Frank Bergemann wrote:
> Hi,
>
> i am confused about MSI's pages saying
> (http://global.msi.com.tw/index.php?
> func=proddesc&prod_no=664&maincat_no=133)
>
> CPU
> • Supports dual Socket A (Socket-462) for AMD® Athlon™ MP processors
> • Supports up to Athlon MP 2800+ or higher
>
> The board supports 133MHz - with overclocking up to 150 MHz.
> In the BIOS the CPU multiplier is at maximum 12.5
>
> So it is maximum 150 MHz * 12.5 = 1875 MHz CPU clock.
> This sounds as if MP 2200+ (clocked 1800Mhz) is the maximum.
>
> So how comes MSI states K7D Master-L supports "MP 2800+ or higher"???
>
> - thanks for your help
>
> rgds
> Frank
http://en.wikipedia.org/wiki/List_of...icroprocessors
Athlon MP "Barton" (Model 10, 130 nm)
Athlon MP 2800+ 2133 MHz 512 KiB 266 MT/s 16x 1.60V 60.0W May 6, 2003 AMSN2800DUT4C
That processor has a multiplier of 16x, so 133 * 16 = 2128MHz. (Equals 2133MHz roughly)
FSB is FSB266, as the FSB is double data rate, and twice the input clock of 133MHz.
Note that some motherboards, only have multiplier control for
four of five FID bits. The MSB of the FID is not controllable.
The board may use an "overclock controller chip", having only
four FID bits for setting the multiplier. An overclock
controller chip, stores the FID between board resets, so the
board will remember the multiplier to use, when the user
sets a custom value.
But the actual processor itself, does have five FID bits to drive
to the socket. So the processor can set 16x if it wants. It is the
motherboard controls, that may not be able to set a "custom" value
of 16x.
A bridge mod or a socket mod, can be used to program the MSB of the
FID. Which would be another way to modify it. You would flip the
upper FID bit, and then you can rely on the BIOS setting to control
the lower four FID bits. Then, you need a table that maps FID
setting, to actual multiplier. And while testing, you run at
clock=100MHz, to reduce the impact of using too high a multiplier
value by accident. Once the multiplier is verified as being correct,
the clock can be bumped up to the value you want.
http://www.ocinside.de/go_e.html?/ht...md_pinmod.html
I think my motherboard only has a four bit FID control, and I never
go over 12.5x with it. (I have to use a custom setting, because I have
a mobile processor, and the default multiplier is too low.) Fortunately,
my FSB goes to FSB400, so 200 * 12.5 = 2500MHz gives me enough room for
overclocking experiments, without needing a high multiplier.
Also, on dual socket S462 boards, you should be a bit careful about
how the power is provided to the Vcore regulators. If you had two
60W processors, and the Vcore regulators were 90% efficient, then
the input power required is 133.3W . If that came from the 5V rail,
that would be 26.7 amps. The main ATX power connector has four
wires to carry +5V, and each is rated for 6 amps. For a total of
24 amps under best conditions. There have been some motherboards in
the past, that had problems with the power connector, as a result
of the loading coming from the processors. (And the reason they had
a problem, is the actual loading on the wires was not even. On
my own S462 motherboard, there is a 1 amp difference in current
draw, between the highest current flow and the lowest current
flow, of the four 5V wires. On badly designed motherboards, if
that imbalance is large enough, the connector pins burn.)
What the previous paragraph means in practice, is if you've been
using two high end processors in your motherboard, then pull and
check the condition of the ATX power connector pins. All the pins
should be shiny.
Paul
Similar Threads
Linear Mode
