Testing / comparing : Intel D975XBX2 / Asus P5B DX ***56K WARNING***

[aicjofs]

what you're actually doing is telling the VRM "When the processor asks for it's default VID, (1.3250v in my case) really give it 1.4000v". You'd think that if the processor asks for a lower voltage that the VRM would scale accordingly but it seems that if you set a vcore other than the default, the VRM always delivers that voltage even if the processor asks for a lower one.

This has been my only issue with this board. If you overvolt to say 1.5V I can't seem to find a way to get it to change withiin windows.

I run pretty high overclocks and run my computer 24/7 but during that time the CPU is idle much of the time. In the past I am able to use a CrystalCPUID or RMClock as SpeedStep/CoolnQuiet and you have controller over multi/voltage changes.

We all know the multi is being stepped down during halts but voltage doesn't which I personally like and never had a problem with while doing heavy overclocking. If I send a requested new lower VID(in for example Crystal) the voltage stays the same and doesn't lower in Speed Fan, IDCC, etc etc, but CPU-Z picks up the requested change lowering (from 1.22 to say 1.16) so the request is making it but the VRM output is constant which I would like changed.

This BIOS implementation isn't "poor"(just not the best word), but it's just an non flexible and non optimal setup if the VRM's output a constant if you are overvolting. Hopefully someone can tell me something I have overlooked on how to get this VRM to deliever 1.5V and stepdown voltage when idle in a manner controlled by the user.

A better way I think to do it is 2 BIOS options. One for VID and another for % OverVID. In this case if you set 1.35VID and 10% Overvid the VRM outputs 1.485V and the system will always take the requested VID from BIOS, OS or where ever and apply 10%. This way you can use a Speedstep type functionality(except RMClock, Crystal) and yet maintain the multipliers and voltage you want for your particular overclock.

EDIT: and oh yeah on the CPU week stepping/rev thing. Either Intel or AMD, I think it does matter.... however week to week variance I think is a lot smaller then people make it out. a week 45 or 46 I think is small, a week 20 to a week 50, I think "could" be significant, and revision/stepping can definetly bring change, not always but it definetly can.