16 Jul
2004
16 Jul
'04
3:59 p.m.
On Fri, 2004-07-16 at 07:28, Mark Knecht wrote:
Florin Andrei wrote:
Well, the thing is, without APIC (default with Fedora single-CPU
kernels) there was some overlap in the way the devices were assigned
interrupts. Nothing bad (especially since the EMU10K1 got it's own IRQ),
but still i thought there is room for improvement.
So, since IO-APIC usually provides more interrupts, i thought, well, if
there are more IRQs available, the kernel might find a better way to
assign them.
The reality is quite the opposite. With IO-APIC the IRQs suffer from
more overlapping than without it. Which is kind of strange to me.
BTW, the mobo is based on the NForce v1 chipset.
--
Florin Andrei
http://florin.myip.org/
Without APIC, the nvidia module was alone on its own interrupt, the
EMU10K1 was alone, the ide and eth modules were on separate interrupts,
etc. Quite ok.
With IO-APIC, nvidia, EMU10K1 and bttv were on the same interrupt, ide2,
ide3 and eth0 were on the same interrupt.
Please note - The 'optimal' numerical assignment of interrupts with
using the APIC model has nothing to do with the older, non-APIC, order.
Please do not confuse the idea that 'interrupt #9 is best' with the
numbers assigned on an APIC system. These are completely different models.