Crash course with vCore offset

Soldato
Joined
12 Jan 2005
Posts
3,357
Location
Devon
Going from scratching my head this morning, to ripping my hair out this afternoon and then sighing relief 20 minutes ago.

Yesterday I had a 2500k running happily at 4.4Ghz IBT stable at 1.24v constant. After that I decided I wanted to have some sort of power saving as I don't always need that sort of clock speed.

As an 'old skool' clocker, I do find this sort of thing refreshing. Being able to tweak your clock for power when you need it, and power-sipping efficiency when you don't. Also it seems that the chip needs less volts when using this method. Maybe it's using the VRM's more effectively?

LLC set to 0%, -.130 offset on a Maximus IV Extreme.

2500kOffsetIdle.jpg


2500kOffsetLoad.jpg
 
So you tried getting to 4.4ghz @ 1.224v fixed, and with 1.224v with off set, and only the offset was stable?

Saying you need less volts with offset would be very strange, so I would like to see some data on that.

Personally I've never found the need for offset, as you can just enable speedstep and c-states and the voltage drops to 0.132 (a bit higher for sandy) anyway.
 
Seems that way, any less than 1.24v fixed and it'd crash.

I found not using offsets made the voltage too high in general. 1.36v or so at 4.4GHz. So I was getting close to 80 degrees at load.
 
Back
Top Bottom