Originally Posted by TahoeDust
I had never seen that. Good video. EVGA did a really good job with their voltage regulation. Very close to the "Flat" he said was kind of unrealistic. I only see a .014v voltage swing even when running Prime95 w/ AVX pulling from 20w to 260w.
Did you check this with an oscilloscope? (this will show transient spikes and dips)
Or even with reading the cpu on -die sense voltage?
Gigabyte also shows "flat" voltage the same way you do, if you read the ITE 8792E vcore sensor, which is the same readout if you use the onboard read points.
This reading is the MLCC caps reading (aka behind the socket).
But hooking up a probe to the VCC_Sense and VSS_Sense lines coming from the VRM show a completely different story--this measurement matches VR VOUT within 10mv.
If "75% less vdroop" on the eVGA board gives flat voltage readout from idle and even running prime95 small FFT AVX, then it's doing the exact same thing Gigabyte LLC Turbo is doing.
LLC Turbo is 0.4 mOhms of vdroop. Default vdroop for 8 core CFL is 1.6 mOhms.
0.4 mOhms is 75% lower than 1.6 mOhms.
The math matches right up.
So it's highly probable that eVGA is doing the same reading (socket MLCC).
Also, "No vdroop" cannot have a sustained load voltage higher than idle voltage. No vdroop is a 0 mOhm loadline.
Vdroop is very simple:
mOhms * Amps = vdroop.
But only Gigabyte and some MSI (and Asrock?) boards have amps monitoring from the VRM. (IR 35201 / intersil).