Thomas Boroske wrote:

> > > - Is it now, with reasonable experience and work put into it,
> > >   possible to get a working compiler toolchain ? With my current
> > >   install it wasn't possible to get a working (C++) setup
> > >   (that's 1.2G with some compiler patches, I think).
> >
> > Apart from the StrongARM bug, yes.

> I missed part of the "revision k" discussion though, so what's the
> problem in a nutshell ? Will there be a workaround eventually ?
> Point is, I doubt I'm prepared to buy another SA card just
> for RBSD.

It can't always be worked round. It's a different bug to the one
reported at boot---that's worked around already.


> Come on - no-one ever did a measured STM or LDM loop ? I can't
> believe it.

Nope.Apart from some simple tests doing memory-intensive C-code that was
a lot quicker than a RiscPC.

> Well, "normally" is a bit difficult. Of course, the PCI bus can do
> that but is the system also up to it ? I mean, it depends a bit
> on the PCI bridge you use, write buffering or not.

It has write buffering FIFOs, read-ahead, etc.

> Again, are you telling me no-one ever tried how many MBs/sec you
> can write to that PCI video card you ship with the system, and
> compared it with a PC ?

Nope. It just goes as quick as the video card will go, simple as that.

> So you've got all the drivers running ? Or is the system so compatible
> that you can simply use the general NetBSD drivers for PCI cards ?

Pretty much. Particularly the bus-DMA drivers, they just compiled
straight in and worked first time :-)

	N.