[Info-vax] "x86 has only a few years left in the market place"

[Scott Dorsey]

=?UTF-8?Q?Arne_Vajh=c3=b8j?=  <arne at vajhoej.dk> wrote:
>On 5/12/2018 9:15 PM, Scott Dorsey wrote:
>> seasoned_geek  <roland at logikalsolutions.com> wrote:
>>> On Saturday, May 12, 2018 at 11:27:18 AM UTC-5, Stephen Hoffman wrote:
>>>>
>>>> Nobody likes x86-64.  Not even the folks at Intel, judging by the
>>>> number of times they've tried to replace it.
>>>
>>> INTEL doesn't have the skills to replace it. That would be why they were caught red handed stealing Alpha tech.
>> 
>> Intel keeps coming out with better architectures... maybe not the iAPX 432,
>> but the 8096, i860 and i960 were definitely huge steps ahead of the x86
>> architecturally and, at the time they were introduced, in performance.  But
>> Intel couldn't sell them.
>
>And then a company called HP convinced them to go with a thing
>called Itanium (or what later got the name Itanium) and we
>know the end of that story now!

No, you can't blame HP for that, it was very much an idea that came 
from Intel in-house.  SGI was on the line long before HP was.

And really, the VLIW thing seemed like a good plan, but there had been so
few VLIW systems out there and nobody had done real simulations and ported
code to simulated systems.  By the time they got simulators running and
compilers and code it became very clear that there were severe cache issues
and the chip guys were having to do dramatic redesigns to throw cache at
all of the problems.

Intel took a gamble, it turned out not to be a good one, but they weren't
able to drop the thing in time.  In the case of the 432, they took a gamble,
it turned out to be a performance nightmare, and they dropped the thing
before shipping more than a few samples.  But by the time the McKinley
chip was around, the amount of money involved made it hard to do that.
--scott

-- 
"C'est un Nagra. C'est suisse, et tres, tres precis."