Terry Dwyer wrote:
> The board seems to have been intended for an AMDXP1800+ since this
> is the default BIOS setting; it offers various fsb speeds of 100/33 to
> 227/45
> & I'm using it at 11.5x167/33 for 1921.09 MHZ with an XP2600+.
> Reviews on the net indicated the MB is not really suitable for this CPU
> due to cooling problems which seems to be true since I've seen BIOS
> temperatures
> up to 78 C. It has overheating protection according to the manual but
> there is
> no BIOS setting for the trip point which has activated during heavy
> floating
> operations. This was cured by using a PSU with a side cooling fan.
>
> Linux dmidecode shows:
> BIOS version= ASUS A7V333 ACPI Bios revision 1006
> BIOS date= 04/17/2002 .
>
> The BIOS offers CPU multipliers of 9.5 to 14.0x; I thought the CPU would
> have
> been locked but it runs ok at 12.5x167= 2087.5 which corresponds to an
> XP2800+;
> as experiment the 13.0-14.0 multipliers were tried but these gave lower
> speeds:
> 13.0x167 -> 835.26
> 13.5x167 -> 918.76
> 14.0x167 -> 1002.30 MHZ
> Is this a BIOS problem given its old date or a CPU quirk?
>
The problem is, many motherboards (including my S462 board) only have
logic drive for four of the five FID bits. The most significant FID
bit is not driven by the overclock controller, which means the user
is stuck only being able to access half of the table of multiplier
settings.
First, consult "Barton FSB:167" table on this page. That will give you
a basic table of multiplier values versus five bit FID.
http://fab51.com/cpu/barton/athlon-e23.html
Next, go here.
http://www.ocinside.de/go_e.html?/html/workshop/pinmod/amd_pinmod.html
Select "Socket View", "AMD Barton", "Vcore Selection" (i.e. don't show
the Vcore wire mod), then "...12.5xFSB" for the rightmost one. What you'll
see in your browser, is a picture of a S462 socket, with five wires shoved
into the holes in the socket, after which the processor is mounted into
the socket. This is a "wire mod", a way of forcing the FID.
Now, set the right most menu item to 20xFSB, and you'll notice only
the left most wire mod changes. That wire is pivoting between logic 0 and
1, for the MS FID.
How you'd use that, is if you wanted 13.5, you'd select 13.5x on the
ocinside.de site, then push a single strand from a piece of stranded
wire, where the left most wire is shown in the socket picture. Then,
you'd go into your BIOS, and select 13.5x. The motherboard would drive
FID0,1,2,3 to the bit pattern necessary for the 13.5x setting, and the
"wire mod" done to the FID4 bit, would complete the job.
If you're handy with a soldering iron, there are people who have
soldered switches to all the changeable stuff, for convenient control.
The "wire mod" method is not very convenient, which is why I was never
tempted to try it.
I don't know how overheat is detected on that board. Some boards have
used the CPU thermal diode, and monitored that with a Winbond 8 pin chip.
The board could also use a thermistor, and a software based method of
shutting down the machine. That method is slower than the Winbond chip.
The Winbond chip would be wired into the logic tree for PS_ON#.
One older thread mentioned the board using too much Vcore for the
processor, and the table here shows some of the possible chips.
There are three 1800+ chips in the table, and the earliest of them
uses 1.75V.
http://web.archive.org/web/20031018050306/http://www.qdi.nl/support/CP...ISocket
You could try
http://www.cpuid.com/cpuz.php and get a copy of CPUZ and
check the processor version. Then see if the motherboard is using
the wrong voltage for it. Reducing the voltage might help a bit
with the high temperatures. You could also change coolers, as a
solution.
Paul
>> Stay informed about: ASUS A7V333 questions 
FAQ
Profile
Private Messages
Log in
