On Thu, Nov 09, 2000 at 06:17:47AM +0000, Paul Jakma wrote:
> On Wed, 8 Nov 2000, Jay Estabrook wrote:
>
> >
> > Richard Henderson completed the work by doling out interrupts to CPUs
> > on a sort-of static round-robin allocation at register_interrupt()
>
> how is IRQ distribution done on i386 SMP? iirc it's a lot more
> dynamic on it, (irq's are routed to cpu randomly?)
That's the way the IO-APIC hardware works, IIRC; only one CPU gets it,
and the APIC tries to use some heuristic to determine which one. I'm
not really sure I want the chip designers to decide something like
that for *me*... ;-}
> what's holding the alpha port back from a similar dynamic IRQ
> scheme? i imagine alpha PIC must be at least as sophisticated as i386
> APIC if not more.. :)
Our hardware doesn't work that way, and I'm not sure we'd want the
"sophistication"; ask some of the folks who've had to make that code
work how hard it was... :-\
As I said, Andrea's first working attempts used a broadcast of the
interrupt to *all* CPUs, with the first one "getting there" doing the
processing. Unfortunately, on Alpha, there's a *lot* more overhead in
terms of delivering an interrupt, so those CPUs that do NOT get there
first take a real hit in terms of overhead (pipeline flushing, PAL mode
entry/exit, etc).
--Jay++
-----------------------------------------------------------------------------
Jay A Estabrook Alpha Engineering - LINUX Project
Compaq Computer Corp. - MRO1-2/K20 (508) 467-2080
200 Forest Street, Marlboro MA 01752 Jay.Estabrook@compaq.com
-----------------------------------------------------------------------------
_______________________________________________
Axp-list mailing list
Axp-list@redhat.com
https://listman.redhat.com/mailman/listinfo/axp-list
This archive was generated by hypermail version 2a22 on Fri Dec 1 08:00:07 2000 PST
Send any problems or questions about this archive to webmaster@alphalinux.org.