Adaptec 29320?

[James Van Artsdalen]

> Date: Tue, 3 Feb 2004 08:09:05 +0100
> From: Bill Squire <billsf at curacao.n2it.nl>
> 
> Keep hearing about "Win64". Those few I know who develop on that platform
> or maintain Windows servers say its quite a farce (remember NT on Alpha?)

Well, it is Windows.  What can I say?  32-bit Windows makes a good X11
server and it's handy to keep one around for clients who might be
scared if they saw vi.

I keep a matrix of four disks with 32-bit and 64-bit Windows and
FreeBSD for testing.  It's usually desirable to not report a bug if
it happens in all four environments.

> Maybe give it a try but i'd suggest a better SCSI host adapter.

I'm after a controller that works in all four test environments
and can directly address more than 4 GB.  I'd prefer serial ATA
but there are none yet.  SCSI is a last resort.

I'm willing to use a different controller.  But when two new PCI-X
cards fail like this I'm suspicious of the Tyan's AMI BIOS, which I
consider suspect.  That BIOS might be programming the PCI-X bridge
wrongly, hence my curiosity as to whether anyone has gotten the 29320,
or any other PCI-X card for that matter, to work in the K8W S2885.

> Highpoint apparently supplies FreeBSD
> drivers. <http://www.highpoint-tech.com/USA/brr1540.htm> I hear
> 3ware works well and Promise seems 'promising' for SATA RAID. (and
> quite cheap too) Anyone use the Promise SATA 150 TX4 ?

That HighPoint cannot directly address more than 4 GB of RAM and I
have 8 GB.  Likewise for Promise.  The 3ware driver may not support
3ware's sATA card and in any case that driver doesn't support more 4 GB
(requires the use of bound buffers).

HighPoint's Rocket Raid-1820 is PCI-X and probably can address the
entire memory space.  But this is supposedly based on the Marvel chip
and what drivers HighPoint has seem to be binary only, useless.
SuperMicro's PCI-X sATA supposedly uses the same chip, without the
boot ROM.