There is nothing to worry about Vcore 1,44V during bios. It could be even higher. In normal use (if not overclocked manual voltage) Zen2 adjust its voltages rapidly so it is not constantly so high. Constant vaues 1.35V might be actually problematic due silicon aging. Idle temps 49C is also not alarming so no worries either. Of course lower voltage and temperatures are better and I also use currently -offset 0,075V which drops temps, voltage and current consumption. I dont see perf drop until the offset is more than -0.100 but I like to have some margins with this. Zen2 with small 7nm process do cause higher temp peaks than generally used to, but it is still fine as default. I just want to see a bit lower values so small negative offset is applied. Just note that too high negative offset drop performance and occasionally also cause instability depending on chip&CPU combo. This performance drop is described as clock stretching (some overclocker has made clarifications about this behaviour). When CPU detects too low voltages it will delay clock periods and effective CPU speed drops even it is not read correctly in monitoring. User will see just lower perf in tests like CB20 .
yep, I'n not really "worried". I'm just under the impression the system isn't working the best way.
Using default settings Vcore values are quite often above the 1.35 V.
Is it dangerous ? I don't think so. But I can't see the point, since performance are affected.
I found that an offset of -0.100 V is the sweet spot for me. Maybe with colder temperatures I will try again to see if situation changes.
Normal voltage is 1.300V so 1.440V is too high for idle even my MSI board got that wrong and a lot of Asus even Maximus boards get this wrong.
@Leonoid007 I do not like AMD's PBO as a lot of other people complains about it's not really good optimized.
@Max(IT) the frequency kept jumping too much and even down to an all-core of 3.8GHz when everything was on auto and now I put in the Offset that it should have 1.300V with an offset of like +0.125V and I do all-core 3.99GHz (40x in Bios) in CB20 with no issues.
View attachment 165158
Excuse me, you put a POSITIVE offset in ?
You worte
+ 0.125V ...
Is that a typo ?
It's been 13 months since Matisse released. That being so, I'd expect you to know that up to 1.5V at low loads is perfectly normal to see in HWInfo. Only Ryzen Master sees true idle voltage, and up to 1.5V is allowed at lower current (and often necessary for shittier chips) for single core boosting to the advertised Max Boost Speed. At this point, we're beating a dead horse that should have stayed dead a long time ago.
If you apply a negative offset that begins to reduce light load Vcore below what your chip
needs for single threaded boost, you will start to see a loss of single thread performance. Whether that's worth it, is up to you. I'm not sure what it takes for
@Max(IT) to listen to what other people have said and realize that this:
View attachment 165159
...means
Auto. It does not mean 1.144V, 1.5V, a green unicorn, or whatever the hell happens to be displayed in that box to the left. It means
Auto. The BIOS suffers from the same monitoring-software-syndrome that plagues every software that isn't Ryzen Master. It simply takes an instantaneous reading of Vcore when it starts, and that's what you see. Much more often than not, that "value" is over 1.4V. But in this case, the reviewer on the X570-TUF happened to see 1.144V. No Matisse chip is sustaining its advertised Max Boost single thread with 1.144 goddamn volts. Neither is a Matisse chip going to sustain all-core at 1.440V out of the box, because that's suicidal for silicon life.
When you set a Vcore
offset , you are still affecting the Vcore at all times by that set amount regardless of what Vcore might be displaying at any given moment or load. Unless you
listen for once, open up HWInfo and watch your SVI2 TFN Vcore
while you are running CB R20, this is going nowhere and no one will be any wiser as to what's actually going on with your stock settings and your -0.1V offset.
Dude, I can be no expert in Matisse yet, but I'm an IT engineers since 1993 so I surely know how to do tests.
I didn't read that value, but the value reported also in HWInfo64, and during CB20 I saw the SVI2 TFN went up to 1.5 V.
You should learn how to speak with others without being offensive.
I checked SVI2 TFN under load and with an offset is 1.30 V all cores and 1.35 V single core. Without the offset it goes above 1.4 V (thus the higher temperature).
Now I did an extensive testing and I can report what I discovered.
I tried 3 configurations:
A) stock AUTO voltage
B) offset -0.150 V
C) offset -0.100 V (my "daily driver" )
Cinebech results aren't really affected much by the 3 settings, BUT temperatures are.
In option A I had these results:
which is ok but idle temperature is around 49° and under load I reached 85° C.
Options B and C were quite similar in term of results, so I will just report C screenshot:
as you can see, basically NOTHING changed, BUT (a BIG BUT no pun intended
) temperature at idle is about 43° and under load I had a spike at 79° but for most of the time it was around 70° C.
A huge difference.
As far as clock speeds well, the results where almost identical: in single core I had 4,541.8 MHz most of the time (with a spike at 4,616.7 MHz) and in multicore it stays around 4.1 GHz.
Now, I need advice (but not lecturing, please), but these are my results. And to be clear, I did AT LEAST 5 TESTS for each configuration, so be sure to report coherent results.
What do you think ?
