> But isn't this true for most chips?
> Here's a question for which I truly do not know the answer:
>
> Are there many applications which HAVE been compiled specifically
> for the P-III, or even for the P-II ?
Dozens/hundreds. Mostly in scientific, financial, audio/visual
processing, and military applications. Rarely in something you'll find
at EB.
> Were I a vendor in that market, I most likely would compile only
> for the least common denominator (i.e., basic Pentium) rather than
> compile either for EACH Pentium variant (thus confusing customers),
> or for the LATEST variant (thus eliminating customers).
Fortunately, there's no need to either confuse or eliminate customers.
The CPU-sensitive portions can be done either by having multiple code
paths within the body of the app, with the active path determined by a
"what CPU am I running on?" flag that is set at startup, or by
segregating CPU-specific code into loadable modules, and loading the
appropriate one based on the CPU-type check. The user doesn't need to be
aware of any of this. That said, I vaguely recall seeing some high-end
(multi-kilobuck) visual-modeling/simulation package a few years back
that had an exposed-to-the-user add-on or replacement module that you
could use if you were running on a Pentium Pro instead of a regular
Pentium. So, while it may not be necessary to confuse the customer, I
guess some people choose that option anyway ;-)
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|