Excuse me, why did my SOC voltage and memory frequency suddenly spike? Thank you, everyone.

Excuse me, why did my R7 9700X SOC voltage and memory frequency suddenly spike? Thank you, everyone.
BIOS Version: 3222 2025/03/07
Updated AGESA to PI 1.2.0.3a Patch A
Sorry, my English is not good. These are all translated from Chinese by AI.
Snipaste_2025-05-01_23-48-01.png


Snipaste_2025-05-01_23-56-51.png
 
Probably same problem as this:
 
Probably same problem as this:
Thank you. May I inquire whether memory frequency fluctuates in real-world usage scenarios? If such dynamic adjustments occur, would the system remain stable and functional throughout these changes? Your technical insights would be greatly appreciated. Thank you for your attention to this matter.
 
Thank you. May I inquire whether memory frequency fluctuates in real-world usage scenarios? If such dynamic adjustments occur, would the system remain stable and functional throughout these changes? Your technical insights would be greatly appreciated. Thank you for your attention to this matter.
Yes they do but not adding 50% of speed and voltage. If that was real the CPU would be fried by now.

There are a few settings in BIOS that when enabled, and if the system conditions allows it, memory speed and UCLK can drop speeds to save energy.
In your case it would drop from 3000 to 1200~1500 for example. But no way adding 50% on top of top speed.
Also FCLK cannot run stable above 2100~2200 on any CPU and even very few can achieve above 2100~2133.

So this is clearly an error bug on some sensors when reporting from the CPU to HWiNFO.
When I had my AM4 system, some times I had exactly double (+100%) values on the same sensors as you.
After a point, after a few HWiNFO updates and also board BIOS upgrades the double values were gone and I had only +6.67% values.
I still have them (the +6.67%) from time to time on my current AM5 system with the 9700X.

Another observation
The SOC voltage is set to 1.24~1.25V as I see it.
This is a bit high for your speeds. Its safer to lower it under 1.2V

My setting is 1.185V in BIOS and HWiNFO reports around 1.165V

2025-05-02.png
 
Yes they do but not adding 50% of speed and voltage. If that was real the CPU would be fried by now.

There are a few settings in BIOS that when enabled, and if the system conditions allows it, memory speed and UCLK can drop speeds to save energy.
In your case it would drop from 3000 to 1200~1500 for example. But no way adding 50% on top of top speed.
Also FCLK cannot run stable above 2100~2200 on any CPU and even very few can achieve above 2100~2133.

So this is clearly an error bug on some sensors when reporting from the CPU to HWiNFO.
When I had my AM4 system, some times I had exactly double (+100%) values on the same sensors as you.
After a point, after a few HWiNFO updates and also board BIOS upgrades the double values were gone and I had only +6.67% values.
I still have them (the +6.67%) from time to time on my current AM5 system with the 9700X.

Another observation
The SOC voltage is set to 1.24~1.25V as I see it.
This is a bit high for your speeds. Its safer to lower it under 1.2V

My setting is 1.185V in BIOS and HWiNFO reports around 1.165V

View attachment 12652
Thank you for your explanation, it is very helpful to me.
Just now, my Soc voltage reached 1.306, exceeding 1.3 again, and FCLK reached 2125MHz.
I will set the Soc voltage to 1.2V in BIOS according to your suggestion.
Is there any way I can test the stability after setting?
Thank you very muchSnipaste_2025-05-02_22-15-11.png
 
Yes that was again erratic reporting from the CPU to the software.
You can see it also at the sensor right below SOC voltage, the "CPU VDD_MISC (SVI3 TFN)" going from 1.100V to 1.169V, which is exactly by the the same (proportional) amount. around +6.2~6.3% in your case.
If you add that persentage you'll find the error values.
1.100 >> 1.169
1.229 >> 1.306
2000 >> 2125
3000 >> 3188
From what I understand, these erratic reports of sensor values do not affect system stability in any way. Internally the CPU is reading the correct values and self-adjust and self-regulate accordingly.

As for testing system stability, the usual tests and benchmarks that most people use. Also a test to stability is using your system normaly to your every day tasks. Whatever is that you do, including gaming.
Gaming can bring out weaknesses. Especially the demanding ones, like Cyberpunk_2077 for one example.
 
Yes that was again erratic reporting from the CPU to the software.
You can see it also at the sensor right below SOC voltage, the "CPU VDD_MISC (SVI3 TFN)" going from 1.100V to 1.169V, which is exactly by the the same (proportional) amount. around +6.2~6.3% in your case.
If you add that persentage you'll find the error values.
1.100 >> 1.169
1.229 >> 1.306
2000 >> 2125
3000 >> 3188
From what I understand, these erratic reports of sensor values do not affect system stability in any way. Internally the CPU is reading the correct values and self-adjust and self-regulate accordingly.

As for testing system stability, the usual tests and benchmarks that most people use. Also a test to stability is using your system normaly to your every day tasks. Whatever is that you do, including gaming.
Gaming can bring out weaknesses. Especially the demanding ones, like Cyberpunk_2077 for one example.
Thank you so much, you have benefited me a lot.
I followed your advice and set the Soc voltage to 1.21, which will maintain 1.2 in HWINFO
When I was playing <Black Myth Wukong> just now, there was a huge voltage fluctuation again.
But it hasn't exceeded 1.3V anymore.
At least psychologically acceptable.
Snipaste_2025-05-03_01-04-52.png
 
You will get over it.
The acceptance really comes at play when you're realizing that these reported values are not real.
SOC and MISC voltage, MemClock, UCLK, FCLK, are not suppose to fluctuate that much and they dont in reality.
You'll get used to it even if you see 2.4V at SOC.
In fact if you do see 2.4V SOC you'll get over it sooner becasue anything real above 1.4V could fry it instantly.
 
Back
Top