From: Keith R. Williams
Subject: Re: ISA IRQ signal active how long?
Date: Mon, 7 Oct 2002 23:16:18 -0400
X-Newsreader: MicroPlanet Gravity v2.60
In article ,
> "Keith R. Williams" wrote in message
> > In article ,
> > firstname.lastname@example.org says...
> > > Eh. The PC wasn't introduced until 1981. Late 1981, for that
> > > matter.
> > August is "late" in the year?
> Well, it's later than February.
But not even Q4. That's *not* late.
> I wasn't sure of the exact date - but that it was after August 1,
> 1981 (since that was the day that the computer store I worked at opened up).
No, it was about 8/1 (though the search engines are being a pain
tonight). They weren't available for a few months, though.
> > > And that was with a (for IBM!) very short lead time - 6 months,
> > > IIRC.
> > Wrong.
> Possibly. Again, I blame defective ORAM (Organic RAM) for that.
> But it WAS a short lead time, for IBM.
Sure, but it wasn't designed by "IBM". It was an "Independant
Business Unit" and didn't have the onerous controls that IBM
engineers had. There is something to be said on both sides of
this one, BTW. I worked in a group trying to reign this stuff
back in, but that's another issue.
> > Eh. The PIC was in use way before the PC, in the same stupid way.
> > Eh. I had been working for IBM for, umm, seven years at the time.
> > I had some small bit of insider information about the PC some
> > time before 8/81. Several friends disappeared to Boca a couple
> > of years earlier.
> > The mode the ISA used was simply stupid (interrupts aren't the
> > only stupid transgression of the ISA bus).
> On this, you sing to the choir, my friend. To misquote one of my
> favorite books, "Crap! I could swallow a bunch of TTL logic and PUKE a
> better design!"
I've said much the same, many, many times. I was overjoyed when
I first read the MCA spec (gawd was EISA a nightmare). Less so
with PCI, but that's all water under the damned dam.
I was the lone wolf on .chips advocating the demise of the ISA
bus five years ago when it was shot through the heart by PC'97.
> > When doing my mainframe thing (years before the PC), I
> > architected a similar, but simpler bus that was used for power
> > controls in mainframes. I had the exact same issues to worry
> > about so designed my interrupts to be level sensitive (the
> > priority was far more complicated though). I went from there to
> > a group doing Intel chipsets.
> > I repeat: What a hunk-o-junk! I'll repeat: the ISA designers had
> > no understanding of TTL. They were ECL designers, and it shows.
> Eh. Now it's easier to curse them when I run into the (still
> occasional) IRQ contention.
Much of that is not ISA's problem.
> Had that today - but it was a software based problem. Seems like
> Win2K wouldn't admit that a SuperIO could share IRQ 4 between the two serial
> ports ... when IRQ3 was used for the ISA modem! Snarl!!!
ZOT! This is a stupid M$ problem. There is no reason the
hardware can't get this right, on a PCI based system anyway. M$
has never been accused of doing things right.