Intel COO signals willingness to go with AMD64!!

[chrisv]

Pathetic, isn't it? And I don't even get paid to write this garbage.

Not to worry. As soon as I finish my MBA at <wildly overrated local
college for morons>, I'll be joining a large consulting firm here, and
all the practice I've had writing useless garbage will begin to pay
off. .

Years of USENET experience will look great on your resume. 8)

 

[chrisv]

IA64 core inside desktop P4 CPUs would be the best bet for Intel to go 64bit.
x86-64 from Intel is the worst thing they could have done, ever.

They tried to make IA64 work. It looks like they've failed. I give
them credit for trying - at one time, the clean-sheet approach to a
new CPU probably made a lot of sense. Today, X86 is still doing well,
design costs (hardware and software) are spread over a bazillion units
sold, and it's just too tough to beat that combination.

Not only Intel will look weak to
follow the AMD proposed 64bit pseudo/hybrid rule at extending the ancient x86 architecture but it
would mean that IA64 might become a dead project not because its architecture its worse but because
customers are quite retarded to promote an x86-64 hybrid instead.

Damn you're dumb. Those "retarded" customers, who seem to be
clamoring for X86-64, are what a business must satisfy if they want to
succeed.

With X86-64, there's a "cheap and easy" route to 64-bit computing. So
there's little reason go go the "expensive and painful" route of IA64.

 

[chrisv]

Concede-to-amd Technology?
Caught-with-our-pants-down Technology?
Catch-up Technology?
Cough-amd-cough Technology?
Change-course Technology?
...

Coat-Tails

 

[geno_cyber]

They tried to make IA64 work. It looks like they've failed. I give
them credit for trying - at one time, the clean-sheet approach to a
new CPU probably made a lot of sense. Today, X86 is still doing well,
design costs (hardware and software) are spread over a bazillion units
sold, and it's just too tough to beat that combination.


Damn you're dumb. Those "retarded" customers, who seem to be
clamoring for X86-64, are what a business must satisfy if they want to
succeed.

With X86-64, there's a "cheap and easy" route to 64-bit computing. So
there's little reason go go the "expensive and painful" route of IA64.

If x86 was such a great architecture, which is not, then Alpha CPUs would have been x86 based
instead what's happening nowadays is that AMD pushed Alpha CPU technologies inside the ancient x86
architecture that was originally designed without any supercomputer stuff in mind, Intel was trying
to do the same on its own but then bought the rest of the Alpha staff to push EPIC IA64 new
architecture.
The fact that Intel invented x86 in first place and they decided that was time to kill it without
continuing to extend it should be more than enough to know that continuing injecting Alpha CPU
architecture inside such a limited and far from really efficient although inexpensive technology in
order to make it work like it was never meant to , simply won't last forever.
Instead AMD , which surely has less money than Intel to spend on R&D and so couldn't afford
designing and entire new architecture to go 64bit, decided to push the x86-64 hybrid extension and
the nonsense market it's pushing Intel managers to kill IA64 just like Alpha was killed by x86
invasion, which is pretty silly to say the least... Intel killed and bought the majority of the
Alpha staff and market thanks to x86 but then its plans were to use Alpha technology and expertise
to design a whole new architecture, the IA64 and stop x86. Instead AMD had to go cheap and decided
to propose an x86-64 extension on its own... and IA64 , which would have been the real revisioned
and upgraded Alpha architecture, is going to die thanks to x86-64 from AMD because Intel managers
feel that they must follow this route to please customers now , despite the fact that thanks to this
route the real PC efficiency and architecture will be far worse than what IA64 would be able to
deliver in a few years from now and surely worse than if Alpha CPUs would still exist in the market.

There's nothing to be happy or satisfied about this big push of the x86-64 if it means the death of
IA64 and no IA64 desktop CPUs, just like x86 was worse compared to Alpha this x86-64 further
extension is worse than IA64.

If people think that x86 it's still a clean design .. oh well, just think about the fact that has
been patched thousand times in its 30 years of life since the first 8bit and 16bit releases....

 

[Felger Carbon]

The AMD64 extensions were a long overdue tidying up of the x86 instruction
set. However, even with a non-tidy instruction set as in x86-32, it was
still doing it's job just fine.

Hear, hear! ;-)

Nothing like an annual production run of 100 million+ CPUs to keep an
architecture on its toes.

 

[Nate Edel]

The AMD64 extensions were a long overdue tidying up of the x86 instruction
set. However, even with a non-tidy instruction set as in x86-32, it was
still doing it's job just fine.

Errr, yes and no.

As more and more systems were approaching 2gb or more physical memory, the
4gb address space limit of the x86 architecture was getting tight. PAE
works, but it's creaky...

 

[Felger Carbon]

The fact that Intel invented x86 in first place and they decided

that was time to kill it without

continuing to extend it should be more than enough to know that continuing injecting Alpha CPU
architecture inside such a limited and far from really efficient

although inexpensive technology in

order to make it work like it was never meant to , simply won't last forever.
Instead AMD , which surely has less money than Intel to spend on R&D and so couldn't afford
designing and entire new architecture to go 64bit, decided to push

the x86-64 hybrid extension and

the nonsense market it's pushing Intel managers to kill IA64 just like Alpha was killed by x86
invasion, which is pretty silly to say the least... Intel killed and bought the majority of the
Alpha staff and market thanks to x86 but then its plans were to use Alpha technology and expertise
to design a whole new architecture, the IA64 and stop x86. Instead

AMD had to go cheap and decided

to propose an x86-64 extension on its own... and IA64 , which would have been the real revisioned
and upgraded Alpha architecture, is going to die thanks to x86-64

from AMD because Intel managers

feel that they must follow this route to please customers now ,

despite the fact that thanks to this

route the real PC efficiency and architecture will be far worse than what IA64 would be able to
deliver in a few years from now and surely worse than if Alpha CPUs

would still exist in the market.

Geno, I think you should consider writing shorter sentences. It would
make your logic easier to follow. Honest.

 

[David Schwartz]

Geno, do you think a Xeon with 64-bit extensions is a better product
than one without them? Your argument is basically that because x86 is bad,
it shouldn't be made better.

I still think the best move for Intel would be to bundle a Pentium and
an Itanium into a single package. In fact, they should have done this *long*
ago, improving the 32-bit and 64-bit processors in parallel.

This would given both 32-bit code and 64-bit code a place to go,
supporting operating systems that run natively on either or both processors.
Initially, the 64-bit processor could be used as an accelerator for
multimedia operations and gradually running natively the increasing supply
of 64-bit code. Ultimately, the 32-bit processor would wind up as an I/O
coprocessor for the OS and run the dwindling supply of legacy code.

Sadly, a variety of marketing and technical constraints mean this utopia
won't happen. For one thing, the increased transistor count of such a beast
would increase its cost to the point where people would have to be willing
to pay a lot for future-proofing. AMD proved you don't have to.

AMD proved you can have both 64-bit and 32-bit support for very little
extra cost. Intel was just out-engineered, mainly because they didn't think
people would want 64-bits on the desktop, even if they could have it for
nearly nothing.

DS

 

[Yousuf Khan]

Errr, yes and no.

As more and more systems were approaching 2gb or more physical memory, the
4gb address space limit of the x86 architecture was getting tight. PAE
works, but it's creaky...

Wasn't talking about the address limitations, was simply talking about the
instruction set itself, as well as its registers and stuff.

Yousuf Khan

 

[RusH]

(e-mail address removed) wrote in

[cut the crap]

So does this mean I will get more FPS on IA64 ?
x86-64 BAD architecture, BAAAAD architecture, SPANK SPANK !

Pozdrawiam.

 

[George Macdonald]

IA64 core inside desktop P4 CPUs would be the best bet for Intel to go 64bit.
x86-64 from Intel is the worst thing they could have done, ever. Not only Intel will look weak to
follow the AMD proposed 64bit pseudo/hybrid rule at extending the ancient x86 architecture but it
would mean that IA64 might become a dead project not because its architecture its worse but because
customers are quite retarded to promote an x86-64 hybrid instead.
If Intel followed customers wishes with P4 then it should have reverted back to Pentium III
architecture since the first P4 releases had lower IPC and overall performance.... and that would
have been a huge mistake for Intel which would have let AMD to conquer the market.

Yup - they made a mistake... but it was in trying to make a DSP type ISA
into a general purpose computer ISA.

If and only if Intel will reveal an hidden IA64 core inside IA32 sooner than expected (there were
unofficial roadmaps floating on the 'net quite some months ago showing IA64 going to desktop by
2006/2007 with dual-core CPUs ....so an hybrid IA32/IA64 architecture was probably on schedule long
time ago at Intel to let the desktop market move to 64bit...) it will have done the right thing,
otherwise it will be a big mistake in the next few years, unless Intel plans to trash x86-64 hybrid
thing anyway to move to IA64 but if it's going to be viceversa then x86-64 will be the worst thing
Intel could have done to itself.

Uhh x86-64 *is* new - it transforms a microprocessor ISA into a real
computer with enough registers to do real computing.:-[]

The IA64 EPIC architecture it's much better than any hybrid x86-64 thing, it will be in the long run
and it's time to stop the x86 extensions, it still is an old architecture, something new was needed
and it's still needed and IA64 is still the answer, even if it's taking too much time for
engineers/designers/coders to debug and improve it, x86 won't last forever and it's already not such
an efficient architecture even though heavily used.

What's so new about VLIW?

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[geno_cyber]

Yup - they made a mistake... but it was in trying to make a DSP type ISA
into a general purpose computer ISA.

What do you mean by DSP type ISA ?
The only DSP like parts in x86 CPUs are MMX,SSE,SSE2,SSE3 and 3DNow! , the rest is still general
purpose and highly patched x86 architecture.

 

[Yousuf Khan]

What do you mean by DSP type ISA ?
The only DSP like parts in x86 CPUs are MMX,SSE,SSE2,SSE3 and 3DNow! , the rest is still general
purpose and highly patched x86 architecture.

He's talking about the IA64, not x86.

Yousuf Khan

 

[Dale Pontius]

If x86 was such a great architecture, which is not, then Alpha CPUs would have been x86 based
instead what's happening nowadays is that AMD pushed Alpha CPU technologies inside the ancient x86
architecture that was originally designed without any supercomputer stuff in mind, Intel was trying
to do the same on its own but then bought the rest of the Alpha staff to push EPIC IA64 new
architecture.
The fact that Intel invented x86 in first place and they decided that was time to kill it without
continuing to extend it should be more than enough to know that continuing injecting Alpha CPU
architecture inside such a limited and far from really efficient although inexpensive technology in
order to make it work like it was never meant to , simply won't last forever.
Instead AMD , which surely has less money than Intel to spend on R&D and so couldn't afford
designing and entire new architecture to go 64bit, decided to push the x86-64 hybrid extension and
the nonsense market it's pushing Intel managers to kill IA64 just like Alpha was killed by x86
invasion, which is pretty silly to say the least... Intel killed and bought the majority of the
Alpha staff and market thanks to x86 but then its plans were to use Alpha technology and expertise
to design a whole new architecture, the IA64 and stop x86. Instead AMD had to go cheap and decided
to propose an x86-64 extension on its own... and IA64 , which would have been the real revisioned
and upgraded Alpha architecture, is going to die thanks to x86-64 from AMD because Intel managers
feel that they must follow this route to please customers now , despite the fact that thanks to this
route the real PC efficiency and architecture will be far worse than what IA64 would be able to
deliver in a few years from now and surely worse than if Alpha CPUs would still exist in the market.

There's nothing to be happy or satisfied about this big push of the x86-64 if it means the death of
IA64 and no IA64 desktop CPUs, just like x86 was worse compared to Alpha this x86-64 further
extension is worse than IA64.

If people think that x86 it's still a clean design .. oh well, just think about the fact that has
been patched thousand times in its 30 years of life since the first 8bit and 16bit releases....

I'll take you on a different front.

I'll agree that X86 is a cluttered mess. But that STILL doesn't mean
that IA64 is better. IMHO IA64 was something like an academic that
rebadged VLIW as EPIC and sold it to management. At any particular
time, the *newest* chip out there will tend to have the best benchmarks
as IA64 does, now. But for BILLIONS in investment and a decade in
development, they didn't leapfrog squat, just got to the 'normal'
leading position. Non cost-effective.

I look at X86-64 in somewhat the same vein as Kidder's "The Soul of
the New Machine," where they managed to clean some dirt out of an
old architecture as they designed a superset. There's dirt that just
can't be taken out, but improvements can be made. (and were)

Dale Pontius

 

[Tony Hill]

On Fri, 30 Jan 2004 22:06:03 GMT, (e-mail address removed) wrote:

If people think that x86 it's still a clean design .. oh well, just think about the fact that has
been patched thousand times in its 30 years of life since the first 8bit and 16bit releases....

And the problem with x86 is WHAT exactly? Just what is it about x86
that you think is bad and that isn't either already fixed or in the
process of being fixed.

Unless you can answer that simple question, all you're other arguments
are rather meaningless.

 

[Mike Tomlinson]

If x86 was such a great architecture, which is not,

then Alpha CPUs would have
been x86 based
instead what's happening nowadays is that AMD pushed Alpha CPU technologies
inside the ancient x86
architecture

IIRC, AMD used the Alpha EV6 bus, but not much else.

that was originally designed without any supercomputer stuff in
mind, Intel was trying
to do the same on its own but then bought the rest of the Alpha staff to push
EPIC IA64 new
architecture.

Funny that, I was always under the impression Intel bought Alpha to kill
it. I'm sure that Compaq's sale of Alpha to Intel came far too late to
integrate any of the Alpha technology into Itanic, anyway.

 

[Robert Myers]

On Fri, 30 Jan 2004 22:06:03 GMT, (e-mail address removed) wrote:



And the problem with x86 is WHAT exactly? Just what is it about x86
that you think is bad and that isn't either already fixed or in the
process of being fixed.

http://denali.cs.washington.edu/pubs/distpubs/slides/visits/case_for_denali.pdf

slide 14.

Now, maybe I could search further and find out about how it is being
fixed, but since I have brought a fact to a discussion that is
otherwise pretty much devoid of facts, I think I've done my part.

RM

 

[Yousuf Khan]

Funny that, I was always under the impression Intel bought Alpha to kill
it. I'm sure that Compaq's sale of Alpha to Intel came far too late to
integrate any of the Alpha technology into Itanic, anyway.

I wonder if the Alpha folks that Intel now employs are already being
assigned to the CT crew, because it doesn't look like they'll have much left
to do with Itanium after this? That would mean that the entire Alpha crew
would've been responsible for upgrading x86, both from the AMD side and the
Intel side.

Yousuf Khan

 

[daytripper]

I wonder if the Alpha folks that Intel now employs are already being
assigned to the CT crew, because it doesn't look like they'll have much left
to do with Itanium after this? That would mean that the entire Alpha crew
would've been responsible for upgrading x86, both from the AMD side and the
Intel side.

I wouldn't write off the EV gang quite yet. More than one rumour has them
designing dual core ia64 parts...

 

[Robert Myers]

I'll take you on a different front.

I'll agree that X86 is a cluttered mess. But that STILL doesn't mean
that IA64 is better. IMHO IA64 was something like an academic that
rebadged VLIW as EPIC and sold it to management. At any particular
time, the *newest* chip out there will tend to have the best benchmarks
as IA64 does, now. But for BILLIONS in investment and a decade in
development, they didn't leapfrog squat, just got to the 'normal'
leading position. Non cost-effective.

Easy for you to say. If anyone accurately foresaw the importance of
OoO and just exactly _why_ it would be so important _before_ it was
introduced into common usage, I should be very much indebted to anyone
who can direct me to an appropriate link (who knows, maybe such a link
exists). Run-time scheduling may or may not prove in the long run to
play the critical role that it currently does, so I'm not going to
make any emphatic statements that purport to be true for all time.
People have tried every conceivable scheme for scheduling, and right
now, on-die runtime scheduling appears to be the winner.

IA64 is saddled with an instruction set that makes OoO very hard, but
not impossible. OoO also makes nonsense of the premise of the IA64
ISA, which is that all the scheduling was to be preprogrammed, using
predicated instructions to make whatever run-time adjustments were
necessary. The scheme works _much_ better than people give it credit
for. The problem is that you only need a cache miss rate of less than
one percent to produce a factor of two slowdown in code execution
given the current mismatch between processor speed and memory latency.
Only the slightest miscalculation can bring you to ruin, and that's
why IA64 needs such a gigantic cache to perform decently.

Intel designed IA64 so that it would be very hard to clone. That, and
making sure that it could not be construed as subject to any of their
cross-licensing agreements, not performance, was their primary design
goal. As it stands _at_the_moment_, Intel seems to have succeeded
beyond its wildest expectations in those respects. It also happens to
have produced a world-beating processor for certain applications. It
can't be cloned, it isn't subject to cross-licensing agreements, and
it can be virtualized.

What makes you think you're so smart?

RM