Difference between moible and standard AMD Athlon processor?

[David Farber]

I bought an AMD AXMA3000FKT4C processor for my ASRock K7VT4A+ motherboard. I
didn't notice that it was an "M" processor until I was ready to install it.
Is this processor compatible with a regular desktop motherboard? If so, is
the performance the same?

Thanks for your reply.

 

[Conor]

I bought an AMD AXMA3000FKT4C processor for my ASRock K7VT4A+ motherboard. I
didn't notice that it was an "M" processor until I was ready to install it.
Is this processor compatible with a regular desktop motherboard? If so, is
the performance the same?

Thanks for your reply.

Some motherboards need a BIOS upgrade. They use less power and run
cooler but have the same performance. Ideal for overclocking.

 

[Paul]

I bought an AMD AXMA3000FKT4C processor for my ASRock K7VT4A+ motherboard. I
didn't notice that it was an "M" processor until I was ready to install it.
Is this processor compatible with a regular desktop motherboard? If so, is
the performance the same?

Thanks for your reply.

I bought a 2600+ AthlonXP-M, and this is my experience.

1) Multiplier is unlocked. It helps to have a motherboard where
you can set the multiplier in the BIOS.
2) The first time you start up the motherboard, the CPU may be running
at 600 to 800MHz (I don't remember the exact figure - the default
multiplier is a relatively low value).
3) The VID coding is slightly different than a desktop processor.
That means at startup, it won't request the exact same voltage
as a desktop processor would. Since your processor is a 3000+ though,
the default voltage it asks for, will be pretty high, and probably
not that much different than normal. Some BIOS use the VID they read
from the processor, to determine the voltage adjustment range, and
that can present problems for people who undervolt, for cool operation.

For VID codings, the table on fab51.com is good. See the table in the
section entitled "L11 : Code to CORE Voltage Definition". My processor
is a "1.450(Q)", meaning AMD wanted it to run at 1.45V when used as
a 2600+ mobile. But when plugged into my motherboard, the Vcore regulator
runs it at 1.575V, just like the fab51.com table shows. This is OK, as my
eventual voltage selection was 1.65V for best stability. If I go much
higher than that in voltage, in fact my motherboard won't POST.

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

There is a table of processors here.

http://web.archive.org/web/20031018050306/http://www.qdi.nl/support/CPUQDISocketA.htm

Notice that a 3200+ Barton, has an actual core clock rate of 2000MHz.
With my brand new mobile processor, I set the input clock to the CPU
to 200MHz. That makes the FSB run at FSB400. For a multiplier, I
set the multiplier to 10x, so that 200x10 = 2000MHz, so that my
processor would be overclocked to be a top of the line 3200+.
I set the Vcore voltage in the BIOS to 1.65V, and tested for stability
with Prime95 (Orthos is also good for this).

http://sp2004.fre3.com/beta/beta2.htm (Orthos tester)

The multiplier is also known as the FID (frequency ID). It is a five bit
value, allowing 32 possible codings. My motherboard has hardware control
signals for only the lower four bits of the FID. This means my motherboard
has limited control of the multiplier. I can select multipliers in the
"low" range. The low range goes up to a multiplier of 12.5. To select a
higher multiplier than that on my motherboard, I have to stick a wire in
the S462 socket, to short a couple holes together. That would expose multipliers
from 13x upwards.

So in my case, my choices are easy ones. I use FSB400, so that I won't need
a high multiplier value to get good use from the processor. If you used
a low FSB, like FSB266, then you'd need a higher value of multiplier, to
get the core cranking along.

The Asrock motherboard in question, supports FSB333. That gives 166MHz feeding
the processor. To run it like it was an AthlonXP 3200+, the multiplier would
need to be 12x. 166 * 12 = 2000MHz. Just a matter of seeing if the motherboard
can set the multiplier or not. OK, I checked the manual and it has jumpers for
FID Should be loads of fun... No need for wire tricks at all.

http://www.asrock.com/mb/overview.asp?Model=K7VT4A+&s=462

Note - if you don't trust the table in the manual, in terms of
defining the multiplier values correctly, try running the processor
with the CPU input clock running at 100MHz. Then, if you accidentally
select a much higher than normal multiplier value, then there is
less chance of harm to the processor. 100x24 is 2400MHz, and my processor
can run at that speed, and so should yours be able to run that fast.
I didn't see any benefit from the high clock, which is why mine runs
at 2000MHz all the time. It is more stable that way. The AthlonXP is
a fine processor, but the FSB is what holds it back. Now that
AMD has the memory controller on the processor, on their later sockets,
the FSB is no longer an issue.

Once you've verified how the multiplier jumpers work, then you can set the
CPU input clock to its final value of 166MHz (or whatever you can get away
with). Using 100Mhz initially, is just while you're verifying the multiplier
jumpers. Not all multiplier settings will result in the processor running,
and I believe values like "3X" or "4X" don't work (fab51.com table has a
strikeout line through those choices).

HTH,
Paul

 

[David Farber]

Some motherboards need a BIOS upgrade. They use less power and run
cooler but have the same performance. Ideal for overclocking.

So if it does need a BIOS upgrade, how dysfunctional might it be until I do
it? Can I assume there is no possibility of damaging the CPU before the
upgrade?

Thanks for your reply.

 

[David Farber]

I bought a 2600+ AthlonXP-M, and this is my experience.

1) Multiplier is unlocked. It helps to have a motherboard where
you can set the multiplier in the BIOS.
2) The first time you start up the motherboard, the CPU may be running
at 600 to 800MHz (I don't remember the exact figure - the default
multiplier is a relatively low value).
3) The VID coding is slightly different than a desktop processor.
That means at startup, it won't request the exact same voltage
as a desktop processor would. Since your processor is a 3000+ though,
the default voltage it asks for, will be pretty high, and probably
not that much different than normal. Some BIOS use the VID they read
from the processor, to determine the voltage adjustment range, and
that can present problems for people who undervolt, for cool operation.

For VID codings, the table on fab51.com is good. See the table in the
section entitled "L11 : Code to CORE Voltage Definition". My processor
is a "1.450(Q)", meaning AMD wanted it to run at 1.45V when used as
a 2600+ mobile. But when plugged into my motherboard, the Vcore regulator
runs it at 1.575V, just like the fab51.com table shows. This is OK, as my
eventual voltage selection was 1.65V for best stability. If I go much
higher than that in voltage, in fact my motherboard won't POST.

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

There is a table of processors here.

http://web.archive.org/web/20031018050306/http://www.qdi.nl/support/CPUQDISocketA.htm

Notice that a 3200+ Barton, has an actual core clock rate of 2000MHz.
With my brand new mobile processor, I set the input clock to the CPU
to 200MHz. That makes the FSB run at FSB400. For a multiplier, I
set the multiplier to 10x, so that 200x10 = 2000MHz, so that my
processor would be overclocked to be a top of the line 3200+.
I set the Vcore voltage in the BIOS to 1.65V, and tested for stability
with Prime95 (Orthos is also good for this).

http://sp2004.fre3.com/beta/beta2.htm (Orthos tester)

The multiplier is also known as the FID (frequency ID). It is a five bit
value, allowing 32 possible codings. My motherboard has hardware control
signals for only the lower four bits of the FID. This means my motherboard
has limited control of the multiplier. I can select multipliers in the
"low" range. The low range goes up to a multiplier of 12.5. To select a
higher multiplier than that on my motherboard, I have to stick a wire in
the S462 socket, to short a couple holes together. That would expose multipliers
from 13x upwards.

So in my case, my choices are easy ones. I use FSB400, so that I won't need
a high multiplier value to get good use from the processor. If you used
a low FSB, like FSB266, then you'd need a higher value of multiplier, to
get the core cranking along.

The Asrock motherboard in question, supports FSB333. That gives 166MHz feeding
the processor. To run it like it was an AthlonXP 3200+, the multiplier would
need to be 12x. 166 * 12 = 2000MHz. Just a matter of seeing if the motherboard
can set the multiplier or not. OK, I checked the manual and it has jumpers for
FID Should be loads of fun... No need for wire tricks at all.

http://www.asrock.com/mb/overview.asp?Model=K7VT4A+&s=462

Note - if you don't trust the table in the manual, in terms of
defining the multiplier values correctly, try running the processor
with the CPU input clock running at 100MHz. Then, if you accidentally
select a much higher than normal multiplier value, then there is
less chance of harm to the processor. 100x24 is 2400MHz, and my processor
can run at that speed, and so should yours be able to run that fast.
I didn't see any benefit from the high clock, which is why mine runs
at 2000MHz all the time. It is more stable that way. The AthlonXP is
a fine processor, but the FSB is what holds it back. Now that
AMD has the memory controller on the processor, on their later sockets,
the FSB is no longer an issue.

Once you've verified how the multiplier jumpers work, then you can set the
CPU input clock to its final value of 166MHz (or whatever you can get away
with). Using 100Mhz initially, is just while you're verifying the multiplier
jumpers. Not all multiplier settings will result in the processor running,
and I believe values like "3X" or "4X" don't work (fab51.com table has a
strikeout line through those choices).

HTH,
Paul

Hi Paul,

Thanks for the in depth information. I'd really like to make this as simple
as possible. I believe the XP3000 runs at 2200MHz. So should I set the fsb
to 100 and the multiplier to 22 to start? Then if successful there, change
the fsb to 133 and the multiplier to 16.5? I was trying to figure out if
anyone had tampered with the bridges on the CPU. Would that be obvious? I am
not trying to overclock this system. It would be great to have it operate
with its default settings. I did not see my processor mentioned on the fab51
web site.

 

[Paul]

Hi Paul,

Thanks for the in depth information. I'd really like to make this as simple
as possible. I believe the XP3000 runs at 2200MHz. So should I set the fsb
to 100 and the multiplier to 22 to start? Then if successful there, change
the fsb to 133 and the multiplier to 16.5? I was trying to figure out if
anyone had tampered with the bridges on the CPU. Would that be obvious? I am
not trying to overclock this system. It would be great to have it operate
with its default settings. I did not see my processor mentioned on the fab51
web site.

As for settings, I would use as high an FSB as you can manage.
Say that is 166MHz, for FSB333. Then, select a multiplier.
2200/166 = 13 or so. Try 13. Try 13 while clock is 100, for 1300 as
your first attempt. What you'll be verifying, is the jumper choice
of 13 is actually 13. Then crank the clock to 166, and that should
give 166 x 13 = 2158 on your second attempt. To reach that speed,
you may also want to use a bit more Vcore. You can use a memtest86+
floppy or CD initially, for stability testing. Once you are comfortable
with your efforts, then you can try booting into Windows, and trying
stuff like Orthos, for the real stability test. A Linux Live CD, like
Knoppix, can also be used for booting, as it doesn't need a hard drive,
and you can run the machine with just an (incorruptable) CD drive.
A Linux Live OS, is a "read only" OS, in the sense that temp files
are stored in a RAM disk. I would only boot into Windows, if I was
fairly certain the thing was really stable.

Paul

 

[Paul]

As for settings, I would use as high an FSB as you can manage.
Say that is 166MHz, for FSB333. Then, select a multiplier.
2200/166 = 13 or so. Try 13. Try 13 while clock is 100, for 1300 as
your first attempt. What you'll be verifying, is the jumper choice
of 13 is actually 13. Then crank the clock to 166, and that should
give 166 x 13 = 2158 on your second attempt. To reach that speed,
you may also want to use a bit more Vcore. You can use a memtest86+
floppy or CD initially, for stability testing. Once you are comfortable
with your efforts, then you can try booting into Windows, and trying
stuff like Orthos, for the real stability test. A Linux Live CD, like
Knoppix, can also be used for booting, as it doesn't need a hard drive,
and you can run the machine with just an (incorruptable) CD drive.
A Linux Live OS, is a "read only" OS, in the sense that temp files
are stored in a RAM disk. I would only boot into Windows, if I was
fairly certain the thing was really stable.

Paul

Come to think of it, maybe my board was set for 2200MHz (200x11).
It's been a while since I used that machine (the CMOS battery has
even run flat on it).

The main thing about the XP-M, is the unlocked multiplier. If a
motherboard has absolutely no support for multipliers, you'd be stuck
using 6X as the default, and have no way to get the rated speed.
(Wire tricks to the rescue...) With a motherboard that has jumpers,
it should be pretty easy to get something better than the defaults.

I don't know if there is much point tampering with a Mobile. Maybe
someone could take a regular AthlonXP, and use the bridges to change
it to a Mobile. I don't know if it becomes unlocked by doing that
or not. I suppose if it did, then everyone would be taking their
"superlocked" newer processors, and doing that to them. So if the
multiplier responds on yours, that is about all I'd be concerned about.

The default VID value should be another proof it is the real thing.
Finding a readout of "1.575V" in the BIOS, is not something you'd
typically find on a desktop processor.

The difference between XP-M 3000+, 2600+, 2500+, might be how
easy it is to get the higher operating speeds. I think there is
a slight correlation to top overclock, with the model used. But
you could basically take any of those DTR mobiles and get something
pretty impressive from them. So if yours is unlocked, and has
no trouble reaching your target speed without an excessive setting
for Vcore, I'd be pretty satisfied with it. Mine has run pretty
cool, so no complaints there.

Paul

 

[David Farber]

Come to think of it, maybe my board was set for 2200MHz (200x11).
It's been a while since I used that machine (the CMOS battery has
even run flat on it).

The main thing about the XP-M, is the unlocked multiplier. If a
motherboard has absolutely no support for multipliers, you'd be stuck
using 6X as the default, and have no way to get the rated speed.
(Wire tricks to the rescue...) With a motherboard that has jumpers,
it should be pretty easy to get something better than the defaults.

I don't know if there is much point tampering with a Mobile. Maybe
someone could take a regular AthlonXP, and use the bridges to change
it to a Mobile. I don't know if it becomes unlocked by doing that
or not. I suppose if it did, then everyone would be taking their
"superlocked" newer processors, and doing that to them. So if the
multiplier responds on yours, that is about all I'd be concerned about.

The default VID value should be another proof it is the real thing.
Finding a readout of "1.575V" in the BIOS, is not something you'd
typically find on a desktop processor.

The difference between XP-M 3000+, 2600+, 2500+, might be how
easy it is to get the higher operating speeds. I think there is
a slight correlation to top overclock, with the model used. But
you could basically take any of those DTR mobiles and get something
pretty impressive from them. So if yours is unlocked, and has
no trouble reaching your target speed without an excessive setting
for Vcore, I'd be pretty satisfied with it. Mine has run pretty
cool, so no complaints there.

Paul

Hi Paul,

I initially set the multiplier to 20 and the fid to 100MHz. It powered up
and then I entered the setup screen. The settings there said that the CPU
was running at 1,100 MHz, 100MHz with a multiplier of 11. I thought that was
strange. Then I noticed there was a multiplier setting as one of the
software choices. The choices were limited to numbers that would not bring
the processor close to its rated speed so I shut it down and reset the fid
to 133MHz then powered it back on. Again, the CPU status was reading a
multiplier of 11 but this time the software multiplier choices had a higher
upper limit. I set it for 16.5 and then reboot. Now, the setup status says
the CPU is running at 2,200MHz.

The reality seems to be is that the multiplier jumpers on the motherboard
are being ignored by the BIOS. Maybe there's a manual setting I'm missing
but I'm happy with the way it is working now. Of course I haven't installed
the hardware yet but I like the start so far.

Thanks for your support.