Ryzen 3000 series support

Thanks for the info Ray62. On my X399 system my RAM reads at 1.37v in OS, on my
7600k system the same RAM reads at 1.345v, both ASRock boards. On the ASUS
board I had my 7600k in for a while the RAM read 1.38v and I am sure if I tried
it in a dozen different boards I would get a dozen different readings but all
only off by 0.0(x)v. This is perfectly normal but, in some cases where it reads
lower than XMP spec you will occasionally need to give the RAM 0.01v-0.04v more
to keep it stable. This is usually only the case with higher frequency RAM
kits, I have found I often have better luck with lower rated kits by actually
lowering voltage. My previous 2666 kit would not go over 2666 @1.35v no matter
what I did and increasing voltage made no difference. Lowering the voltage to
1.3 got me to 2933 stable and 1.25v got me all the way to 3200 on that kit.

The TLDR of it is simple, with RAM it pays to explore both sides of the
spectrum. Sometimes a lower voltage puts less strain on the SoC and IMC and
allows better stablity/overclocking, other times the RAM is just not able to
hit higher speeds at it's default voltages and needs a bit of a bump up. It all
depends on what is holding back the OC.

Both the RAM and the IMC are subject to the silicon lottery.

Edited by Xaltar - 11 Sep 2019 at 4:18pm

Irfaanm wrote: I'm on the 2.70 bios and still cannot run my ram at XMP 3200mhz :( B450m steel legend / r5 3600 Manual OC i can get it to 3134mhz

Happens to me2, if I set XMP I have a lot of BSOD's. I've used DRAM Calc to set it to 3200 with custom timmings.

That looks correct but you might get the boot problem all of us are having

There should be a new round of updates coming with AMD having just sent out a new AGESA.

Sweston wrote: That looks correct but you might get the boot problem all of us are having

Please don't post things like this in response to other people's queries, it
doesn't help them and only serves to scare them for no reason. While the boot
issue is fairly wide spread it isn't actually all that common. I have spent the
last 2 weeks testing a half dozen different boards from 300 series to 500
series and have not been able to replicate the issue on any of my test setups.
My suspicion is that there is another factor at play here, not just the
board/UEFI.

I feel for all of you having issues, I have spent a lot of time and effort
trying to help find solutions for you but alas, have not been able to produce
the issue on any of my test setups, even with components I have borrowed from
friends. I have tried every BIOS I can find for Ryzen 3000 on all the boards I
have to test with and none of them will cooperate, or I guess not
cooperate?

Given how many people have been effected by this issue (globally across all
brands it is a significant enough number to be a concern for AMD) I suspect the
next update may bring with it a solution.

Ray62 wrote: So what voltage was the RAM receiving instead? And, how did you check that? With measurement, directly at the RAM power supply contacts? With the bios H/W monitor? With a tool like HWInfo or Ryzen Master? ...

Using hwinfo since the bios monitor only seems to state the set voltage, and not the measure of actual dram voltage.
With XMP (1.35V), voltage range is 1.328V - 1.344V
With voltage manually set to 1.36V, voltage range is 1.328V - 1.360V
When manually setting the voltage to 1.37V, fluctuates between 1.344V to 1.370V
This behavior is not a big issue, since a simple increase in voltage resolves the instability, but for people who're not technically inclined, this would make them believe that their mb or ram is defective.

Xaltar wrote: There should be a new round of updates coming with AMD having just sent out a new AGESA.

Looking forward to the ABBA bios update
Hoping for some additional frequency boosts since my 3900x only manages 4.25GHz on all cores (single core boost is more respectable at 4.62Ghz)
However, not looking forward to the temperature increase it may entail.
Already hitting mid 80C with a dual radiator when on 75% cpu load