Post #1,349
7/10/01 6:37:20 PM
|
Umm could you expand on that?
What does PCI, AGP substitute for the er eternal Interrupt? ESP?
(I think I understand 'interrupts' - the PDP-8 was described in first sentence as, an interrupt-driven minicomputer... so what am I missing?)
(Ignorant guess: PnP / Windoze makes their own up, outta hype + saved glitches from deceased tasks - stored in an .idx file (#237) for later usage. Each one has to be Registered, of course, so that no OS-interrupts may be used, or ones from an unRegistered second machine :-)
|
Post #1,361
7/10/01 8:39:36 PM
|
Brother, can you spare an IRQ?
I think he means that PCI cards use a different method of assigning IRQs than ISA cards. It depends on the BIOS really and the motherboard hardware. Some IRQs are assigned to certain slots, it seems, and if you move the card you can change the IRQ it uses.
| "I can see if I want anything done right around here, I'll have to do it myself!" | Moe Howard |
|
Post #1,374
7/10/01 10:13:14 PM
|
PCI's designed to share IRQs. BUT...
like most PC hardware, there are exceptions.
Here's a good page with info on IRQ sharing on the PCI bus, with an emphasis on problems NICs can have.
[link|http://personal.bellsouth.net/sdf/h/b/hburgiss/dsl/survival/irq.htm| [link|http://personal.bellsouth.net/sdf/h/b/hburgiss/dsl/survival/irq.htm|http://personal.bel...ival/irq.htm]]
Basically, IIRC, the PCI bus is more sophisticated than the old ISA bus in that it can discrimiate between two devices raising the same interrupt signal. Edge- versus level-triggering seems to raise a flag with me. Check Google for more details...
HTH.
Cheers,
Scott.
|
Post #1,382
7/10/01 10:43:36 PM
|
Good crib sheet..
Thanks. Saved.
Interesting re SBLive stealing an Irq for DOS use - about which Windoze is oblivious!
Also noted my E-machine has modem in 1st PCI slot adjacent to AGP! but - works like charm.
Good reminder too, to make sure I remembered to disable the secondary IDE and second floppy in BIOS in the new boxes.
Now as to edge VS level- triggering (??) - fer us scope types that rings a bell, but also a bit sophisticated for a toy box I'd a thunk: that is, for it to reliably discriminate / flip - for one trigger-level setting VS a slightly different criterion. At a 100 - 200 MHz bus rate, say !
(If you find that ref, be curious how they Do that.)
Thanx,
A.
|
Post #1,384
7/10/01 11:03:08 PM
|
Edge versus Level triggering of interrupts...
I misrememberd. ISA uses edge, PCI uses level.
[link|http://mailman.real-time.com/pipermail/linux-kernel/2001-March/023682.html|Here's] a bit I found about it. I'm no expert on this stuff. There should be better info out there - especially on Intel's site.
HTH.
Cheers,
Scott.
|
Post #1,390
7/11/01 1:05:13 AM
|
See what you mean. Heh! an analog! e- problem..
If my reading between lines is close enough:
What he's calling "edge" triggering is.. that sort of detector, commonly using a Schmitt Trigger circuit: a feedback device which enhances the risetime via.. yup pos feedback! to give a thus faster dV/dt as a trigger pulse. What he calls "level triggering" implies an intentionally 'slower" device which is not concerned with dV/dt so much as just "change state when V= x.yz V".
Guess I'll poke around a bit; hard to see how same line from an "Irq detector" under s/w control - could parse the two - but it seems that the Vectra didn't.. too well. How'd ya find That!?
(It's amazing the amount of stuff you can see with a real analog GHz scope on these boards - let alone decipher)
A.
|
Post #1,395
7/11/01 3:07:09 AM
|
IRQ Sharing In Action
Here's my IRQ layout, as reported by Windows 2000:
IRQ Number\tDevice
9\tMicrosoft ACPI-Compliant System
11\tNVIDIA RIVA TNT2/TNT2 Pro
11\tIntel 82371AB/EB PCI to USB Universal Host Controller
11\tVortex Multifunction PCI Parent
11\tAureal Vortex 8830 Audio (WDM)
11\t3Com EtherLink XL PCI TPC NIC (3C900B-TPC) #2
11\tMass Storage Controller
15\tSecondary IDE Channel
8\tSystem CMOS/real time clock
13\tNumeric data processor
6\tStandard floppy disk controller
4\tCommunications Port (COM1)
3\tCommunications Port (COM2)
1\tStandard 101/102-Key or Microsoft Natural PS/2 Keyboard
Note all that stuff going through IRQ 11. I'll also free up another IRQ when I gets me a PS/2 -> USB adapter for this here IBM keyboard. --
Peter
Shill For Hire
|
Post #1,663
7/13/01 12:16:19 AM
|
Re: IRQ Sharing is a two edged sword.
It's nice that a number of devices can share an interrupt - you can have more devices.
It's bad because when an interrupt occurs, you don't know who (i.e. which device) caused it or why. On an interrupt, additional interrupts are automatically disabled - interrupt handlers aren't themselves interrupted. The interrupt handlers of the associated device drivers for an IRQ are started one at a time, probably in the same sequence each time, and interrogate the associated device to see if it was the one that caused the interrupt. If not, the next device's interrupt handler is started. If so, the event is recorded, the interrupt condition in that device is reset, the next operation if any may be started, and interrupts re-enabled. If other devices had a pending interrupt the interrupt handling process (as above) would start again. Otherwise, whatever process was originally interrupted would resume.
The disadvantages of sharing interrupts are that 1) it takes longer to figure out what caused an interrupt and 2) that some devices, in effect, get preferential treatment. For good performance, it's important to arrange the sequence of device drivers so that the most frequently interrupting devices are serviced earlier.
Alex
|
Post #2,173
7/17/01 11:49:10 PM
|
Amiga also shared IRQs
The IRQ number was related to the priority of the interrupt. I recall that the serial port operated on 2 IRQs - the higher priority IRQ was for receiving, the lower one for sending.
Darrell Spice, Jr.
[link|http://home.houston.rr.com/spiceware/|SpiceWare] - We don't do Windows, it's too much of a chore
|
