<Besides, aren't a whole bunch of people busy writing software even as we speak on enamalators? That way when Merced boxes ship in volume 2nd half 2000 a bunch of software is available with boxes at a great price? Won't happen?>
"You can imagine just how hard it is to get software developers to even budge on a new architecture. ... How tough do you think
it will be to get software developers to port their software onto real IA-64 hardware, much less IA-64 emulators?"
How tough? Unfortunately VERY tough. IA-64 introduces a new
paradigm and with it a new way of developing to take advantage of
the architecture. What's going to hurt the most is compilers will
be good but a simple recompile just won't hack it... It will
require quite a bit of IA-64 tutorials to get folks to make the
most of IA-64, to get the hoped for "speed up."
Speaking from experience... I took an existing VAX application with
several hundred thousand code lines, had to modify 3 lines in
one module... recompiled it and it ran quick as a bunny on Alpha.
<But one confusing thing... "low-cost doesn't make sense because software isn't available" .. what does that mean?>
"Low-cost means high volume. High volume makes no sense without a
large software base."
Why that's silly... look at your most recent MPR there... there is
a nice chart detailing *expected* costs of the entire Intel product
line a few years out... it pegs Merced selling for $2000 and McKinley
selling for $2000 and DeerTick selling for $1000.. tell me, if
low-cost equates to high volumes, why isn't Merced or McKinley
coming down on that chart? Can we expect Merced to drop to $1400
when McKinley is introed? We could spend a lot of time here but
both you and I know it has many determining factors, demand
, MANUFACTURING COST ;-), size (related to Man. cost ;-), etc.
Besides, if low-cost equates to high-volume, how is Samsung getting
away with their Alpha pricing, a loss leader?
"That's why I'm not particularly worried about low-cost Alpha boxes. Where's the
software? Or do you think corporations will buy these low-cost Alpha boxes just
to let their personnel run Netscape on their intranets?"
A myth on your part... the same market that a Merced workstation is
targetted at will be the same market Alpha has plenty of software
available for. In fact, when Merced ships there will probably be
many more Alpha applications available at the workstation level
for Alpha than there will be for Merced...
What software are you talking about anyhow? Seems Merced available
applications and Alpha available applications for the server
should be better on Alpha end... All major RDBMS, etc. what
are you thinking here?
As for workstation support:
compaq.com
many of those are available on Alpha too.. which brings me back
to Merced targets at FRS.. why would you purchase a $9000 Merced
workstation when better performance can be had for $3500 on
an Alpha?
Check out this PDF.. Pro-E magazine benchmark supplement.
ftp://ftp.compaq.com/pub/products/workstations/bench99-final.pdf
The XP1000 nails down several segments, as the price comes down
and performance goes up with the EV67 750 MHz uplift.. and EV68
to follow... that Pro-E XP1000 only had a 500 MHz CPU in it
of the older 21264.. the 21264a will do much better.
Rob |
