🚀 go-pugleaf

wbg wrote:
> 
> z@deltrak.demon.co.uk> <5qvvfa$9h2@news.nternet.net>
> <c1.01.2Gw8XV$17q@cretin.summerfield.network>:
> Organization: Hevanet - Portland's Internet Provider   Voice: 503-228-3520
> Distribution:
> 
> John Summerfield (summer@summerfield.network) wrote:
> 
> <WRT IRQs>
> : In any event, 16 (15 actually) is plenty: the AT (ISA) bus design is the
> 
> : John Summerfield
> 
> I must take exception to your assertion, Mr. Summerfield. If we both had
> the time, I could send you the details of at least half a dozen Wintel
> configurations that have passed through our shop recently where 15 IRQs
> were not enough. When you have Plug 'n Pray devices that have it all
> figured out which IRQ they *must* own, sound and video cards that seem to
> think they can't function without using up *two* IRQs, and compounding it
> all, The World's Most Widely Distributed Computer Virus (Win95)
> exercising its *own* perverted notions of which device should have what
> IRQ, then "plenty" can rapidly become a most ephemeral concept.

-- 

The problem goes back to the design of the original IBM PC, which used
unsharable edge triggered interupts.  If sharable low lever interupts
were used, more than one device could use an interupt, and the software
could sort out which device needed sharing.  I used to service Data
General mini computers, which had all of *ONE* interupt line, but there
was arbitration logic on the boards, so that the requesting device with
the highest got first crack at being serviced, then the next highest
etc.  With this one interupt, the computer could service dozens of
devices and hundreds of com lines.  On my first computer, an IMSAI 8080,
I had all my devices sharing one interupt, even though the computer
supported up to 8.

Due to the use of edge triggered interupts, and other bad design
decisions, I had the impression that someone on the PC design team, was
out to sabatoge the project.