4800Mhz at <65c

 

[Mussels]

Uhh, cause you're hitting 90C and throttling?

That um, explains it all

Lower your PPT and use a curve, your cooling cant handle it.

 

[OneMoar]

naa doesn't matter I get about the same scores at 125/95/140 at <80c
I have the thermal throttle limit manually set to 90c

 

[DemonicRyzen666]

why does it seem like PPT means more than anything for temperature?

EDC and TDC don't change much on my chip, maybe all core over clock or single thread changes a little bit. Like not all cores are consistent at the same clock then. From 100w down to 90w ppt yields lower temps while scoring about the same in cinebenchr23.

Last night I tested the 5600x setting on here the 90w/ 75TDC / 100 EDC then tried 100w/75TDC/100 EDC score didn't change too much, temps were 10C lower with 90w PPT.
The tried 90w / 85TDC / 110 EDC scores about the same maybe only 5c lower temps.

I don't know if I should trust HWinfo, it shows that if I set 125w PPT but EDC at 120 it will stop at 120w PPT and won't go any further in HWinfo.

 

[OneMoar]

PPT is basicly personal pref as much as your cooler can handle because all-core is heavily affected by this
lower edc = higher boost clocks
tdc no idea what this does really

 

[Athlonite]

I have the thermal throttle limit manually set to 90c

I think it's HW64 showing wrong temps because I'm getting the same sort of reading and I know damn well it aint that hot

 

[Mussels]

why does it seem like PPT means more than anything for temperature?

EDC and TDC don't change much on my chip, maybe all core over clock or single thread changes a little bit. Like not all cores are consistent at the same clock then. From 100w down to 90w ppt yields lower temps while scoring about the same in cinebenchr23.

Last night I tested the 5600x setting on here the 90w/ 75TDC / 100 EDC then tried 100w/75TDC/100 EDC score didn't change too much, temps were 10C lower with 90w PPT.
The tried 90w / 85TDC / 110 EDC scores about the same maybe only 5c lower temps.

I don't know if I should trust HWinfo, it shows that if I set 125w PPT but EDC at 120 it will stop at 120w PPT and won't go any further in HWinfo.

PPT is the all core wattage limit, basically.

These chips lose efficiency after 100W, where the heat builds up fast with minimal performance gain - so it comes down to choosing a value between 100W and 142W (the chips max) that works within your cooling.
Overheating and hitting 90C is going to lower your clocks and performance
keeping the wattage lower is gunna be colder AND faster.

 

[DemonicRyzen666]

PPT is the all core wattage limit, basically.

These chips lose efficiency after 100W, where the heat builds up fast with minimal performance gain - so it comes down to choosing a value between 100W and 142W (the chips max) that works within your cooling.
Overheating and hitting 90C is going to lower your clocks and performance
keeping the wattage lower is gunna be colder AND faster.

I haven't gotten above 82C on prime95 with just Small FFT with AVX on, cinebenchr23 is 10C lower it's also like 300mhz higher lol
My current issues seems to be my VRM loop 1 is hotter then VRM loop 2 whey they should mirror each other. I wonder if it just has bad contact.
Just now change my EDC from 125 to 85 = better Cinebenchr23 score both single and Multi. guess I'll try to lower the wattage and see what it gets.

 

[Taraquin]

why does it seem like PPT means more than anything for temperature?

EDC and TDC don't change much on my chip, maybe all core over clock or single thread changes a little bit. Like not all cores are consistent at the same clock then. From 100w down to 90w ppt yields lower temps while scoring about the same in cinebenchr23.

Last night I tested the 5600x setting on here the 90w/ 75TDC / 100 EDC then tried 100w/75TDC/100 EDC score didn't change too much, temps were 10C lower with 90w PPT.
The tried 90w / 85TDC / 110 EDC scores about the same maybe only 5c lower temps.

I don't know if I should trust HWinfo, it shows that if I set 125w PPT but EDC at 120 it will stop at 120w PPT and won't go any further in HWinfo.

For 5600X there is little point going above 76W default limit, especially uf you use CO. Running -30 CO allcore and +50 pbo I get multicore boost in CB to 4.6-4.65GHz if temp is below 71-72C. If I set 90W limit consumption gets to 85W and allcore sits at 4.675-4.7GHz. This gives me 200points in CB23, but noise and heat is up significantly.

CB23:
- 200 pbo and CO -27,-29,-30x4, 90W: 76C, 88W, 4.75-4.6GHz, 12050p, voltage max 1.34v, fan rpm 1200.
-50 pbo and CO -30x6, 90W: 74C, 85W, 4.7-4.675GHz, 12050p, voltage max 1.22v, fan rpm 1200
-50 pbo and CO -30x6, 76W: 70C, 76W, 4.625-4.6GHz, 11850p, voltage max 1.22v, fan rpm 1200

I use the latter since sustained clocks are higher due to lower temp. My CPU gets a 50-100MHz drop in allcore frequency if temp goes from 71 to 76C.

 

[ntdouglas]

Is this in line with most 5600x.

 

[oobymach]

I found with the 5600x running -30 allcore gave a system freeze/fubar as soon as a demanding task ended. -25 with the +50 pbo runs fine though until it gets hot.

I stopped messing with pbo because of the throttling I kept running into, p95 second round is a killer when all workers are running it I got throttling regardless of the pbo settings.

Atm I'm running a 1.2v manual 4.65ghz and temps have never been lower, idling in the 30's and p95 maxes out at 62 degrees.

 

[oobymach]

Decided to try pbo again with the 5600x after running into issues with the 1.2v manual, running +200mhz with -8 allcore and 125watt 75a 100a and the maintained boost during p95 is higher now by over 200mhz and p95 happily solved 5 rounds of workers before I stopped it.

 

[ntdouglas]

Decided to try pbo again with the 5600x after running into issues with the 1.2v manual, running +200mhz with -8 allcore and 125watt 75a 100a and the maintained boost during p95 is higher now by over 200mhz and p95 happily solved 5 rounds of workers before I stopped it.

Nice. Try running occt and see what your all core is.

 

[oobymach]

Not familiar with occt, ran into an issue with the -8 allcore (system freeze but mouse still moved), tried -5 allcore but boost frequency was lower so now I'm testing with single core -5 -8 -5 -10 -10 -10, sustained boost is about the same with the -8 allcore, 4800-4850mhz on the first round of p95, 4675-4700mhz on the second round.

On auto all cores would drop to 4400-4500mhz on the second round.

 

[DemonicRyzen666]

seems like the curve just gives a more effect clock speed on all cores doesn't seem to do much for single thread for me.

 

[oobymach]

seems like the curve just gives a more effect clock speed on all cores doesn't seem to do much for single thread for me.

After looking into it more you're supposed to have your 2 best cores with the lowest undervolt, after checking ryzen master mine are cores 2 and 3 so I revised the +200 boost with -10 -5 -5 -10 -10 -10 and have been playing with it for a few hours, gaming, stressing, internet and watching vids and it's stable so far.

I had a bsod with the -5 -8 -5 -10 -10 -10 and a freeze with -5 -5 -5 -12 -12 -12 so still trying to find a reasonable stable curve.

 

[Taraquin]

After looking into it more you're supposed to have your 2 best cores with the lowest undervolt, after checking ryzen master mine are cores 2 and 3 so I revised the +200 boost with -10 -5 -5 -10 -10 -10 and have been playing with it for a few hours, gaming, stressing, internet and watching vids and it's stable so far.

I had a bsod with the -5 -8 -5 -10 -10 -10 and a freeze with -5 -5 -5 -12 -12 -12 so still trying to find a reasonable stable curve.

when you get bsod run event viewer in windows, you should find which core that failed there. No need to change the whole curve. I bet some of your cores can do anove -20.

seems like the curve just gives a more effect clock speed on all cores doesn't seem to do much for single thread for me.

Yeah, it does not affect single cores.

Curve optimizer only does the following:
At a given speed your CPU runs at a given voltage, negative CO tells the CPU to use less voltage at that given speed. If you use the standard powerlimit a 5600X runs at 76W max. Running stock if temp is kept below 70-71C (temp increases consumption) you can expect around 4.3GHz in cinebench. If you can run -30 on allcores like I do I get 4.6-4.65GHz allcore, in both cases it uses 76W. If temp rises to 75C speed drops to 4.55-4.575GHz.

You can remove the powerlimit and get a allcore speed of the same as your single core speed if cooling is good enough, but if you then use negative CO your consumption will drop. If I use -30 CO I can set powerlimit to 65W and the CPU runs at 4.3GHz allcore like it did without CO at 76W

 

[oobymach]

when you get bsod run event viewer in windows, you should find which core that failed there. No need to change the whole curve. I bet some of your cores can do anove -20.

That would be great only I have no Idea what to look for or where to look, or how to understand what I click on when I do find something, most of it looks like binary gibberish and I have no idea how to read binary.