X370 Taichi VCore nasty spikes

Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic

 baskura wrote:

The voltage spikes you are seeing are NORMAL for Ryzen and you should not be concerned of spikes into the 1.5v range when using the Auto setting for V-core voltage.
What is causing these spikes? It's XFR (Extended Frequency Range) which is a feature of Ryzen which is ONLY enabled when running at stock (i.e. Auto settings). 
...
If you overclock your Ryzen CPU/lock your voltage in any way, XFR is disabled which is why you no longer see the voltage spikes. 
...
This is not an Asrock bug, it's a Ryzen feature, it happens on both my Asrock Taichi and Asus Crosshair VI.
...

From what I read, I am in agreement: Max vCore of ~1.55v under single core XFR boost is by design.

What Gigabyte did when they killed a bunch of chips with high vCore ~1.70v, I do not know.  Was it on auto mode?
What I can say is that from P3.00 to P3.10 on the TaiChi went from 1.55v to 1.50v max. vCore under XFR single-core.

You say "lock your voltage" = disable XFR.  That's not true.  As long as you don't change clock speeds or actively disable CPB, then you can indeed lock or offset the vCore and XFR will still work fine.  
I set my vCore to fixed 1.30v and XFR worked fine but vCore SVI2 TFN did not scale down when idle.
Offset mode is working for me with -0.10v and forcing LLC to L4: I see a max. of 1.369v and it drops under idle.

It is a fact that voltage is always kept on the higher than optimum setting so that all chips that leave the factory will work at their advertised max. specs.  For the vast majority, the chips will work below the auto voltages.  It's been the same for Intel, AMD and anyone else.
My old Intel Core2Duo used more volts than necessary, and I could overclock 25% on less than the stock auto max. volts!

One thing that seems to be true is that these modern CPUs are very hard to read data from in the OS, and specifically Windows.
They are self-regulating and so are doing things at rates the OS beyond what the OS can report.  I guess results are sort of averaged out, or a mini-snapshot of the real truth.
The on-chip SMU may request a vCore, may or may not actually get it, and may distribute less than this voltage to the internal parts of the chip (core, cache etc.).  I have a feeling much of the SMU's control is removed when the clock frequency is altered and the chip is made 'dumb' and will stick to distributing voltage by a table or by whatever comes in - in the case of overclocked frequency, whatever the vCore in BIOS is will go to the cores.

It's still a little annoying that AMD aren't just completely clearing this up once and for all.  Of course, they don't want to give out too much info on how their intellectual property works, but this is just general user information.
If they have stated that with overclocking don't go over a max of 1.45v, then it looks very bad to the uneducated when they see 1.55v and they aren't even overclocking!

C'mon AMD, educate us :)