What makes P4 2.6MHz different from, say, P4 2.8MHz?

[Jen]

I'm just curious, and certainly no expert. Also, new to this group, so
maybe it's already been answered.

What makes a P4 2.6 different from another P4 of the same class - the
same architecture, FSB speed, cache, etc.

Say it's a P4, northwood, socket 478, FC-PGA2 package, 512 cache,
130nm process, but one is 2.6MHz and one is 2.8MHz.

Are there differences in the chip fabrication masks? Are they made on
different fabrication lines? Made in different plants?

Are they all made on the same fab line, but are sorted into different
bins based on some sort of speed or parametric testing?

If this is so, are you really just getting a slightly defective part
if you buy a 2.4 verses a 3.2?

Is speed info, serial number, etc. just burned into the chip after it
has been packaged?

[The little lost angel]

On Tue, 30 Sep 2003 22:16:34 GMT, Jen <jen@home.org> wrote:Are they all made on the same fab line, but are sorted into differentbins based on some sort of speed or parametric testing?

That's the way they are sorted. Though in many cases as the design and
process matures, most chips will tend to fall into the higher speed
bins. The slower chips are then just chips taken from the higher speed
bins and marked as slower chips to meet orders. This is also the
reason why lower speed chips can generally be overclocked (run above
spec) better.
If this is so, are you really just getting a slightly defective partif you buy a 2.4 verses a 3.2?

Not really defective, just not capable of performing at 3.2Ghz but
perfectly okay at 2.4

BTW, it's Ghz not Mhz, hasn't been a x86 CPU operating in the 2~3Mhz
range ever I think :PppP

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

[Jen]

On Wed, 01 Oct 2003 04:33:00 GMT,
a?n?g?e?l@lovergirl.lrigrevol.moc.com (The little lost angel) wrote:
On Tue, 30 Sep 2003 22:16:34 GMT, Jen <jen@home.org> wrote:Are they all made on the same fab line, but are sorted into differentbins based on some sort of speed or parametric testing?That's the way they are sorted. Though in many cases as the design andprocess matures, most chips will tend to fall into the higher speedbins. The slower chips are then just chips taken from the higher speedbins and marked as slower chips to meet orders. This is also thereason why lower speed chips can generally be overclocked (run abovespec) better.If this is so, are you really just getting a slightly defective partif you buy a 2.4 verses a 3.2?Not really defective, just not capable of performing at 3.2Ghz butperfectly okay at 2.4BTW, it's Ghz not Mhz, hasn't been a x86 CPU operating in the 2~3Mhzrange ever I think :PppP

Brain fade - I started writing 3200MHz, etc, but then edited the whole
message to change to 3.2, etc.

Thanks for the other information. There must be a certain amount of
variation when Intel/AMD/other does their parametric testing. The
chips that are still within tolerances, but not quite "perfect", go
into the slower speed pile.

Is the speed rating, and other ID information burned into some
on-chip programmable memory after the chip is put into a package?

[Tony Hill]

On Tue, 30 Sep 2003 22:16:34 GMT, Jen <jen@home.org> wrote:I'm just curious, and certainly no expert. Also, new to this group, somaybe it's already been answered.What makes a P4 2.6 different from another P4 of the same class - thesame architecture, FSB speed, cache, etc.

In a word (err, two words): not much.
Say it's a P4, northwood, socket 478, FC-PGA2 package, 512 cache,130nm process, but one is 2.6MHz and one is 2.8MHz.Are there differences in the chip fabrication masks? Are they made ondifferent fabrication lines? Made in different plants?

No, no and no.
Are they all made on the same fab line, but are sorted into differentbins based on some sort of speed or parametric testing?

Yup, that's about it, at least to a certain extent. Market demand
also plays a major role in it. It turns out that Intel can make a lot
more profit by selling some 3.2GHz chips at very high prices and
2.4GHz chips at much lower prices than they could if they sold all
their chips at 3.2GHz. So, what this means is even if ALL the chips
coming off the line were capable of running at 3.2GHz, Intel would be
better off selling them at a variety of speed grades and price points,
rather then just selling 3.2GHz parts at a reduced price.
If this is so, are you really just getting a slightly defective partif you buy a 2.4 verses a 3.2?

Sort of, though saying "defective" makes it sound like the chip
doesn't work right. All chips work only within a certain range of
temperature, voltage, current, etc. They also have some limits in
clock speeds. One chip might be slightly "cleaner" than another, so
it will work without errors at up to 3.2GHz (plus a little extra
margin of error), another chip might work perfectly at 2.4GHz, but
when you start to clock it to 2.6GHz or beyond it might have some
troubles.

As mentioned above though, marketing also comes into play. It's quite
possible that if you buy a 2.4GHz chip it could easily run at 3.2GHz
with no troubles at all, but for marketing reasons Intel decided they
needed some chips to run at 2.4GHz. Nothing illegal or even really
immoral about this, Intel isn't "downgrading" the chips or anything
like that. They just happen to only be testing the chip to meet their
2.4GHz standard, and if the chip happens to exceed that standard, it
doesn't really matter.
Is speed info, serial number, etc. just burned into the chip after ithas been packaged?

Yup, or at least it is with some chips. With the AthlonXP you can
actually see this. There are some small "bridges" on top of the
processor carrier which determine the clock multiplier of the chip
(and therefore it's speed), it's voltage, whether it will work in
dual-processor setups, etc. AMD will burn out various connections to
make these settings, all done post-packaging. Intel may do the same
sort of thing at a different stage in the packaging process (I don't
really know where/when they set the clock speed and such settings),
but the concept is the same.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca