The AMD Zen 3 Ryzen 5000 Discussion Thread

[Rogue]

i just upgraded to 5600X from 3800XT. Reset bios setting and only enabled DOCP and set PBO to enabled.

Just wondering if i got a bad chip - ran heavyload and hwinfo together overnight to test if it's stable but somehow it crashes, monitor shows black screen when i woke up in the morning. Got to power cycle my rig before it can power up again. Anyone have the same issue with Zen3 chip? Or i need to reformat my system?

Adequate cooling or "stock"? Singapore is not known for mild climate... Have a saved record of temperatures/clockspeeds throughout test period and look look see see.

 

[aikishi86]

Adequate cooling or "stock"? Singapore is not known for mild climate... Have a saved record of temperatures/clockspeeds throughout test period and look look see see.

I'm using arctic liquid cooler. usually idle around 35-39degC, normal ambient without AC. With AC on at night, it goes all the way down to 31degC lowest. And it was about 60degC when i was playing cyberpunk 2077. I didnt get the temp history yet when running heavyload + hwinfo as i need to be in office today. Let's see if i can retrieved it tonight.

Btw, i just realised that some discussion on top that hwinfo caused sudden reboot or shutdown. I'm not sure if it's the cause of my issue. I think i'm still using ver. 6.XX. I'm going to try tonight when i update to hwinfo 7.02 if it still happen.

 

[soltrdc]

i just upgraded to 5600X from 3800XT. Reset bios setting and only enabled DOCP and set PBO to enabled.

Just wondering if i got a bad chip - ran heavyload and hwinfo together overnight to test if it's stable but somehow it crashes, monitor shows black screen when i woke up in the morning. Got to power cycle my rig before it can power up again. Anyone have the same issue with Zen3 chip? Or i need to reformat my system?

Try undervolting. My undervolted 5600X runs 24/7 at 4.6 GHz and barely hits 70C. I'm using a 240mm Jonsbo AIO.

Edit: sounds like no brainer but just in case, make sure you have a good spread of thermal paste and seating of the heatsink!

 

[Mach3.2]

I'm using arctic liquid cooler. usually idle around 35-39degC, normal ambient without AC. With AC on at night, it goes all the way down to 31degC lowest. And it was about 60degC when i was playing cyberpunk 2077. I didnt get the temp history yet when running heavyload + hwinfo as i need to be in office today. Let's see if i can retrieved it tonight.

Btw, i just realised that some discussion on top that hwinfo caused sudden reboot or shutdown. I'm not sure if it's the cause of my issue. I think i'm still using ver. 6.XX. I'm going to try tonight when i update to hwinfo 7.02 if it still happen.

You using negative offset on curve optimiser?

Could be your undervolt not stable, try to reduce your negative curve offset.

 

[aikishi86]

You using negative offset on curve optimiser?

Could be your undervolt not stable, try to reduce your negative curve offset.

Actually i didnt overclock or undervolt. I only set DOCP or PBO enabled. The rest all at stock

 

[Mach3.2]

Actually i didnt overclock or undervolt. I only set DOCP or PBO enabled. The rest all at stock

Hmm that's odd. I'm out of ideas, but could be hwinfo as you've prostulated. Do keep us posted.

 

[tatose]

Try setting "Core Performance Boost" in your uefi/bios to disabled. Sometimes motherboard makers set unrealistic power values as default. You might also need to disable PBO (not enable).

 

[soltrdc]

Undervolting helps with temps, cos stock settings give a generous allowance for the voltage curve while you can run the same clocks with lower voltage. DOCP shouldn't be an issue as it's usually within stable limits. Furthermore 5600X only has one CCX, eliminating one avenue where infinity fabric comes into play. But just to be sure you might want to try swapping the RAM for another set to try.

Since we're on the topic, here are my undervolt/PBO settings for reference:

PBO limits: manual
PPT 104W TDC 68A EDC 102A
Max boost, scalar, max temp all Auto
Curve optimizer: per core, negative 10 on preferred cores (use ryzen master to identify), negative 15 on the rest

This effectively allows me to maintain 4.6 GHz on a full load without going above 70C on a 240 mm AIO. For the AIO radiator I'm using Cooler Master SickleFlow fans with fan curve of 50% baseline, 60C/70%, 75C/100%.

 

[soltrdc]

Agree with this (1)

Try setting "Core Performance Boost" in your uefi/bios to disabled.

Not with this (2)

Sometimeses motherboard makers set unrealistic power values as default. You might also need to disable PBO (not enable).

For (1), it's a customization that some motherboards BIOS offer that they sustain boost clocks for longer time by sustaining voltage for longer time. True that disabling this can sometimes help temps/stability, especially if you are also running AMD own PBO

For (2), "unrealistic" power limits are a myth. I have so far used a 5900X on an Asus B550 and a Gigabyte X570 board. The Asus one sets max PPT to 395W, the Gigabyte one 1200W. In both cases, the chip automatically limits itself to about max 200W with PBO on, all Auto. In other words, motherboard limits are just what the motherboard is capable of providing, the CPU will not use more than what it needs. How much will it use is a combination of what limits you give it and cooling. If you can cool it infinitely, then it will boost to it's max infinitely. Basically that's it.

If you don't want any overclock, I suggest still turn on PBO, set the PPT/TDC/EDC to the default values (these can be Googled, or simply open Ryzen Master with PBO off and switch to advanced view to see the values), leave everything else auto, and use curve optimizer to undervolt, it will certainly help temps.

Note: while I referenced the 5900X above, it will apply for any Zen 3 chip. Am currently doing the same on a 5600X. The only difference is in 5900X and 5950X there are two CCX so have a couple more options to play with.

 

[tatose]

Agree with this (1)


Not with this (2)


For (1), it's a customization that some motherboards BIOS offer that they sustain boost clocks for longer time by sustaining voltage for longer time. True that disabling this can sometimes help temps/stability, especially if you are also running AMD own PBO

For (2), "unrealistic" power limits are a myth. I have so far used a 5900X on an Asus B550 and a Gigabyte X570 board. The Asus one sets max PPT to 395W, the Gigabyte one 1200W. In both cases, the chip automatically limits itself to about max 200W with PBO on, all Auto. In other words, motherboard limits are just what the motherboard is capable of providing, the CPU will not use more than what it needs. How much will it use is a combination of what limits you give it and cooling. If you can cool it infinitely, then it will boost to it's max infinitely. Basically that's it.

If you don't want any overclock, I suggest still turn on PBO, set the PPT/TDC/EDC to the default values (these can be Googled), leave everything else auto, and use curve optimizer to undervolt, it will certainly help temps.

Well, on theory it should reconcile to the thermal limits. However, we have cases of whea errors because of firmware/hardware stuff. You might get lucky with default settings from motherboard maker, not everybody is lucky.

 

[soltrdc]

Well, on theory it should reconcile to the thermal limits. However, we have cases of whea errors because of firmware/hardware stuff. You might get lucky with default settings from motherboard maker, not everybody is lucky.

The point is in a typical desktop setting, the CPU never hits or even approaches those limits. So if you're getting WHEA errors, I highly doubt it's because of the mobo power limits, which in the case of PBO is specifically the 3 parameters of PPT/TDC/EDC. Other parameters such as boost scalar, max boost override, max temp can also be controlled via motherboard but those limits are capped at the same as AMD's own, i.e. max scalar 10X and max boost override 200MHz.

Enabling PBO and manually setting limits to the default, ensures you are getting the stock out-of-the-box limits + the benefit of undervolting to improve temps. Improvement of temps also helps with stability, since the CPU clocks don't yo-yo between hitting the max capability of your cooler and what it has the potential to do. In the Intel camp, there's the phenomenon where the i9 10900K hits a higher max boost than the 10850K but overall sustained clock is lower by about 100MHz because that's exactly what happens.

 

[tatose]

Point is, motherboard default is actually not at the appropriate settings. And they are unrealistic considering AMD official spec. The normal user will not think of googling the official values and manually set them. So many issues with Ryzen3000/5000 arises from this.

 

[soltrdc]

Lol, I thought that's the point of these forums, as are many avenues like other forums and Reddit. What's the point of discussion if we don't learn anything? Well, I'm just sharing from my experience and what I learn elsewhere as well, a (free) way to improve performance. PBO is baked into Ryzen, you paid for it anyway so why waste it. Anyway, to each his own, your rig your call. I put the info out there, it's for anyone who reads to do what they wish with it

 

[soltrdc]

Reference image for what I was referring to earlier. This is a steady state after 15 min of 100% load. I remember wrongly, it's not 4.6 GHz steady but 4.5 GHz. In other words, by undervolting and playing with just the PPT/TDC/EDC, I get a constant boost of just 100MHz below the max stock boost of a 5600X. The CPU at this point is not thermally limited but power limited, hence the 100% of PPT/EDC. CPU power at this point is 73.5W, 8.5W higher than the 65W stock TDP of the 5600X and temperature is steady at 70C. My radiator fan curve is 70%/70C to 100%/75C, so this is also not pushing the limits of my cooling potential.

Now this is very different from overclocking, because with PBO it is not a static 4.5 GHz. The CPU is still able to reduce clocks or even set cores to sleep when cores/threads are not loaded. Btw do not be alarmed at the red colors of PPT and EDC, the color scheme simply is applied according to how close the values are to the limit. Even at stock, these will show red when the CPU hits the (low) limits. Undervolting plays a big part since it literally reduces the heat produced at the same clock speed as a non-undervolted core. I only apply a modest undervolt of -10 for preferred cores and -15 for the rest, which translates to a range of -30 to -50 mV preferred cores and -45 to -75 mV for the rest. This is the other genius part of PBO undervolting - it doesn't apply a fixed offset but a range, allowing the CPU some wiggle room in terms of stability. I run this CPU at 100% load, 4.5 GHz, 70C for a full 24 hours and nothing is going wrong, no crash no WHEA no whatever. Pretty sure it'll go for days without issue. I even have room on my fan curve to cool the CPU even further if I want to, but higher speeds = more noise and I think 70C is very decent and well within the limit of the default thermal throttle temp (95C).

Okay, I think this is a lot of info on PBO for Zen 3 already. Again, it's just for reference. Whatever anyone does with their CPU is none of my business, just sharing some information which might be useful for someone.

 

[aikishi86]

Seemed like updating the hwinfo to 7.02 works, was only on 6.40 previously... max temp running heavyload with hwinfo was only 77.6. Idling between 33-37degC room temp. Down to 31degC when i on my AC

Was playing abit of RDR2 last night too. Max temp only hitting between 58-62degC. Guess this issue is "resolved" for me now. Thanks all for the advice.

 

[soltrdc]

Congrats, glad you solved the issue!

So it wasn't even a hardware problem to begin with but an app-specific one. I'm gonna guess it's a memleak crash, but it's just speculating. End of the day, good to know that the hardware is doing fine.

 

[The_King]

anyone got any advise on Asus ProArt B550-CREATOR?

Is it just because of better I/O? or it claim of superior 24/7 of quality, stability and compatibility?

thanks

 

[midlusionz]

Those with USB issue got it sorted out with new bios firmware?

 

[soltrdc]

I had the USB issues a few months back. Have been keeping updated with all BIOS release, and currently have no such issues anymore. My previous board was Asus ROG Strix B550-I with BIOS 2201. Current board is Gigabyte X570 Aorus Master with BIOS F33i. My keyboard/mouse are all using Logitech Unifying receiver. Both boards are fine, no USB dropout issues anymore.

 

[midlusionz]

I had the USB issues a few months back. Have been keeping updated with all BIOS release, and currently have no such issues anymore. My previous board was Asus ROG Strix B550-I with BIOS 2201. Current board is Gigabyte X570 Aorus Master with BIOS F33i. My keyboard/mouse are all using Logitech Unifying receiver. Both boards are fine, no USB dropout issues anymore.

ah nice! now whenever my usb stuff not working i get paranoid haha but turn out to be some other issue haha