a8n-sli premium lowest cpu vid

hi

trying out crystalcpuid (288) and i am trying to lower my vid but i cant go under 1.100v???
bios show's lower vid is possibly, 0.800v.

anyony know why?

bios 1009 and i have tested both auto and manual bios vid settings.

tia coco

In article ,
wrote:

hi

trying out crystalcpuid (288) and i am trying to lower my vid
but i cant go under 1.100v??? bios show's lower vid is
possibly, 0.800v.

anyony know why?

bios 1009 and i have tested both auto and manual bios vid settings.

tia coco

Asus traditionally does not allow undervolting. The BIOS
will prevent you from reaching the minimum VID value for
the processor.

Paul

Paul skrev:
In article ,
wrote:

hi

trying out crystalcpuid (288) and i am trying to lower my vid
but i cant go under 1.100v??? bios show's lower vid is
possibly, 0.800v.

anyony know why?

bios 1009 and i have tested both auto and manual bios vid settings.

tia coco

Asus traditionally does not allow undervolting. The BIOS
will prevent you from reaching the minimum VID value for
the processor.

Paul

any way of going around this?

tia coco

In article ,
wrote:

Paul skrev:
In article ,
wrote:

hi

trying out crystalcpuid (288) and i am trying to lower my vid
but i cant go under 1.100v??? bios show's lower vid is
possibly, 0.800v.

anyony know why?

bios 1009 and i have tested both auto and manual bios vid settings.

tia coco

Asus traditionally does not allow undervolting. The BIOS
will prevent you from reaching the minimum VID value for
the processor.

Paul

any way of going around this?

tia coco

I think I should clarify that statement a bit.

In previous generations of motherboards, the only way to set Vcore,
was via the BIOS. The VID pins on the bottom of the processor would
define a default Vcore voltage, and the Asus motherboard would have
some logic signal drivers that could override the VID pin value. That
is how the BIOS could set the voltage. In that case, Asus usually
followed the CPU manufacturer's advice for allowed voltage values,
and that is why the VID could not be set in the BIOS to the lowest
possible value.

For those old motherboards, you could modify the hardware signals
controlling VID. This would void the warranty, and at least one
poster here succeeded in doing such a modification.

The newer processors are different, and desktop processors are
gaining more power management features than they used to have in
the past. In the case of your processor, there is the Cool and
Quiet feature. Such a feature could be called a "soft VID", because
the VID value can be controlled by writing a register inside the
processor.

Before we go any further, get a copy of this document.

http://www.amd.com/us-en/assets/cont...docs/30430.pdf

For each processor, there are defined "P-states". AFAIK, a P-state
is part of ACPI, and is a power state for hardware. Let's take
an example from that AMD document, on PDF page 8

ADA3000AEP4AP

Max P-State 2000MHz (10x multiplier, if clock=200), VID_VDD = 1.5V
Intermediate P-State #1 1800MHz (9x multiplier, if clock=200), VID_VDD = 1.4V
Min P-State 800MHz (4x multiplier, if clock=200), VID_VDD = 1.3V

Now, _something_ in your computer knows those values. It could be that
the BIOS passes ACPI objects to the OS, via an ACPI table during POST.
Or the AMD CPU driver software could have these values recorded. I
don't really know the mechanics of this process well enough, to tell
you how those FID/VID pairs are controlled. In any case, a P-state
table for each processor type, must be stored on the computer, in order
that each processor have a unique table of values. Obviously newer
processors have a lower Min P-state VID value than the older processors.

What I can suggest, is crystalcpuid will not be able to set the voltage
on the sample processor above, to less than 1.3V. Because if
crystalcpuid honors the ACPI p-state values, it won't try a voltage
below 1.3V.

The above table of values represent a characterization from AMD, that is
guaranteed to work for their processor. The characterization is
conservative, and as you suspect, the processor might continue to
run if the voltage is dropped further. Of course, your CPU frequency
should be dropped in proportion to the voltage - you cannot expect to
run a 10X multiplier, and a 0.8V Vcore value. The above table shows you
the general operating trend.

The fact that the following program also honors the "P-state transitions",
suggests that at the OS level, only ACPI objects are available to make
changes to hardware settings. It could be that there is no driver or API
available to make arbitrary changes to the soft-VID directly. You can
see there is a P-state table in this tool, but again, I don't know
where the characterization data in the table is coming from.

http://cpu.rightmark.org/products/rmclock.shtml

If you experiment with that utility, I know the poster "Figo" is
curious about it too :-) So post back with your test results.

Paul

Paul skrev:
In article ,
wrote:

Paul skrev:
In article ,
wrote:

hi

trying out crystalcpuid (288) and i am trying to lower my vid
but i cant go under 1.100v??? bios show's lower vid is
possibly, 0.800v.

anyony know why?

bios 1009 and i have tested both auto and manual bios vid settings.

tia coco
Asus traditionally does not allow undervolting. The BIOS
will prevent you from reaching the minimum VID value for
the processor.

Paul
any way of going around this?

tia coco

I think I should clarify that statement a bit.

In previous generations of motherboards, the only way to set Vcore,
was via the BIOS. The VID pins on the bottom of the processor would
define a default Vcore voltage, and the Asus motherboard would have
some logic signal drivers that could override the VID pin value. That
is how the BIOS could set the voltage. In that case, Asus usually
followed the CPU manufacturer's advice for allowed voltage values,
and that is why the VID could not be set in the BIOS to the lowest
possible value.

For those old motherboards, you could modify the hardware signals
controlling VID. This would void the warranty, and at least one
poster here succeeded in doing such a modification.

The newer processors are different, and desktop processors are
gaining more power management features than they used to have in
the past. In the case of your processor, there is the Cool and
Quiet feature. Such a feature could be called a "soft VID", because
the VID value can be controlled by writing a register inside the
processor.

Before we go any further, get a copy of this document.

http://www.amd.com/us-en/assets/cont...docs/30430.pdf

For each processor, there are defined "P-states". AFAIK, a P-state
is part of ACPI, and is a power state for hardware. Let's take
an example from that AMD document, on PDF page 8

ADA3000AEP4AP

Max P-State 2000MHz (10x multiplier, if clock=200), VID_VDD = 1.5V
Intermediate P-State #1 1800MHz (9x multiplier, if clock=200), VID_VDD = 1.4V
Min P-State 800MHz (4x multiplier, if clock=200), VID_VDD = 1.3V

Now, _something_ in your computer knows those values. It could be that
the BIOS passes ACPI objects to the OS, via an ACPI table during POST.
Or the AMD CPU driver software could have these values recorded. I
don't really know the mechanics of this process well enough, to tell
you how those FID/VID pairs are controlled. In any case, a P-state
table for each processor type, must be stored on the computer, in order
that each processor have a unique table of values. Obviously newer
processors have a lower Min P-state VID value than the older processors.

What I can suggest, is crystalcpuid will not be able to set the voltage
on the sample processor above, to less than 1.3V. Because if
crystalcpuid honors the ACPI p-state values, it won't try a voltage
below 1.3V.

The above table of values represent a characterization from AMD, that is
guaranteed to work for their processor. The characterization is
conservative, and as you suspect, the processor might continue to
run if the voltage is dropped further. Of course, your CPU frequency
should be dropped in proportion to the voltage - you cannot expect to
run a 10X multiplier, and a 0.8V Vcore value. The above table shows you
the general operating trend.

The fact that the following program also honors the "P-state transitions",
suggests that at the OS level, only ACPI objects are available to make
changes to hardware settings. It could be that there is no driver or API
available to make arbitrary changes to the soft-VID directly. You can
see there is a P-state table in this tool, but again, I don't know
where the characterization data in the table is coming from.

http://cpu.rightmark.org/products/rmclock.shtml

If you experiment with that utility, I know the poster "Figo" is
curious about it too :-) So post back with your test results.

Paul

hi

have now tested rightmark cpu clock, same result! lowest possibly vid is 1.100v, obvious
taken from a p-state table in the cpu!

what about disabling cool-n-quiet in bios? lower vid? (going to test later)

(i have already uninstalled amd x2 cpu driver, not needed with crystalcpuid or rightmark
cpuclock)

coco

CoCo skrev:
Paul skrev:
In article ,
wrote:

Paul skrev:
In article ,
wrote:

hi

trying out crystalcpuid (288) and i am trying to lower my vid but i
cant go under 1.100v??? bios show's lower vid is
possibly, 0.800v.

anyony know why?

bios 1009 and i have tested both auto and manual bios vid settings.

tia coco
Asus traditionally does not allow undervolting. The BIOS
will prevent you from reaching the minimum VID value for
the processor.

Paul
any way of going around this?

tia coco

I think I should clarify that statement a bit.

In previous generations of motherboards, the only way to set Vcore,
was via the BIOS. The VID pins on the bottom of the processor would
define a default Vcore voltage, and the Asus motherboard would have
some logic signal drivers that could override the VID pin value. That
is how the BIOS could set the voltage. In that case, Asus usually
followed the CPU manufacturer's advice for allowed voltage values,
and that is why the VID could not be set in the BIOS to the lowest
possible value.

For those old motherboards, you could modify the hardware signals
controlling VID. This would void the warranty, and at least one
poster here succeeded in doing such a modification.

The newer processors are different, and desktop processors are
gaining more power management features than they used to have in
the past. In the case of your processor, there is the Cool and
Quiet feature. Such a feature could be called a "soft VID", because
the VID value can be controlled by writing a register inside the
processor.

Before we go any further, get a copy of this document.

http://www.amd.com/us-en/assets/cont...docs/30430.pdf


For each processor, there are defined "P-states". AFAIK, a P-state
is part of ACPI, and is a power state for hardware. Let's take
an example from that AMD document, on PDF page 8

ADA3000AEP4AP

Max P-State 2000MHz (10x multiplier, if clock=200),
VID_VDD = 1.5V
Intermediate P-State #1 1800MHz (9x multiplier, if clock=200),
VID_VDD = 1.4V
Min P-State 800MHz (4x multiplier, if clock=200),
VID_VDD = 1.3V

Now, _something_ in your computer knows those values. It could be that
the BIOS passes ACPI objects to the OS, via an ACPI table during POST.
Or the AMD CPU driver software could have these values recorded. I
don't really know the mechanics of this process well enough, to tell
you how those FID/VID pairs are controlled. In any case, a P-state
table for each processor type, must be stored on the computer, in order
that each processor have a unique table of values. Obviously newer
processors have a lower Min P-state VID value than the older processors.

What I can suggest, is crystalcpuid will not be able to set the voltage
on the sample processor above, to less than 1.3V. Because if
crystalcpuid honors the ACPI p-state values, it won't try a voltage
below 1.3V.
The above table of values represent a characterization from AMD, that is
guaranteed to work for their processor. The characterization is
conservative, and as you suspect, the processor might continue to
run if the voltage is dropped further. Of course, your CPU frequency
should be dropped in proportion to the voltage - you cannot expect to
run a 10X multiplier, and a 0.8V Vcore value. The above table shows you
the general operating trend.

The fact that the following program also honors the "P-state
transitions",
suggests that at the OS level, only ACPI objects are available to make
changes to hardware settings. It could be that there is no driver or API
available to make arbitrary changes to the soft-VID directly. You can
see there is a P-state table in this tool, but again, I don't know
where the characterization data in the table is coming from.

http://cpu.rightmark.org/products/rmclock.shtml

If you experiment with that utility, I know the poster "Figo" is
curious about it too :-) So post back with your test results.

Paul

hi

have now tested rightmark cpu clock, same result! lowest possibly vid is
1.100v, obvious taken from a p-state table in the cpu!

what about disabling cool-n-quiet in bios? lower vid? (going to test later)

(i have already uninstalled amd x2 cpu driver, not needed with
crystalcpuid or rightmark cpuclock)

coco


grrr, not even without bios cool'n'quiet support does the vid get any lower than 1.100v!!
crystalcpuid works perfekt without cool'n'quiet enabled....stated otherwise in the faq!

setting the cpu to 4x and 1.000v vid in bios cpu options is fully ignored....

coco