Hey man, just checking if you were able to find this? iii) Set the "UCLK DIV1 MODE" to "UCLK ==MEMCLK" ?
we sport the same board and can't find the equivalent for that.
3200 32GB Trident Neo
I hope the help I gave you on Discord has borne fruit.
Discord is a lot easier to discuss this sort of thing than a forum that I only occasionally look at (ego surfing is not part of my psyche ).
I wasn't able to, but I am running stable and low temps just by simply undervolting vcore (VDDCR CPU Voltage) by setting it from Auto to Offset and then inputting -0.12 (I believe), and still maintain normal single core boosts. In other words, no overclocking, no messing with Ryzen Master, etc. This brings vcore down from hovering at 1.45v at idle, when it's got no business being that high, to hitting 1.33-1.35v~ max.
Other people, even with the exact same board as the OP have tried following his instructions and had no luck, or very different results. Their machine just refuses to post, or they have unforeseen issues. So trying it on a completely different board (and different CPU) seems out of the question to me. I'd rather just use undervolt with the offset and let the CPU run the way it's supposed to.
Got it. Thank you. I've been fiddling around since this is just the 2nd AMD system that I build. The first one was way back 2012 and oh boy, never expected for it to be this complicated to OC. Been an intel ever since but tide has turned.
If you are running that way, you can always try out the EDC bug and see what that does for you.
Some fun time info on that.
My guide is specifically NOT overclocking the CPU.
All the configuration suggestions I have made remain within the spec of the 7nm architecture as stated by TSMC (a maximum of 1.3 Volts).
That I get better results for lower temps and lower power consumption than those who actually do overclock is an added bonus.
Had to clear CMOS following this guide, thank you for that !!
My RAM is rated for 16-19-19-39 at 3600 (1800Mhz), but Ryzen Master showed 15-15-15-51, so I thought "hmm, I'll leave it for later". Everything worked fine but then, after successful reboot, I wondered "ok, maybe this guy knows what he's talking about ?". So I entered the 15-15-15-51 and, of course, it didn't boot anymore.
Also, after applying these settings the single core boost clock was lower and so was FPS in single-threaded games.
DO NOT FOLLOW THIS. Probably my fault for following random guy's guide on the internet.
(I Have a Ryzen 3900x and a gigabyte X570 MB, updated drivers, bios etc)
Ryzen Master doesn't apply anything automatically on boot, if you set bad timings in the software it shouldn't prevent the system from rebooting normally after crashing and you certainly shouldn't have needed to clear your CMOS. You're going to need to be specific about what you did so we can figure out how you broke it.
And perhaps OP didn't make it clear enough, but maximising FPS isn't the goal here. While undervolting does lower temperatures and allow for higher sustained speeds under load, the current limitations of Zen2 and Ryzen Master prohibit per-core overclocking, so single-core performance is something we sacrifice in favour of all-core performance and preventing degradation. I've seen reports elsewhere of chips degrading at just 1.325 volts. If you value single-core boosts above all else then congratulations, you don't have to do anything.
On a bit of a side note, if your RAM is rated for 16-19-19-39 @3600MHz but your software is reporting faster timings, that suggests you didn't actually enable DOCP and it was only running at the stock 2133MHz and timings. You should double check that DOCP is enabled in your UEFI. However, whether it's set to stock or DOCP shouldn't make a difference to how stable it is when you set the exact same timings manually, RAM is just strange sometimes.
In my case, I found that manually setting the timings and leaving them on DOCP made no change to stability, so I stuck with DOCP. From the OP's other comments ITT this doesn't seem unique to me and it might work for you if you want to try undervolting again.