Hello,
> > If someone is just about to flash the QQ, would you please
> > try to capture lspci -xxx before and after the flash
> > on otherwise identical configuration and look
> > whether something has changed and where?
> It did change - a lot. But that could also be because
> I changed stuff in my configuration.
> ...
> So, I hope not to bother you by sending both configs with this mail
Well it would be much easier on the unchanged
config, but I tried nevertheless:
Seems you have played with secondary master drive.
The memory setting (RAS/CAS delay, RAS precharge)
were also flipped - did you do that manually,
or is that something that the new BIOS also tweaks?
I summarize the changes for the Intel stuff - I have
no docs on HPT or Matrox:
Host bridge
- aperture base address changed
- 0e00000 - 0efffff was disabled, are read only now
- RAS to CAS delay changed from 3 to 2 clocks
RAS precharge 3 -> 2
- multiple bit error -> single bit error at address 0
(there is a "ignore during init" note, so it probably is OK)
AGP bridge
- AGP I/O base default now (f000, was c000)
- mem base d400, was e400; mem limit d7ff, was e7ff
- prefetchable mem also changed
ISA
- lower BIOS (e0000) enabled
- PIRQ A routing changed (irq14 -> irq10)
- PIRQ D routing changed (irq10 -> irq9)
IDE 1
- secondary channel:
slave IDE timing enabled and set
prefetch and posting enabled now
iordy is sampled and not forced internally for drive 1
fast timing enabled for drive 1
- UDMA/33 enabled for secondary drive 0
- secondary drive 0 UDMA timing set
USB
- interrupt line (ignored on PIIX4) changed from 0a to ff
ACPI
- burst timer reload enabled for device monitor 0 and 1
The rest is the same with both BIOS-es.
I have no idea what the ACPI stuff means. The only interesting
change is the IRQ routing stuff - the original routing of
PIRQ A to irq 14 is really weird. So either this, or the HPT
or VGA changes did make the trick.
I don't see anything that is APIC related here, so if there
is less APIC errors with the QQ BIOS, the reason must
be somewhere else.
I am no Intel nor PCI guru - I only looked up the numbers.
Maybe some voodoo masters :-) can analyse this better.
Regards
--
Stano
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
