paulj said:
> > In article
<
[email protected]>, "JMac"
> > <
[email protected]> wrote:
> >
> > > OK, bear with me, I only do this kind of
thing on rare
> > occasions.
> > >
> > > I'm thinking of upgrading my A&V266-e with
a new CPU. If I
> > got it figured
> > > out I can put an Athlon XP 2600+ fsb 266,
in this board.
> > I'd need to change
> > > the Duron/Palomino jumper and upgrade BIOS
to 1011.
> > >
> > > When looking for a CPU to purchase I see 2
types a mobile
> > version, Athlon
> > > XP-M 2600+ 2.0GHz Barton and Athlon XP
2600+ 2.133GHz
> > 266FSB.
> > >
> > > I'm thinking I don't want the mobile CPU?
> > >
> > > Any thoughts the group may have I'd love to
hear. Anything
> > else I need to be
> > > aware of before doing this?
> > >
> > > Thanks!
> >
> > It was pretty hard to find details for this one.
Fortunately
> > A7Vtroubleshooting had an answer.
> >
> > I thought the 2600+ XP FSB266 was relatively hard
to find. So,
> > your first problem for that one, could be locating
a source
> > for
> > one.
> >
> > In terms of general performance characteristics
(and this is
> > true
> > of any processor), processors are starved for
memory
> > bandwidth.
> > This gets worse, the higher the multiplier that is
used.
> >
> > The A7V266-E is a FSB266 board, so the ratio
between the core
> > and the FSB is worse than it would be with a board
that
> > supports FSB400. What this means is, if you upgrade
to the
> > fastest officially supported configuration, much of
the
> > core speed of 2GHz+ is wasted, as the processor is
waiting
> > for memory a lot of the time. While computations
which are CPU
> > bound still benefit, as soon as there is even a
tiny
> > dependency
> > on external memory, performance drops.
> >
> > The upgrade path you take, really depends on how
much money
> > you
> > have to spend, and the practical matter of how much
life there
> > is left in the AthlonXP family versus Athlon64.
Certainly, if
> > you stick with your current motherboard, you'll get
a little
> > boost from a processor upgrade. If you go for a
motherboard
> > that
> > support FSB400/DDR400 plus a new processor, that
will be
> > better
> > than using the A7V266-E. A step above that, is to
go with
> > Athlon64 and the possibility of the low latency
processor
> > memory
> > controller (and dual channel, if using S939). But,
the upgrade
> > price of Athlon64 is a stretch past the other two
options.
> >
> > Now, in terms of the processor upgrade, I much
prefer the
> > flexibility that comes with the mobile 2600 XP-M.
Since
> > it is unlocked, you can dial up settings suitable
for a
> > clock of 133MHz or a clock of 200MHz. That means
the processor
> > could be happy in the A7V266-E now, and could be
put in a
> > FSB400 motherboard later.
> >
> > This posting has some info on what works or doesn't
work on
> > the
> > A7V266-E with respect to a Mobile processor.
> >
> >
http://www.a7vtroubleshooting.com/f...oard=a7v;action=display;num=1099432970;start=
> >
> > So, the five position multiplier DSW, doesn't drive
the most
> > significant bit of the multiplier setting. And,
that is why
> > you may find that only the "low" multipliers are
available.
> >
> > There is a "wire trick", to select the "high"
multipliers. The
> > high multipliers are needed on a FSB266 board, as
to get 2GHz
> > rates, you need something like 133x15. Since 15 is
in the high
> >
> > multiplier set, you might need the wire trick to
enable the
> > multipliers.
> >
> >
http://www.ocforums.com/showthread.php?threadid=279392
> > (I like this picture as it shows the real mccoy.)
> >
http://img49.exs.cx/img49/8428/13multi2.jpg (wire
trick 13x+
> > multis)
> >
> > I believe this adapter does much the same thing,
only it
> > has switches on it.
> >
> >
http://www.upgradeware.com/english/product/xptmc/xptmc.htm
> >
> > (Multiplier table)
> >
http://www.upgradeware.com/english/product/xptmc/setup.htm
> > (Compatible motherboards)
> >
http://www.upgradeware.com/english/product/xptmc/compatibility.htm
> >
> > This site shows the various socket mods.
> >
http://www.ocinside.de/go_e.html?/html/workshop/pinmod/amd_pinmod.html
> >
> > L3 Sysclk FID4 is mentioned here. The pin used for
the wire
> > trick.
> >
http://fab51.com/cpu/barton/athlon-e23.html
> >
> > The multiplier pins are actually marked as NC for
"No
> > Connection"
> > in the AMD datasheet for a Barton. If you use the
ocinside web
> > site,
> > and select various voltages, you can see pins J7,
L7, L5, L3,
> > L1
> > (which are VID[4] through VID[0]), change strapping
values.
> > You can
> > align that info to the pinout on PDF page 66. Page
66 is a top
> > view, which is the same as looking into the socket.
> >
> >
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26237.PDF
> >
> > In that datasheet, the multiplier pins are marked
"NC" (no
> > connect)
> > in the top view. Socket hole AJ27 is the FID4
multiplier pin.
> > AH28 = VSS = logic 0 pin. AH26 = VCC = logic 1 pin.
You can
> > see
> > the pivot from one position to another on the
ocinside socket
> > view, using multiplier 12.5 and 13. For multipliers
>=13,
> > connect AJ27 to AH26 (logic 1).
> >
> > Software tools like CPUMSR can also be used to
change the
> > multiplier
> > when booted into Windows. Presumably, what would
happen, is
> > the
> > computer would boot at 600 or 800MHz or so, then
you would
> > use this program to set the multiplier register
inside the
> > Mobile processor.
> >
> >
http://www.cpuheat.wz.cz/html/AXP_multiplier_FAQ/AXP_Multiplier_FAQ.htm
> >
http://www.cpuheat.wz.cz/html/main.htm (CPUMSR
download)
> >
> > There are a number of boards that support socket
462 and
> > FSB400/DDR400. With those, you could do 200x12.5 =
2500MHz,
> > so low multipliers would be all you would need. You
might
> > need a RAM upgrade, if the board requires DDR333 or
DDR400
> > memory when running at FSB400. Some motherboard
manuals
> > have a table of what combinations of RAM speed and
FSB
> > are possible.
> >
> > HTH,
> > Paul
Hi
I took Paul4’s advise last night and did the ’wire trick’ on my
A7V266-E board. I am also now runing a mobile barton core 2600+ chip
(266FSB version.)
I have followed the instructions and my machine is powering up, but
the main board only picks the cpu up at 1866 in the bois startup
screen. If I change the multipler it doesnt make any differece (in
Windows using WCPUID) it still runs at 14 x 133 eg 1833mhz (1.57
volts). It wont run any faster?
The mother board is running in Jumper free mode and i dont know why
this is hapening, can anyone help?
Thanks in advance
Paulj
If Jumperfree mode is refusing to accept multiplier settings,
go to Jumper mode. Set JEN to 1-2 position. Set Sysclk to
133MHz (all switches OFF). Then, use CPU_ratio DSW near
the top edge of the board.
Now, compare the CPU_Ratio info in the manual, to the table
from fab51. Notice that the bit positions in the table are
reversed. So, you can use the fab51 info, to set your
switches,
as long as you remember to reverse the order. I've included
these table entries, to show how fab51 compares to your
manual.
CPU_ratio Switch position 1 2 3 4 5 fab51 barton
FID4,3,2,1,0
8X ON -OFF-ON -OFF-ON C:C:C
8.5X OFF-OFF-ON -OFF-ON C:C::
9X ON -ON -OFF-OFF-ON C:C::
9.5X OFF-ON -OFF-OFF-ON C::C:
10X ON -OFF-OFF-OFF-ON C:::C
10.5X OFF-OFF-OFF-OFF-ON C::::
Here is the full table, quoted from fab51:
The Barton table is the same as the L6 mobile table.
http://fab51.com/cpu/barton/athlon-e23.html
Barton FSB:166
Mult Core L3-FID[4:0]
5.0x 833 CC:CC -
5.5x 917 CC:C: -
6.0x 1000 CC::C -
6.5x 1083 CC::: -
7.0x 1167 C:CCC -
7.5x 1250 C:CC: -
8.0x 1333 C:C:C -
8.5x 1417 C:C:: -
9.0x 1500 C::CC -
9.5x 1583 C::C: -
10.0x 1667 C:::C -
10.5x 1750 C:::: -
11.0x 1833 CCCCC -
11.5x 1917 CCCC: -
12.0x 2000 CCC:C -
12.5x 2083 CCC:: -
13.0x 2167 :C:CC -
13.5x 2250 :C:C: -
14.0x 2333 :C::C -
21.0x - :C::: -
15.0x 2500 ::CCC -
22.0x - ::CC: -
16.0x 2667 ::C:C -
16.5x 2750 ::C:: -
17.0x 2833 :::CC -
18.0x 3000 :::C: -
23.0x - ::::C -
24.0x - ::::: -
3.0x Invalid :CCCC -
19.0x - :CCC: -
4.0x Invalid :CC:C -
20.0x - :CC:: -
When fooling with the multiplier, the best approach is to set
the system clock to 100MHz, until you are sure you have the
right
multiplier set. Say, for example, you set 14x. With the clock
at 100MHz, the BIOS will show 1400MHz. Once you are sure, then
change the clock to 133MHz and start the computer again.
Now, concerning the wire trick. With the wire in place, I
would
only try switch settings in the lower half of the table,
starting
with 13.0x. The reason for this, is I don't know whether the
switch position 5 on DSW is wired to the socket or not. I am
worried about the DSW fighting with the wire in the socket.
To eliminate that as a possibility, stick to multiplier codes
that have :xxxx as the format (from 13.0x to 20.0x in the
table
above), as long as the wire is in the socket.
Tell us how it works out.
Also, I don't know why the BIOS was not able to set the
multiplier.
Unless it was a recognition problem or something. Maybe the
BIOS
won't set the multiplier with the wire in place ? Maybe a
later
version of BIOS would work better ? That is the nature of
working
with Mobiles - everything is an experiment
Paul
