🚀 go-pugleaf

smckinty@sunicnc.France.Sun.COM (Steve McKinty - Sun ICNC) writes:
>
>Bear in mind that the 8530 in a SPARCstation is a character-interrupt
>device. Every received character causes a CPU interrupt and the processor
>must read the chip registers, store the value in memory, etc. Asking it
>to do that at >38.4 is asking a lot, especially for an SS1 or 1+. I have
>seen SS2s running at 64k, but you can expect that to be quite a load
>on the system. If the machine is heavily loaded you may lose characters.

Huh?  Since when did an interrupt every 13us or so become an unbearable 
load?  For well-designed and trim interrupt routines, this is nothing.  I've
designed special-purpose boxes for everything from scoreboards to large
power-grid controllers, and have routinely run uarts at speeds above 38.4K
with even the humble descendents of the 8251 and the 2681, being serviced
by such pokey old processors as Z-80s and 68000's.

The frequency of the interrupt, as long as it is an order of magnitude or more
slower than the clock speed of the processor, does not have to be a 
significant drain upon system resources.  The problem with Sun is that their
goofy interrupt-service routines and port device-drivers in general have
traditionally been extremely inefficient.  Back in the good ole' days we'd
run 11/780 ports under straight BSD4.2 at 19.2K without problems, but then
"progress" came along in the form of sysV port drivers and Streams and now
every 3rd administrator has a seizure at the very mention of anything over
9600 baud.  *sigh*.

~james