[SOLVED] 5900X Vcore at 1.53v (not VID)

Nov 8, 2021
11
2
15
Hello,

The screenshot shows the result after a few hours playing Civilisation VI. My 5900X has an undervolt using Curve Optimiser per-core, stock power limits, and no 'AutoOC' or additional frequency. I had to undervolt in order for the Vcore to stay under 1.5v as, for some reason, a few days ago it started wanting more (VID was always over 1.5v and Vcore always over 1.5v). I'm curious about the VID (why is the CPU requesting such high voltage when it's meant to be hardware locked?) and concerned about the SVI2 TFN reported figures - I don't have the image, but I had a 10 hour session and Vcore and SVI2 TFN both showed over 1.5v (1.53v), unlike the screenshot below where SVI2 TFN and Vcore don't match.

System specifications are in the screenshot.

What I've tried:
  • Loading Optimised Defaults
  • Disabling every single overclocking function I can find
  • Run a 15 hour all-core stress test with an AVX2 load (I use AVX2 frequently) and the clocks averaged ~4.3 GHz and SVI2 TFN ~1.37v.
Any help would really be appreciated.

Screenshot-4.png
 
Solution
Hello,

The screenshot shows the result after a few hours playing Civilisation VI. My 5900X has an undervolt using Curve Optimiser per-core, stock power limits, and no 'AutoOC' or additional frequency. I had to undervolt in order for the Vcore to stay under 1.5v as, for some reason, a few days ago it started wanting more (VID was always over 1.5v and Vcore always over 1.5v). I'm curious about the VID (why is the CPU requesting such high voltage when it's meant to be hardware locked?) and concerned about the SVI2 TFN reported figures - I don't have the image, but I had a 10 hour session and Vcore and SVI2 TFN both showed over 1.5v (1.53v), unlike the screenshot below where SVI2 TFN and Vcore don't match.

System specifications...
Hello,

The screenshot shows the result after a few hours playing Civilisation VI. My 5900X has an undervolt using Curve Optimiser per-core, stock power limits, and no 'AutoOC' or additional frequency. I had to undervolt in order for the Vcore to stay under 1.5v as, for some reason, a few days ago it started wanting more (VID was always over 1.5v and Vcore always over 1.5v). I'm curious about the VID (why is the CPU requesting such high voltage when it's meant to be hardware locked?) and concerned about the SVI2 TFN reported figures - I don't have the image, but I had a 10 hour session and Vcore and SVI2 TFN both showed over 1.5v (1.53v), unlike the screenshot below where SVI2 TFN and Vcore don't match.

System specifications are in the screenshot.

What I've tried:
  • Loading Optimised Defaults
  • Disabling every single overclocking function I can find
  • Run a 15 hour all-core stress test with an AVX2 load (I use AVX2 frequently) and the clocks averaged ~4.3 GHz and SVI2 TFN ~1.37v.
Any help would really be appreciated.

Screenshot-4.png
When running a heavy load using AVX2 instructions it's highly unlikely to be running with such high voltage as the algorithm will pull back on voltage to keep core current and temperature reasonable. You can check that with HWInfo when running your application that uses AVX2 instructions.

But AMD's never given hard and fast specs for operating voltage either, probably since it's so dynamicaly varying in use. They've only said that it will raise voltage to 'as high as' 1.5V when it boosts a core to maximum clocks in light, bursty loads. So I think of that as a line in the sand and probably a good idea to prevent it from going above.

I had the same problem on my B550m TUF Gaming motherboard. I lowered VCore slightly with a negative offset until it never hit above 1.5V. Since then, with BIOS updates it's stopped doing it and I have Vcore set at full AUTO. Just do not ever use a fixed voltage.
 
Solution

TRENDING THREADS