[LAU] RME Digi96/8 PAD pci pro

[Reuben Martin]

On Wednesday 25 March 2009 1:44:18 pm Giso Grimm wrote:
> Paul Davis wrote:
> > On Wed, Mar 25, 2009 at 12:34 PM, Grammostola Rosea
> > <rosea.grammostola at gmail.com <mailto:rosea.grammostola at gmail.com>> wrote:
> >
> >     Hi,
> >
> >     What do you guys think of the RME Digi96/8 PAD pci pro? How is the
> > card compared to an maudio delta 66  for example?
> >
> >
> > personally speaking, i would avoid this card. even though the RME
> > products that came after this are, in my opinion, peerless, this one
> > came before RME really understood how to do PCI bus-mastering. as a
> > result it is an inefficient h/w design that requires the device driver
> > to do a bunch of work that in the digi9652 and everything that followed
> > was done by the hardware. i am sure that the sound quality is good,
> > probably even a little better than the m-audio but certainly comparable
> > - its just not up the same internal design specs as RME's later products.
>
> And what is actually even more visible to the user are the strange
> buffer size constraints: Only the whole memory can be used, thus
> nperiods*fragmentsize*samplesize is fixed. Each of those parameters has
> only few possible values. For typical settings (32Bit/sample and ADAT
> mode) this results in fragment sizes of either 64 or 256 samples (thanks
> to the jack developers now at least nperiods can be decreased to values
> below the hardware nperiods, though). Synchronizing and hardware routing
> is much less flexible than with later RME products.
>
> Personally I would avoid that card (I bought an RME Digi96/8 a few years
> ago, everything said here holds for the whole Digi96/8 series).
>
> Giso

I've use one for quite a while now. I have no complaints, but it is an old 
design. The upside is that it works. The support for the newer hardware is 
sketchy. Just because a driver may "support" it doesn't mean there are any 
userspace tools for it. I built a box for a friend using the HDSP AES32 which 
is supported by the MADI driver. I was able to get it to work, but barely. The 
only software I could find for working with the matrix was a small CLI program 
intended to be a demo, (By a guy who hasn't touched the MADI driver in years) 
which for some reason got preserved in the SVN for pd of all places and just 
happened to work on the AES the same as the MADI. Not exactly the most 
intuitive interface for my friend who is totally clueless using a CLI shell. 
So just because the design may be kinda poor doesn't mean it's not useful.

-Reuben