Question AMD Ryzen 5 5600G - CPU voltage and frequency fluctuates on idle

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

lukakalis99

Honorable
Oct 15, 2016
31
0
10,530
Hi all,

I have build a whole new PC couple of days ago. Everything fresh and new and out of the box. Nothing OCed.

Motherboard: AOURS B450 Elite v2
CPU: AMD Ryzen 5 5600G
RAM: 16GB DDR4 3200Mhz Kingston Fury Beast
PSU: MWE CoolerMaster 700W
SSD : 480GB Kingston
GPU: Nvidia GTX 1660

I have a problem where I noticed via AIDA64 and CPU-Z and HW Monitor all of them [to be sure] that I have big CPU voltage and frequency fluctuates on idle.
Like Voltage is from 0,924v to 1,126v and Clocks 2994Mhz up to 3114Mhz.

Keep in mind that when I run AIDA64 stress test the voltage jumps to 1,218v and Clock to 3897Mhz CONSTANT !

The temp on idle are max 38C and on full load AIDA64 test its max 64C.


I researched for days to check if this is okay, as I see very big FPS spikes in my games and I really don't like that as I payed a pretty good amount of $ for this build.
 

lukakalis99

Honorable
Oct 15, 2016
31
0
10,530
Probably an artifact or flaw in Aida64's monitoring.

Run HWInfo64 at the same time and graph CPU Die (average) temp and CPU (Tctl/Tdie) sensors and compare to those. I'd trust HWInfo first...but you also have to understand the difference between those two readouts, just hover over the sensor names and read.

FYI: When I had the old cooler [before placing this new Cooler Master EVO 212 Black]I never saw this spikes.
P.S. I ran this test on first time placing the Cooler then I pulled it of again to check for some thermal paste leakage or if it was to little, it was okay either whey I pasted a new layer of new cooler master gold paste. And again those spikes :

Okay just done it.
Please take a look : View: https://imgur.com/a/6GcVwpf
 
FYI: When I had the old cooler [before placing this new Cooler Master EVO 212 Black]I never saw this spikes.
P.S. I ran this test on first time placing the Cooler then I pulled it of again to check for some thermal paste leakage or if it was to little, it was okay either whey I pasted a new layer of new cooler master gold paste. And again those spikes :

Okay just done it.
Please take a look : View: https://imgur.com/a/6GcVwpf
The HWInfo readings look quite good...the core "max" readings hovering around 66C in the period it was open. Even Tctl/Tdie is good at 69C max, 63.1C average. If that was taken while running an Aida stability test I'd have to say it's very good, although I'm not familiar with 5600G processors. As far as the Aida temperature reporting, though, I imagine there's an anomaly in the way Aida64 reports temperature that allows those crazy spikes in the chart.

HWInfo is well known to work properly with Ryzen CPU's and the author keeps it current with frequent updates. I note your version is 7.12...7.14 is out now. Right click on a temp sensor and create a graph for it to monitor over time.
 
Last edited:
....
So I guess om okay then. I mean they are 1 second rear spikes but its reading a spike of 120c.
...
My processor (a 3700X) has a Tjmax of 95C , a thermal trip limit of 115C and a high temp cutout (HTC) limit of 115.5C. If any part of the processor ever spiked as high as 120C I'm sure it would shut down instantly. It's never done that, I hope the motherboard leaves fans and AIO pump running if it does.

That's almost certainly an anomaly in Aida's temp reporting.
 

lukakalis99

Honorable
Oct 15, 2016
31
0
10,530
My processor (a 3700X) has a Tjmax of 95C , a thermal trip limit of 115C and a high temp cutout (HTC) limit of 115.5C. If any part of the processor ever spiked as high as 120C I'm sure it would shut down instantly. It's never done that, I hope the motherboard leaves fans and AIO pump running if it does.

That's almost certainly an anomaly in Aida's temp reporting.


Okay my dude thanks :D
 
Any feedback maybe :D ?
Best overclock guide for Ryzen is...don't. Not in any conventional sense, at least.

The boost algorithm for Ryzen gets almost all the performance the silicon is capable of with ambient air cooling. The best you can do is use PBO with something called Curve Optimizer and slight undervolt using offsets. Just search on YouTube, several good how-to's are out there. Conventional, all-core fixed overclocking as often as not ends up hurting light-threaded performance important for gaming UNLESS you do heroic things with sub-ambient cooling.

The other thing is to realize the boost algorithm is temperature sensitive so improved cooling only helps performance. But with diminishing returns obviously: a Hyper 212 isn't really the 'best' anymore but it may be good enough for a 5600G. But that's if AMD left anything to get with their lowest low-end Zen 3 part.

Don't look at clocks expecting to figure anything out; look at bottom line performance with a good benchmark program. CB20 or 23 is the best one, both multi-thread and single-thread. Just be sure to get a good baseline first and run it in a consistent way: nothing else running in background and after a fresh restart to get most repeatable results.
 
Last edited:
  • Like
Reactions: Unolocogringo

lukakalis99

Honorable
Oct 15, 2016
31
0
10,530
Okay thank you all for the feedback, but also I was wondering since I have now 2 x 8GB of the same RAM can I add [without loosing any performance] a third 16GB stick with this 2 of 8GB [FYI its the SAME ram stick but its no 8GB its 16GB]
 

DimkaTsv

Commendable
Nov 7, 2021
171
25
1,640
Okay thank you all for the feedback, but also I was wondering since I have now 2 x 8GB of the same RAM can I add [without loosing any performance] a third 16GB stick with this 2 of 8GB [FYI its the SAME ram stick but its no 8GB its 16GB]
Even same brand - same title ram is not techincally same, unless binned by 4 sticks. Especially unless you have T-topology motherboard, which, i am pretty sure, you don't
So, yes, you can try to run it, but <Mod Edit> can happen and you should be prepared... Also feels like inefficient using of RAM slots there. 16-8-0-8 division may be even worse than 8-8-8-8...
 
Last edited by a moderator:

DimkaTsv

Commendable
Nov 7, 2021
171
25
1,640
@DimkaTsv Okay, sooo, what would be the best option then ?
Well.. either buy second 16 GB stick and try 0-16-0-16 configuration... You may hit problems, but shouldn't be much if they are same brand/serial number and if you just run them by XMP, so chances for troubles are far far less likely.
Or you can buy second 2x8 kit and try to get 8-8-8-8 config... But you may be forced to downclock RAM, because of topology stuff. If same brand/serial chances are less, but still exist.
But i definitely wouldn't just used 16-8-0-8 or 8-16-0-8, because of raw difference in ranking. 16 GB ram sticks are mostly 2 rank ones, and 8GB are mostly 1 rank. so controller have different take on them. And with daisy chain topology traces go through one stick slot to get to second (like 0-8 0-8, and you will make 16-8 or 8-16 in first channel and 0-8 in second channel),so you have much more chances to get incompatability this way...
Technically you can try 8-8-0-16 (or 0-16-8-8) first. if it will not work than just do 4x8 or 2x16

So it is more of a question of compatability. Just try it if you already have all 3 sticks, and if it will not work out... than decide on "do nothing" "4x8" and "2x16" choises