AMD Athlon XP version

I've noticed that when I change the CPU speed on my AOpen AK79D-400
Max motherboard the
processor model designation as reported by BIOS during POST, and as
reported by Windows
XP Pro Device Manager, also changes. For example, when running the
processor at 2.00 GHz I see
AMD Athlon(tm) XP 2600+, whereas if I drop the speed down to 1.67 GHz
I see AMD Athlon(tm) XP 2600+.
I had always thought that the "2600+" or whatever was part of the
processor version designation, burned into the
chip during manufacture.

What am I missing?

TIA

Ed

Ed wrote:
> I've noticed that when I change the CPU speed on my AOpen AK79D-400
> Max motherboard the
> processor model designation as reported by BIOS during POST, and as
> reported by Windows
> XP Pro Device Manager, also changes. For example, when running the
> processor at 2.00 GHz I see
> AMD Athlon(tm) XP 2600+, whereas if I drop the speed down to 1.67 GHz
> I see AMD Athlon(tm) XP 2600+.
> I had always thought that the "2600+" or whatever was part of the
> processor version designation, burned into the
> chip during manufacture.
>
> What am I missing?
>
> TIA
>
> Ed

The only thing that is "burned into the chip", is the model number.
As in "model 10" means it is a Barton. All other parameters are
controlled by the bridge resistors on the top of the chip.

This is one of the reasons it is so easy to counterfeit chips.
The easiest, for example, was changing a 2500+ to look like a 3200+.
All you had to do for that one, was change the label affixed to
the top of the chip. (Sometimes the fonts used, aren't quite right.)

Even the major retailers have sold counterfeit processors on occasion.
So it was a real problem.

Some of the counterfeits were easy to catch. One poster, for example,
plugged in a new chip he bought, and could not understand why the
computer was using such a high Vcore voltage. It turned out, the VID
bridges had been changed by the counterfeiter, to a higher than normal
value. In one case, they even used a matching paint, on the bottom
of the chip, to cover up some conductive material added to the
processor to change a bridge value.

AMD could have used other technologies. They could have added fuses
which could be programmed once at the factory. They could have used
a laser technology, to burn and program characteristics on the line.
They could have added some EEPROM that they set at the factory
(but the extra process steps would have added to manufacturing cost).
Instead, they chose to leave everything in the open, and while it
took a while to figure out, eventually almost all their secrets
were uncovered.

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

http://www.ocinside.de/go_e.html?/ht...md_pinmod.html

In the BIOS, the identity is, as you note, based on the operating
speed at the time. And is why a 2500+ becomes a 3200+, as far as the
BIOS is concerned, just by raising the CPU input clock via the BIOS.

Paul

"Paul" <> wrote in message
news:ftuatp$tt4$...
> Ed wrote:
>> I've noticed that when I change the CPU speed on my AOpen AK79D-400
>> Max motherboard the
>> processor model designation as reported by BIOS during POST, and as
>> reported by Windows
>> XP Pro Device Manager, also changes. For example, when running the
>> processor at 2.00 GHz I see
>> AMD Athlon(tm) XP 2600+, whereas if I drop the speed down to 1.67
>> GHz I see AMD Athlon(tm) XP 2600+.
>> I had always thought that the "2600+" or whatever was part of the
>> processor version designation, burned into the
>> chip during manufacture.
>>
>> What am I missing?
>>
>> TIA
>>
>> Ed
>
> The only thing that is "burned into the chip", is the model number.
> As in "model 10" means it is a Barton. All other parameters are
> controlled by the bridge resistors on the top of the chip.
>
> This is one of the reasons it is so easy to counterfeit chips.
> The easiest, for example, was changing a 2500+ to look like a 3200+.
> All you had to do for that one, was change the label affixed to
> the top of the chip. (Sometimes the fonts used, aren't quite right.)
>
> Even the major retailers have sold counterfeit processors on
> occasion.
> So it was a real problem.
>
> Some of the counterfeits were easy to catch. One poster, for
> example,
> plugged in a new chip he bought, and could not understand why the
> computer was using such a high Vcore voltage. It turned out, the VID
> bridges had been changed by the counterfeiter, to a higher than
> normal
> value. In one case, they even used a matching paint, on the bottom
> of the chip, to cover up some conductive material added to the
> processor to change a bridge value.
>
> AMD could have used other technologies. They could have added fuses
> which could be programmed once at the factory. They could have used
> a laser technology, to burn and program characteristics on the line.
> They could have added some EEPROM that they set at the factory
> (but the extra process steps would have added to manufacturing
> cost).
> Instead, they chose to leave everything in the open, and while it
> took a while to figure out, eventually almost all their secrets
> were uncovered.
>
> http://fab51.com/cpu/barton/athlon-e23.html
>
> http://www.ocinside.de/go_e.html?/ht...md_pinmod.html
>
> In the BIOS, the identity is, as you note, based on the operating
> speed at the time. And is why a 2500+ becomes a 3200+, as far as the
> BIOS is concerned, just by raising the CPU input clock via the BIOS.
>
> Paul

Thanks, Paul. That's interesting. The processor came with an Asus
A7N8X-X mobo that I bought on eBay. It was advertised
as an Athlon XP 2600+. The board turned out to be bad, and the seller
went underground, so I shouldn't be surprised if the
chip was counterfeit. I tried to read the label but couldn't make it
out. I currently have it set to run at 2.00 GHz and it seems
to be stable. Windows Device Manager shows it as a 2600+ at that
speed. VCore is 1.65v. Is that about right?

I still don't understand why the BIOS reported designation for the
processor would change like this. I could see HOW it could be done,
e.g., the BIOS looking up 2000+ or 2600+ or whatever from a fixed
table built into the processor based on user-selected bus frequency,
but what would be the point? It like Ford sticking a 3.8L V6 badge on
the trunk, but if I manage to get to go 0-60 6 seconds or whatever the
trunk badge changes to 4.5L V8.

I was going to see if the Asus A7V266-E in my other machine behaves
the same way. It has an Athlon XP 1800+ (Model 6 according the Device
manager). I normally run it at 1.5 GHz, so I went into setup and
bumped it up to 2.00 or so GHz. After rebooting it still said
Athlon XP 1800+. So, I concluded that it's only AOpen that fudges the
reported processor model designation. But now I see that my
A7V266-E/Athlon XP 1800+ machine has reverted back to 1.5 GHz. Don't
know why.

Anyway, thanks for your interest.

Ed

Ed wrote:

>
> Thanks, Paul. That's interesting. The processor came with an Asus
> A7N8X-X mobo that I bought on eBay. It was advertised
> as an Athlon XP 2600+. The board turned out to be bad, and the seller
> went underground, so I shouldn't be surprised if the
> chip was counterfeit. I tried to read the label but couldn't make it
> out. I currently have it set to run at 2.00 GHz and it seems
> to be stable. Windows Device Manager shows it as a 2600+ at that
> speed. VCore is 1.65v. Is that about right?
>
> I still don't understand why the BIOS reported designation for the
> processor would change like this. I could see HOW it could be done,
> e.g., the BIOS looking up 2000+ or 2600+ or whatever from a fixed
> table built into the processor based on user-selected bus frequency,
> but what would be the point? It like Ford sticking a 3.8L V6 badge on
> the trunk, but if I manage to get to go 0-60 6 seconds or whatever the
> trunk badge changes to 4.5L V8.
>
> I was going to see if the Asus A7V266-E in my other machine behaves
> the same way. It has an Athlon XP 1800+ (Model 6 according the Device
> manager). I normally run it at 1.5 GHz, so I went into setup and
> bumped it up to 2.00 or so GHz. After rebooting it still said
> Athlon XP 1800+. So, I concluded that it's only AOpen that fudges the
> reported processor model designation. But now I see that my
> A7V266-E/Athlon XP 1800+ machine has reverted back to 1.5 GHz. Don't
> know why.
>
> Anyway, thanks for your interest.
>
> Ed
>

What they have to work with, is reading the multiplier (FID). And
then they have a table like this to work with. You can see that the
Barton 3200+ and 2500+ have the same multiplier, and the same
model number (10), and the only difference is the CPU clock.
There are two pins on the bottom of the CPU, FSB_SENSE[1:0],
and those should indicate which processor is which (166MHz or 200MHz).
What I don't know, is if FSB_SENSE is readable in software, or has
to be supported in hardware on the motherboard, to enumerate
it properly.

Family Core P.R. Pkg CPU Cache Mult Core Tmax Power
Freq Clk Volts

XP Model 10 2200 (3200+) OPGA 200 512 11x 1.65V 85oC 60.4W
Barton 2100 (3000+) OPGA 200 512 10.5x 1.65V 85oC 53.7W

XP Model 10 2167 (3000+) OPGA 166 512 13x 1.65V 85oC 58.4W
Barton 2083 (2800+) OPGA 166 512 12.5x 1.65V 85oC 53.7W
1917 (2600+) OPGA 166 512 11.5x 1.65V 85oC 53.7W
1833 (2500+) OPGA 166 512 11x 1.65V 85oC 53.7W

XP Model 8 2167 (2700+) OPGA 166 256 13x 1.65V 85oC 62.0W
Thoroughbred 2083 (2600+) OPGA 166 256 12.5x 1.65V 85oC 62.0W

XP Model 8 2133 (2600+) OPGA 133 256 16x 1.65V 85oC 62.0W
Thoroughbred 2000 (2400+) OPGA 133 256 15x 1.65V 85oC 62.0W
CPU ID 0681 1800 (2200+) OPGA 133 256 13.5x 1.60V 85oC 57.0W
1733 (2100+) OPGA 133 256 13x 1.60V 90oC 56.3W
1667 (2000+) OPGA 133 256 12.5x 1.60V 90oC 55.7W
1533 (1800+) OPGA 133 256 11.5x 1.60V 90oC 55.7W
1467 (1700+) OPGA 133 256 11x 1.60V 90oC 55.7W

XP Model 8 1800 (2200+) OPGA 133 256 13.5x 1.65V 85oC 61.7W
Thoroughbred 1733 (2100+) OPGA 133 256 13x 1.60V 90oC 56.4W
CPU ID 0680 1667 (2000+) OPGA 133 256 12.5x 1.65V 90oC 54.7W
1667 (2000+) OPGA 133 256 12.5x 1.60V 90oC 54.7W
1600 (1900+) OPGA 133 256 12x 1.50V 90oC 47.7W
1533 (1800+) OPGA 133 256 11.5x 1.50V 90oC 46.3W
1467 (1700+) OPGA 133 256 11x 1.50V 90oC 44.9W

XP Model 6 1733 (2100+) OPGA 133 256 13x 1.75V 90oC 64.3W
Palomino 1667 (2000+) OPGA 133 256 12.5x 1.75V 90oC 62.5W
1600 (1900+) OPGA 133 256 12x 1.75V 90oC 60.7W
1533 (1800+) OPGA 133 256 11.5x 1.75V 90oC 59.2W
1467 (1700+) OPGA 133 256 11x 1.75V 90oC 57.4W
1400 (1600+) OPGA 133 256 10.5x 1.75V 90oC 56.3W
1333 (1500+) OPGA 133 256 10x 1.75V 90oC 53.8W

But it became common practice, when you got a 3200+, to find
it running (and indicating) 2500+. So for whatever reason, there
seemed to be lots of processors running at the wrong speed.
So people didn't think about it too much, when they were told
to turn up the clock to get it running at the right speed.
That is what made passing off 2500+ processors as 3200+
processors, so easy. Just change the label, and leave it to
the user to "turn-em up".

This is back in the day, when AMD did a decent job of doing a datasheet.
For the AM2/AM2+, it seems they've fired all their tech writers, and have
just given up on documentation. Finding the info you want for the recent
stuff, is getting harder and harder. (They don't document bridges in
the datasheet, to make it harder to reverse engineer.)

http://www.amd.com/us-en/assets/cont...docs/26237.PDF

Paul