Intel Itanium

[Christopher Browne]

On Thu, Mar 24, 2011 at 3:58 PM, William Muriithi
<william.muriithi-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org> wrote:
> On 24 March 2011 10:54, Lennart Sorensen <lsorense-1wCw9BSqJbv44Nm34jS7GywD8/FfD2ys at public.gmane.org> wrote:
>> On Thu, Mar 24, 2011 at 05:08:36AM -0400, Walter Dnes wrote:
>>>   Here's an anonymous post from approximately 5 years ago with a cynical
>>> take on Itanium...
>>>
>>>
>>> Is Microsoft responsible for Intel's Itanic disaster?
>>> Authored by: Anonymous on Tuesday, February 21 2006 @ 04:12 PM EST
>>> There was no Itanic disaster.
>>>
>>> Remember, when Intel started the Itanium bluff, they hand zero presence
>>> in 64-bit computing and very limited presence in high-end computing.
>>> Leading 64-bit players were MIPS/SGI, Alpha/Dec, PA-RISC/HP, and yes,
>>> IBM and Sparc.
>>>
>>> In fear of Itanium, SGI spun-off MIPS and became a Wintel reseller.
>>> Compaq feared Itanium would quickly kill DEC's Alpha and gave it to
>>> Intel when they bought DEC. HP redirected their hardware resources away
>>> from PA-RISC to their intel/itanium partnership.
>>>
>>> Without laying out a single transistor, Itanium completely destroyed 3
>>> of the 5 64-bit competitors; and Intel went from being a PC desktop
>>> brand to a wall-street-recognised leader in high-end computing.
>>>
>>> Never was there a more successful bluff in business. I don't see how
>>> people can consider that a failure.
>>
>> If it wasn't for AMD making x86_64, intel would have been in a lot of
>> trouble now.  intel really did think the itanium would work and that
>> they could get rid of x86 (and hence amd and via and such) in the high
>> end server market where the nice profits are to be found.
>>
>> So yes they destroyed 3 of the 5 64bit platforms, but intel didn't
>> have one themselves, so what good was it for them if the itanium didn't
>> work out?
>
> Very well said.  I was about to mention the same thing. Itanium was
> not a bluff at all. They almost got left behind by AMD as they
> desperately tried to push Itanium.  Heck, they even ended up licensing
> AMD technology when Microsoft refused to support their own version of
> x86_64.  Plus, they burnt  through $ 5 billion if I recall correctly.
>
> I felt like if it was not for Linux, there would be no x86_64.
> Microsoft only supported x86_64  after AMD had a decent sales.  If it
> was not for Linux support, I am not sure AMD would have been able to
> push that project successfully.

Yep, you're onto something, for sure.

AMD got a lot of uptake on AMD64 from the Linux community, because
they were able to offer some mighty nice server configurations that
were deeply better than Intel's best 32 bit offerings, and yet not
terribly much more expensive.

Linux was a big win, there, because, what with the porting efforts to
make Linux play well on Alpha, PPC, SPARC, and MIPS, it was
comparatively quick and easy to add an extra 64 bit platform.

Java wasn't yet super-popular, so limited support from Sun wasn't a
big barrier.

A lot of web server applications could run perfectly well, benefiting
from the big memory space and comparatively large register set,
without needing any visible code changes.
After all, LAMP (Linux, Apache, Middleware, Postgres) apps have their
code written in ("one of the languages starting with a P") which don't
care what kind of CPU you're on.

The same wouldn't be remotely true for Windows(tm), where applications
are pretty aware of what CPU architecture they're on.  It's no insult
to Microsoft to say that it's a tougher challenge to port their OS to
a 64 bit architecture.

In any case, the ready availability of "Linux on AMD64" made the
bootstrap process *way* easier for AMD - they had a profitable market
(e.g. - servers have way better margins than desktops) almost
immediately.

It's disappointing that there haven't been more instances of such.  I
would have loved to see cheap-ish boxes running MIPS, ARM, PPC, or
SPARC.

Mind you, none of those included the "emulate IA-32" layer that was
the *other* half of what saved AMD64 from irrelevance - it ran quite
nicely as a "wicked fast IA-32" box, too.
-- 
http://linuxfinances.info/info/linuxdistributions.html
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists