Dan, List, See notes in-line (all the way to the bottom!)...
> From: "Dan Schwartz" <[EMAIL PROTECTED]> > Date: Fri, 4 Jan 2002 03:02:58 -0500 > > Dear Mark, > > The bus timing issue is simple: A 33 mHz Peripheral Connection > Interface (PCI) device will generate a single interrupt for each > event. However, to support legacy devices there's a PCI <-> ISA > bridge chip, to accomodate the parallel, RS-232 serial, keyboard & > mouse ports as well as the ISA boards. > > The problem is that the bridge chips don't handle ISA > interrupts too well: Basically, the 4:1 clock speed difference > causes each ISA interrupt to cause 4 interrupts on the PCI > bus. OUCH! Could you clarify this a little? Could you explain *why* it generates 4 interrupts? I *do* understand that 4 x 8 = 32, but this is the first time I have heard this story about 4 PCI interrupts for each ISA interrupt. I just can't see *why*? If I remember it correctly, ISA Interrupts are active *high*, *non* open collector, edge sensitive. About the worst combination they could have chosen when it comes to sharing interrupts! I thought PCI Interrupts (of which I know less of - I have gradually moved away from hardware) were better designed: open collector (?), active low (?), level sensitive (?). I can't see why the two can't be interfaced reasonably. > Cheers! > Dan > > Please CC any replies to me as I'm a DIGEST subscriber. Thanks! At your service ;-) > PS: I was part of a team that assembled an Altair 8800, S/N 007, in > December 1975 We are evidently from the same generation! On this (eastern) side of the Atlantic there was a less money for systems and the systems were more expensive in those days. In consequent the "heavier" (S-100 etc) systems didn't get such a foot on the ground as on your side. We read the articles in the magazines and looked at the pictures but had to work with the simpler SBC's and development boards. I have done years of "serious" work on "8 bitters" like the KIM (6502)! > >From: Mark Plowman [mailto:[EMAIL PROTECTED]] > >Subject: Re: [Leaf-user] loading PCMCIA modules; PCI vs ISA > > > >Dan, List, > > > >> From: "Dan Schwartz" <[EMAIL PROTECTED]> > >> Date: Wed, 26 Dec 2001 08:53:25 -0500 > >> > ><snip> > >> > >> As for using an ISA-> PCM/CIA card - Or in fact any ISA card > > ^^^^^^^^^^^^ > >> like a NIC - it's generally Not A Good Idea. As it turns out, the ISA > >> bus runs at 8.3 mHz while the PCI to ISA bridge runs at 33 mHz. This > >> means that every ISA interrupt generates *four* PCI interrupts. Yuk. > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > > >Would you care to explain/elaborate this? I can't make head nor tail > >of it. I am a poor soul with a mere twenty years in the > >hardware/software business and I haven't met this one before... ;-) > > > > > >> But, it gets worse: The NIC has to respond to all packets, > >> generating interrupts whether the packet is destined for that machine > >> or not. > > > >And here I was that thinking that software told hardware which MAC > >addresses it had to respond to... ;-) This one seems to have slipped your attention. Care to comment on it? > >> Cheers! > >> Dan > > > >Seasons greetings > > > >Mark New Years greetings Mark _______________________________________________ Leaf-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/leaf-user
