Did you get a shiny new Ryzen 3000 CPU?

[Roddey]

So testing around tonight,

4400 took 1.45 volts to stable, temps stayed low due to the water loop,
View attachment 131929

however, 4375 only takes 1.3 and is stable, temps stayed low again and barely anything in the cinebench scores.View attachment 131930

This is on an Asus board without any new bios, not sure if bios will help but 1.45v to get 4400 seems very high to me

The new ABBA bios gives me this on temps and voltage for a 3800x while running Cinebench 15. Better than the first bios for temps and voltage.

 

[HD64G]

So, the ABBA version agesa might be able to intervene with the auto UEFI settings for voltage and XFR and solve the trivial that some boards were doing with the vcore and clocks.

 

[Roddey]

So, the ABBA version agesa might be able to intervene with the auto UEFI settings for voltage and XFR and solve the trivial that some boards were doing with the vcore and clocks.

Sorry, I dont know how it works, just see results. But I am maintaining a higher sustained clock than the old agesa's would. I made a video showing the sustained clocks and voltage in a userbenchmark test. I noticed that in hwinfo I am getting to 1.488 volts at max boost. Different results than Ryzen Master.

.

 

[tabascosauz]

Sorry, I dont know how it works, just see results. But I am maintaining a higher sustained clock than the old agesa's would. I made a video showing the sustained clocks and voltage in a userbenchmark test. I noticed that in hwinfo I am getting to 1.488 volts at max boost but just for a short time

.

Max boost =! Max load. A core can be at 4.4GHz 1.4v as much as it wants, it won't mean much as it's not being loaded to 100%. When the stress test begins and seems to "catch" the CPU in its 1.488v low load state, it's more because of HWInfo's update interval. Zen 2 has a reaction time of 1-2ms.

Also, HWInfo, like every other software save for Ryzen Master, updates too slow to accurately reflect what Ryzen is doing on voltage.

You can still use offset to drop voltage if you want; it applies the offset broadly regardless of boost, activity or load. So at idle, you drop from 1.5v to 1.45v, at load from 1.3v to 1.25v, for example. The benefits that you will perceive come from the load Vcore reduction, not the idle drop.

 

[tabascosauz]

Okay, just updated to the F42c BIOS that has 1.0.0.3ABBA. Running the same PBO off, -0.075V as on F42a, and idle seems slightly more stable. CPU-Z is not benching single-core properly because the loaded core only runs the bench at x42.8 while said core can clearly hit x44 (the fastest core went from x43.5 to x44, and three other cores are now hitting x43.8 max).

One problem though, Ryzen Master and HWInfo are reporting different temperatures. This is really annoying. I trust Ryzen Master due to its update interval but HWInfo gives me a better idea of max/min and shows me whether WHEA errors are popping up. And this is on the new 6.12 release of HWInfo too. And the histogram issue on Ryzen Master remains, where if you want to use the histogram, Ryzen Master will keep one core at 4.2GHz at all times so your idle temps can't be lower than 45c. And it'll stay that way until you turn off histogram.

 

[X800]

So asus did let the new bios come to TUF B450-PRO GAMING BIOS 1820 Update AGESA 1.0.0.3ABBA .This broke the everything it did blue screen for 10 different things. I did clear cmos and loaded default still blue screens. I did have go back 2 bios versions now everything is back to normal.