Re: Buildtime woes and other stuff
> On Thursday 24 March 2005 7:16am, Oliver Korpilla wrote:
> >
> > I hate this &$%?!@ board!!! (I've overlooked the fact that ATI made not
> > only the onboard graphics but the chipset as well!) Nothing works as
> > intended...
> >
> > If MSI or ATI would be so kind to give me some documentation so I could
> fix
> > this mess myself... But MSI does not feel responsible, and ATI failed to
> > react. Their fglrx driver for x86_64 even fails to compile - and I guess
> > this is what they meant with "This board supports Linux" ...
>
> ATI's dismal record with video drivers for Linux suggests to me that they
> wouldn't be much better when supporting their mobo chipset under Linux
> either. They're just fine if you're only going to run Windows on your
> machine, but I would stay away from them until their Linux reputation
> improves.
I've stopped using ATI graphics drivers - they are regularly out of date,
and they have no process in place for submitting patches. Most of what bugs
these drivers is that they are coded to now deprecated interfaces, which can
easily be fixed, but sending fixes over ATI customer support sucks - I
tried.
> I'm using an older MSI Neo board using VIA chipsets, and it works
> exceedingly
> well under Linux. Everything I have is recognized including SMBus, ACPI,
> and
> Ethernet. However, I don't use SATA, so I can't speak to that. When you
> replace your current motherboard I suggest looking at a VIA-based MSI
> board.
Which I cannot - it's the only MicroATX board I've found, and the chassis is
MicroATX as well. I would need to: Get a new chassis, a new power supply
(now with the chassis), a new motherboard and a separate graphics adapter,
since the other boards I took a look at didn't have integrated graphics.
uATX was among the requirements. But I would have resisted and not bought
the board, if I noticed the chipset was from ATI, too, knowing their FOSS
policies. I'm the only one running Linux, so I'm the one stuck, while the
others are fine running XP on the boxes I procured for them.
> The
> only downside is of course not having built-in graphics on the mobo, I've
> always preferred a separate video card, but YMMV.
Me, too. But was not for me to chose this time.
> Your timing/clock problems are however more widespread. I have related
> problems too with spurious messages about lost timer ticks. They all
> appear
> to stem from problems with the kernel's ACPI driver and/or the kernel not
> handling a changing CPU clock speed well (this happens to me precisely
> because I've got the Corecell activated on the mobo and its dynamically
> changing the CPU speed based on load). I've seen one conversation on lkml
> about it, and it appears to be a "work-in-progress" problem, that will
> eventually be solved within the kernel itself.
I would be perfectly fine with running with constant maximum speed, I want
no changing clock speeds. May it be that the clock remains in "fast mode",
while the CPU clock is reduced? This would explain the performance lags with
kernel compiling!!! FreeBSD surely doesn't use frequency scaling etc. by
default and would be immune to this, I guess, and this is why I've had no
problems in *BSD! How can I switch to "max performance"?
My Google and lkml searches turned up not much revealing, especially no
patches. Can you point me to threads or even fixes???
With kind regards,
Oliver Korpilla
--
Handyrechnung zu hoch? Tipp: SMS und MMS mit GMX
Seien Sie so frei: Alle Infos unter http://www.gmx.net/de/go/freesms
Reply to: