Pc Chips M848A Vcore Settings

I'm sure this question has been asked before.
I want to up my vcore setting, but I cant seem to find it
in the BIOS, can any1 help me?

Ty much


Fievel.

Fievel J Mousekewitz wrote:
I'm sure this question has been asked before.
I want to up my vcore setting, but I cant seem to find it
in the BIOS, can any1 help me?

Ty much


Fievel.


I have a manual here on disk, for the M848A. Apparently
there are different revisions of the board.

How the Vcore control works, is like this -


CPU ---- 5 VID ---------- 5 pins on
signals Vcore regulator

The VID value is controlled by bridges on the top of the
processor chip. So AMD encodes a value, a set of ones
and zeros, and that is what is setting the voltage.
The value is a five bit value, so there is a table of
32 possible settings. (One setting results in the
processor being unpowered, leaving 31 other choices.)

You can see some info here, in the L11 section, on
Vcore value versus five bit code value.

http://fab51.com/cpu/barton/athlon-e23.html

Some motherboards have an "overclock controller" chip.
That chip provides a storage device for FID and VID bit
values. When you set a value in the BIOS, it can be
stored in the overclock controller, and survive a
hardware reset. This chip is an example of such a device,
and on motherboards that support multipler settings or
Vcore settings, chances are some chip like this is used.
Sometimes, the multiplier portion of such a chip, only
controls four signals, when there are five on the processor,
and that is why some multiplier values may be missing in
the BIOS screen choices.

http://web.archive.org/web/200701211.../ATXP1_1.1.pdf

The path on a board with a chip like that, looks like this.
When unprogrammed, the overclock controller can pass the
5 bit value untouched. When the Overclock Controller is
programmed, then it asserts the new value required.

CPU ---- 5 VID ----- Overclock Controller ----- 5 pins on
signals Stores New VID value Vcore regulator

If your motherboard really doesn't have a VID (Vcore)
adjustment, then the other options are "wire mods".
This can be used to boost the voltage, but you need a
table of values to do it intelligently. (You don't want
to deliver too much voltage.) This site provides a diagram,
showing either a top view of the CPU socket, or a bottom
view, where the pins are soldered to the motherboard. By
making changes at the socket level, it is possible to
override the values on the VID signals. That looks like
this. To install tiny jumper wires in the socket, you
have to remove the processor to do it.

socket wiring mod interferes here
|
|
v
CPU ---- 5 VID ---------- 5 pins on
signals Vcore regulator

http://www.ocinside.de/go_e.html?/ht...md_pinmod.html

To understand what a socket mod is doing, you can look at the
pinout in an AMD datasheet, as a cross check. (PDF page 66.)

http://www.amd.com/us-en/assets/cont...docs/26237.PDF

NC NC VCC In the OCinside diagram, the
wire mods involve VSS holes,
VID[4] VID[3] and connecting select VID bits
to VSS, to put a logic 0 on them.
NC VSS VCC

NC VID[2]

VCC VSS VCC

SAO#1 VID[1]

VCC VSS VCC

SAO#0 VID[0]

To work an example, say the existing Vcore is 1.5V as sent from
the factory. If you use the hardware monitor, and the processor
is relatively idle, the voltage will jump up a little bit, above
that value. So the measured value might be 1.56V. It is hard to be
certain, using just the hardware monitor, as to what voltage the
chip normally uses. But if you have the OPN number printed on the
top of the processor, you can look it up as a double check. The
fab51.com page shows a letter, and a particular letter in the OPN,
says what voltage the processor is using.

So, let's say your processor is 1.500V. In the fab51.com table,
that corresponds to "C : : : C", or closed/open/open/open/closed.
The two closed bits, means those bits are already grounded to VSS.
I can use any VID value in the fab51.com table, where the VID4
and VID0 are shown as closed (as I cannot easily change a closed
one). I use 1.65V on my own processor, as an example of a typical
choice. 1.65V might be used on a 3200+.

CCCCC 1.85V need three jumper wires
CCC:C 1.80V need two jumper wires
CC:CC 1.750 need two jumper wires
CC::C 1.700 need one jumper wire
C:CCC 1.650 need two jumper wires
C:C:C 1.600 need one jumper wire
C::CC 1.550 need one jumper wire

If you wanted to change a closed to an open, that would require
mods to L11 bridges.

The two red marks on this image, might be the result of someone
trying to get 1.675V. But the number of possible voltage values,
is controlled by the number of VID bits which are open ":". When
I selected a 1.500V processor as an example, with three open VID
bits, that leaves 8 possible voltage values, so there are seven
new options for setting the voltage. But if only two bits were
open ":", then the table above would only have three new voltage
values possible.

(This figure is a 180 degree rotation of PDF page 66 in the AMD document.)
http://www.ocworkbench.com/2004/arti...Socket_462.jpg

The simple answer might be, use another motherboard, depending
on how squeamish you are. My A7N8X-E Deluxe, for example, allows
me to change the voltage and multiplier, and do a bit of
overclocking. You need to do a fair amount of research, to get
conclusive evidence of how much voltage or frequency change is
possible. (For example, some motherboards only control four of
five multiplier bits, so a wire mod might still be needed on
those boards, to access all possible multiplier values.)

Paul