Bizarre, Low Temperatures on 1950X while OC'd to 4.0 GHz (Lower Than Ambient?)

Setup:
CPU: AMD Threadripper 1950X
Cooler: Corsair h115i AIO Liquid Cooling
Mobo: ASUS Zenith Extreme
RAM: 4x8GB, Corsair 3000 MHz
GPU: ASUS GTX 1080 Ti Turbo
PSU: Corsair HX750i, 750W





 

After loading the pre-set 4.0 GHz profile in BIOS for the ASUS Zenith Extreme mobo with a 1950x installed, and changing the voltage to 1.3875 using Ryzen Master (was unstable at the default OC voltage of 1.35 v), my temperature readings have dropped dramatically—even lower than my ambient temps.  I'm wondering if this low voltage, which is perfectly stable running Cinebench and Prime95, is not providing the sensors with correct temp data to report?

 

It looks like when the CPU is not under any load, the cores drop down to 2.2 GHz with an insanely low voltage of 0.8875—this seems to be when I'm getting the crazy low temperature readings.  Under load, one or more cores clock at 4.0 GHz and the voltage goes to the set 1.3875.  However, even then, the temp after a multi-core run of Cinebench 15 yields 36.5 C in Ryzen Master at about 25 C ambient.

 

Ignore the low memory clocks—I don't have DOCP enabled.

 

(Frequency and voltage while Idle)

Img1.jpg

(Frequency and voltage when at least one core working)

Img2.jpg

 

Cinebench 15 scores show the processor working extremely well, ranging from 3100 to a high of 3311 while OC'd at 4.0 GHz.

 

Img 3.jpg

 

The ASUS mobo normally displays the tCTL CPU temp on the onboard OLED, but it's now giving me ridiculous readings of 9 C or so at idle.  Earlier, the OLED was displaying CPU error codes while in Windows, even while the PC was functioning perfectly (Codes 0C and then D3 "Check CPU").

 

IMG_20170819_135603.jpg

Anyone know what on earth is going on?