AMD survived to this day thanks to IBM first & Intel second.
AMD seems to forget that way to soooooon.
Other MFGs(Cypress & VIA & IDT) came out with clones but because they
are blocked from x86 technology. This is very easy for Intel to kill
them. Intel just simply comes out with a new technology. They are very
slow to react because they have to do reverse enginerring first. That
takes time --> doom from the beginning.

Your history is mostly correct, granted, with a whole lot of bias.
But, for the most part, it has nothing to do with AMD's current
predicament. Also, don't forget that AMD beat Intel to the market with
64 bit architecture and multi-core architecture. And, for the last
years of the x86 lifecycle, AMD was beating Intel's x86 chips. The
original name for the 64 bit instruction set used in the industry was
AMD64.

The way I interpret this history is as follows:
1) AMD was created to be a little brother competitor to Intel in the
x86 world
2) AMD caught up to Intel, took massive parts of its x86 market share,
and produced a better x86 chip (this is circa the P4/Athlon XP days)
3) AMD decided to innovate and invent AMD64, rather than continue to
obsess the consumer with raw megahurtz, which can be quite unrelated
to performance games.
4) Intel, failing to significantly innovate for about 15 years as it
banked on the x86 success, copies AMD.
5) AMD decides to innovate again, and begin pushing multi-core chips
to the consumer. Intel lags.
6) AMD redoes its architecture to ensure better scaling as more cores
are added. Intel continues to turn up the voltage, and obsess
customers with raw megahurtz.

The first post had it spot on. AMD had the lead and got greedy. They
decided to take all the money and not really push on R&D (Why no new
product since 2003).
Now they are screwed.
Although AMD said they scrapped a project, supposedly the next
processor which was meant to be before barcelona. So barcelona should
offer 2 generations of performance. I doubt it will though.

Anyway....

Multi core is not an AMD innovation. 64bit is not an AMD innovation.
You my friend are confusing innovation with evolution.

Intel is not pushing mhz. Their cpus outperform AMD's hz for hz and
clock higher anyway.
Stop worshiping AMD. The work they are doing is great, but it is
simply evolution. Great they are doing native quad core. That is
evolution. What about Native Octo core, 16, 32 etc? Will you be
calling that innovation too? It is evolution. Just because AMD is
first to do it doesn't mean its the right thing to do. Why not just
make 16 core native and be done with it?

Intel is not doing native quad yet because the die size is so large it
is not economical to do so. When you get a defect in your die if you
have a larger core the chances of the defect being on the area where
the core is are greatly increased. Then you have to throw away a whole
core. With the glued approach you can simply throw away 1 dual core
and pick another one to make a quad. Infact chances are for Intel
because they stack so much cache on their cpus it will hit the cache
they'll disable it and sell it as a cheap end product.

The original poster made it quite clear that he thought AMD was
getting a free ride from Intel's design of the x86, and is now doing
poorly because that free ride is running out.

I am simply pointing out that the movement from standard x86 to 64-bit
processing and/or multi-core processing was not AMD falling off of
Intel's coat tails. It was AMD's own work! Their R&D teams beat Intel
to the marketplace in moving beyond x86. Period.

A) Frankly, it would be better for AMD to be loud and obnoxious about
a really great product. It is also very unlike them. They have a
history of not releasing benchmarks or surrounding new releases with
much marketing. Ruiz is quoted as saying that a product should sell
itself. Therefore we can't use their verbosity (or lack thereof) to
read into either the validity of the demo, or the general quality of
the product.

B) Correct. Nothing here to be conclusive about.

C) Good question. Reporters for the Inquirer have are known to
frequently ignore NDAs.

D) No reliability guaranteed. But, there's also no strong evidence for
heavy doubts beyond reasonable skepticism.

After thinking things over, with your questions in mind, I think that
this really must be intentional. AMD has been very clear that while
they may or may not have 3.0+ Ghz Phenoms/Barcelonas, the first
Barcelonas to hit shelves (at the release on Sep. 10) will only be
2Ghz. AMD wants to, and needs to, get Barcelona to shelves as soon as
possible; clock-speeds are irrelevant. AMD has several enormous orders
for Barcelonas waiting to be filled for super-computers, and its
customers are beginning to get miffed. Give the channel the 2Ghz
Barcelonas promised to fill standing orders, and then follow on to the
3Ghz Barcelonas when the fabs are fully ramped. In the mean time, with
all the PR problems AMD is having, the last thing they need is to be
bragging about 3Ghz processors while only 2Ghz processors are on
shelves. This doesn't mean there are no 3Ghz Barcelonas - it means, as
we've known for a while, they won't be the first ones to hit shelves.
And, consequently, they won't be the first ones advertised. Simple.

how do the 2ghz Barcelona's compare to intels current quad cores,
anyone know?

jownewen, yes, Intel currently uses 2 dual core processors and
packages them in the same chip to make a "quad core". AMD will be the
first to release a native quad core with Barcelona. Intel's quad core
product Nehalem is scheduled for release in mid-2008. This has certain
implications for memory intensive applications since Intel's chips
must rely on a relatively slow front-side-bus for inter-chip
communications and operations. However, they are able to make up for
it by having a very large shared cache (between dual cores) and
reorderable smart queues.

By the time AMD can hit 2.6GHz, Intel will have 3.3GHz. AMD better
have something that can seriously outperform Intel or they will be
stuck selling budget processors for another quarter.

However, we have enough information to make our own estimates!

A 2.6GHz Barcelona is roughly equivalent to a 3.26GHz 65nm Xeon.

http://arstechnica.com/news.ars/post/20070807-new-benchmarks-intels-4...
However, according to benchmarks of 45nm Core2s, a 2.33GHz Penryn is
11% faster than a 2.33GHz Conroe... Or a 3.3GHz Penryn is equivalent
to a 3.6GHz Xeon.

So we have a 2.6GHz Barcelona coming out against a 3.3GHz Penryn. The
2.6GHz Barcelona would have been equal to a 2.9GHz Penryn; it's too
bad then that the 3.3GHz Penryn will be 14% faster then the 2.6GHz
Barcelona.

And yes, AMD can ramp up, but then again so can Intel.

http://www.dailytech.com/article.aspx?newsid=7574

michae...@gmail.com schreef:

AMD will have a one-year architectural advantage
Intel will have the advantage of a smaller production process

and for innovation vs. evolution
from dual to quad to octo and so on is indeed evolution
changing from a Mhz/Ghz battleground to Multiple cores on the same die
however is innovation
just like AMD's Fusion (system on a die) with an integrated graphics
core on-die is innovation

You obviously don't understand how processors are made and the factors
that are involved when producing them. A larger die size means the
chances of a defect in the product are greatly increased. Defect =
Thrown in bin.

Links. Search for die size. Barcelona is a LOT bigger than any
previous generations.
http://en.wikipedia.org/wiki/Image:Amd_cpu_die_compare.jpg
http://en.wikipedia.org/wiki/List_of_Intel_Core_2_microprocessors
http://www.intel.com/pressroom/archive/releases/20070328fact.htm

http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_861_...
http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_9485...

I'm not sure why you quoted me on the hz for hz. Intel outperform AMD
clock for clock (which implys at the same speed) I wasn't talking
about the past, I was talking about now.

AMD doesn't have an arch advantage. Stop this native nonsense. Intel's
glued approach works perfectly well as can be seen in many multi-
threaded applications. Doesn't 3dsmax nearly gain 100% with an Intel
quad? What is AMD planning on doing. Gaining 300% by just doubling the
cores? Don't be stupid.

Intel may learn to make a true quad core and release its "micro-die"
chips by the end of this year, and what about same time AMD making a
octa-core using Intel's gluing trick... ;). (I am not underestimating
Intel; they are the masters of glue and very much capable of gluing 2
glued quad, and make a glued, glued 8-core).

evolution - not - innovation:
I think this is the most common jealousy dialog to any master piece.
 "Yea that's exactly I was thinking about"

2 cores on a die..yes I guess I could accept that is innovation,
although we already had 2 processors on one motherboard, so when it
became economical to they thought "hey lets put 2 processors into 1
processor". Now you can make your own decisions to why this has
happened at the time it happened. I'd suggest you take a look over the
last 10-15 years of cpu history

But 4 is simply adding more to the already existing principal. If you
are going to say 4 cores is innovation then imo you must accept that
2000gb hard drives are innovation and that a stick of ram with 16gb is
innovation.

There is no learning to make a true quad core. They know how to do it.
If you know the differences between native and glued I should think it
is obvious that it is not hard to do.

I'd like to see AMD try to glue 2 Barcelonas. I wouldn't like to see
Intel glue 4 quad cores. I can't see that happening either. The heat
would be far too much. We're talking 4 x 107mm2 die. Also 2x 283mm2
die!

Let's just take a look back at history eh?

http://balusc.xs4all.nl/srv/har-cpu-amd-k6.php
http://balusc.xs4all.nl/srv/har-cpu-amd-k7.php
http://balusc.xs4all.nl/srv/har-cpu-amd-k8.php
http://balusc.xs4all.nl/srv/har-cpu-amd-k9.php

283mm2 is quite a bit larger than all of those.

The best part about me looking these up is the other day I saw a
comment from an established AMD fan boy about how K8 yields were awful
in 2004. Now look at the die size of K8 back then.  AMD have since
made larger ones but they are premium products and they also have 3
years of practice behind them.

Then there is roborat64's analysis. There are a lot of comments worth
reading too. I'd personally ignore the money side of it too and look
at the basic economics of production.
http://roborat64.blogspot.com/2007/05/bad-economics-of-65nm-monolithi...

New product + large die = Bad idea.

I don't think it is impossible to glue 2 Barcelona or 2 Duo, the
temperature issue can be solved by a good cooling system (buy
Thermaltake after ATI?).

If the whole thing is an evolution, it look like they both started
from same monkey and at this age Intel evolved into a king Kong and
AMD become a cave man.
In Intel's story its all about GHz (still solving the FSB puzzle?),
and I will laugh if somebody says a 3GHz Celeron beats a 2GHz P4, so
whats the 1Ghz extra muscle?

Just reducing the die won't boost any performance, for that you need
improve or try some new architecture. I don't think the order really
matters (except some rearrangements? at most remove couple of boxes of
cache?), either you can put a new architecture first and reduce the
die size later, or reduce the size first and look for a new
architecture to fit in.