[Info-vax] Poulson info from Dave Cantor

[Michael Kraemer]

JF Mezei schrieb:

> The way I remember it, the argument was made that it was much faster and
> cheaper to put the smarts in software (the compilers) instead of
> hardware (the chip) and that EPIC would be able to grow at a faster pace
> and never be caught up by competitors.

This still would only work if the basic assumption is correct,
i.e. vast speed gains by exploitation of inherent parallelism.
But simple code inspection of average programs will
reveal, that the amount of work eligible for parallel execution
is only modest, with the possible exception of dedicated
benchmark codes like LinPack.
And this modest amount of parallelism can be caught already
by traditional RISC chips with little extra cost.


> 
> Perhaps after the 8086 showed its ability to jump over the brick wall,
> Intel lost interest since IA64 would no longer replace the 8086 as
> industry standard.
> 

Me thinks it was AMD-64 which torpedoed the Itanic.
HP/Intel hoped that only their product would go 64
and replace PA, x86 and all others.
AMD showed, that this is already possible in an
evolutionary way within x86 itself. No need for
a completely new product.