Paul G. Allen wrote:
Interesting. The benchmarks I saw at the time showed the EV8 at about
(IIRC) 4 times faster than the same speed (as in clock speed) PIII (the
PIII being the fastest Intel CPU at the time).
Do you really mean EV-8? EV-8 never reached silicon. So, if they were
throwing around benchmarks, that was simulation.
My 200 MHz 21264 system was faster than my PIII 800MHz system. (I still
have that system with Linux installed on it, but as for performance, it
was eclipsed by Intel and AMD systems years ago.) Yes, the price tag was
a bit higher (OK, more than a bit. ;) )
Really? My memory is fuzzy, but, IIRC, the problem at that point was
that "normal" business customers couldn't tell the difference between NT
on Alpha or x86 in that time frame. Computers had hit the "fast enough"
point for businesses.
And, oh, yes, I remember outfitting a 533Mhz 21164 (EV-5) system with
the "consumer-level" board in 1996. Egad, that was expensive.
Looking at cronology, 500MHz 21264 vs. 166MHz Pentium in 1996 is a very
different battle from a 300Mhz 21264 vs. 60MHz Pentium (P54C) (and Intel
really couldn't even ship many of those, 486's were still the primary
processor).
When it came to number
crunching, nothing beat an Alpha, which made them really nice for
graphics (which is what I worked on at DIGITAL - graphics workstations
using Alpha and PPC running UNIX and NT).
That is 100% true. Alpha crunched numbers like nobody's business. If
you had applications that could soak up the processor, an Alpha ruled.
This was doubly so because the memory subsystem on Alphas could actually
keep the beast *fed*.
Alphas absolutely ruled for CPU design for a *looong* time.
The 21264 systems sold far better than the 21164 systems, so I wouldn't
call the 21264 a failure.
Sold better but killed the company. I still maintain that punting out
EV-6 2 years earlier (and they could have) would have saved the company.
The problem was that the corporate culture created no engineering
incentive to do that. DEC is a good example that all engineering
thinking in management positions is just as damaging as all marketroid
thinking.
It's possible that a sell-off might have still occurred, but it would
have had a company in a very different revenue situation.
I still recall the saying "Alpha. The fastest processor nobody's ever
heard of."
Sad, but true.
It also didn't help that Ken Olsen screwed up getting Alpha into Macs in
1991. But that's a story for over a beer ...
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
