Re: end of Intel?

On Sat, 22 Feb 2025 07:46:38 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:

On 2/21/2025 6:36 PM, bitrex wrote:

They could probably have a good bite of AMD and Nvidia's lunch if they tried
but they're only about 1% of the GPU market share right now, and seem to be
pretty unfocused about what direction they want to go in.

>
This has been Intel's problem from Day 1.  They enter a field,
make a contribution to the state of the art -- and then don't know
how to leverage that, going forward.
>
The 8085 et al. showed how they were unable to leverage past
success in future endeavors -- losing the market to Zilog
(who, subsequently, proved even more inept at leveraging past
success!).
>
"They don't know what BUSINESS they are in..."
>
And, things like the 432 were *decades* ahead of their time
the failure of which has left us with archaic processor designs
(at the chip level) and all of the hassles that come with them.

The 432 and HPPA were super-CISC architectures, lots of microcode, and
were slow. I think they were born of the core memory concept that main
memory is slow. Neitheher seemed to be designed by people who believed
Moore's Law, specifically that ram and cache would continue to get
denser and faster.

Intel had an ARM license and didn't use it. They should have done a
CPU with one x86 core and a dozen ARMs.

The new Pi Pico chip has both ARM and RiscV cpus. Same idea.

>
[NatSemi takes the cake for CPU flops -- owing solely to poor
commercialization, not technological innovation]
>
The 8086 probably was the decision that sealed their inevitable
fate; it tied them (and all of their resources) to a single
architecture that they then had to keep supporting in hindsight
(with each new innovation).  With MS eating up all of the
performance improvements that the hardware provided, they
never had a chance to make a break-out product.

I suspect their top management sabotaged anything that wasn't x86.

>
OTOH, you had folks like GI selling what amounted to "sequencers"
successfully -- albeit to a different (niche) market.  Note that
the 86010 didn't set the world on fire -- by a long shot.  (Thus
confirming past performance is not a valid predictor of future
success!)
>
I recall looking at the die in intel EPROMs and admiring how
tiny they were when contrasted with their japanese competitors.
OTOH, their competitors had realized that technical achievement
wasn't the key to market dominance -- just make the parts
*cheaper* than your competitors and worry about the "process"
catching up, later!
>
Zilog had the momentum, NatSemi the product.  Both failed to
meet their challenges.  (And poor TI, off in the wilderness
making all the wrong predictions about system performance)