boblq wrote:
400 cpus or even 10 times that seems like a drop in the bucket
when compared with consumer markets for millions of units. Does
EDA really matter to Intel? Why?
Yes, because while the volume may be small it has multiple effects:
1) The chips that the EDA folks are using are *very* high margin as they
normally choose the highest performing chips. A license for EDA
software can be $100,000+. Making that 10% more effective by spending
$5,000 on a chip rather than $1000 is a reasonable economic choice.
Intel *loves* those kind of businesses.
2) 400 hosts is on the border of being a medium EDA shop. A smart RF
startup will be about 50 hosts at about 20 engineers. A multiplier of
2-3 means that the market will bear millions of these chips, and they
are *much* higher margin than consumer.
3) The technical advances that go into the higher chips are what
eventually get pushed into the consumer chips. If you can't sell those,
you bear twice the devlopment cost.
4) Marketing
It's the same argument as video cards. While the $500+ nVidia
"Ultrajizzinmypants" and ATI "Superspoogeonmyscreen" cards have *very*
low volumes (probably less than 10,000), they are high margin, test out
technologies, and help with marketing.
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
