Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
Hi Group,
Are there any "officially" approved measurement points
where I could measure Vcore directly on this board?
I've noticed that the utilities to measure Vcore always
return values in 16mV steps, so that I can only ever
read the following values:
1,280 1,296 1,312 1,328 1,344 1,360 1,376 1,392 1,408
1,424 1,440 1,456 1,472 1,488 1,504 1,520 1,536 1,552
1,568 1,584 1,600 1,616 etc.
Most programs also only show 2 decimals, so I didn't realise
this at first. It seems that is the resolution of the A/D converter
(or D/A converter?) or something, so I would like to connect
a high quality digital multimeter to the board directly to see
how stable my Vcore is.
I've noticed with my board that I always get a lower voltage
than the nominal value I set. The board has two settings:-
Vid (0,85-1,55V in steps of 0,025V) and Vid Adjust
(0%, 3,3%, 5%, 8,3% or 10%). The CPU voltage is
Vcore = Vid + Vid Adjust. I am not sure if Vid supplies
other components too, so that when I increase it I am
overvolting something else. Does anyone know?
Anyway, I've played a bit with the settings and plotted the
deviation from nominal. It seems the Vid adjust setting is more
accurate, perhaps the results are of interest, though of course
it could be inadequacies in my PSU (Seasonic 350W).
http://home.arcor.de/david_ffm/oc_psu1.jpg
I have also replotted my overclocking results with the voltages
modified to reflect the measured values rather than the nominal
values, and added a few data points. The curve of stability in
the space of Vcore against CPUfreq looks a bit more plausible.
http://home.arcor.de/david_ffm/oc_res3.jpg
David
Hi Group,
Are there any "officially" approved measurement points
where I could measure Vcore directly on this board?
I've noticed that the utilities to measure Vcore always
return values in 16mV steps, so that I can only ever
read the following values:
1,280 1,296 1,312 1,328 1,344 1,360 1,376 1,392 1,408
1,424 1,440 1,456 1,472 1,488 1,504 1,520 1,536 1,552
1,568 1,584 1,600 1,616 etc.
Most programs also only show 2 decimals, so I didn't realise
this at first. It seems that is the resolution of the A/D converter
(or D/A converter?) or something, so I would like to connect
a high quality digital multimeter to the board directly to see
how stable my Vcore is.
I've noticed with my board that I always get a lower voltage
than the nominal value I set. The board has two settings:-
Vid (0,85-1,55V in steps of 0,025V) and Vid Adjust
(0%, 3,3%, 5%, 8,3% or 10%). The CPU voltage is
Vcore = Vid + Vid Adjust. I am not sure if Vid supplies
other components too, so that when I increase it I am
overvolting something else. Does anyone know?
Anyway, I've played a bit with the settings and plotted the
deviation from nominal. It seems the Vid adjust setting is more
accurate, perhaps the results are of interest, though of course
it could be inadequacies in my PSU (Seasonic 350W).
http://home.arcor.de/david_ffm/oc_psu1.jpg
I have also replotted my overclocking results with the voltages
modified to reflect the measured values rather than the nominal
values, and added a few data points. The curve of stability in
the space of Vcore against CPUfreq looks a bit more plausible.
http://home.arcor.de/david_ffm/oc_res3.jpg
David
The most *accurate* method would 
Facebook
Twitter
Instagram