Re: FYI: I2O Architecture

Matthew Jacob (
Mon, 18 Nov 96 22:05:52 PST

>I haven't seen anything on this mailing list about I2O, so I thought I'd
>drop a pointer.
>The I20 architecture is suppose to allow the distribution of I/O functions
>over multiple processors. A lot of major companies appear to be behind this.
>Unfortunately, to receive the spec. you need to sign a NDA (no idea how
>restricting it is).

And pony up 250$.

Serious technology dope smokers funded by their corporate handlers
only, please. :-)

The backgrounder page makes I2O look like quite a lot of similar
efforts that I've seen over the last N years- perhaps with some
better aspects than others. Such efforts have ranged from the
DDI/DKI to OBIOS (open BIOS specification) to CDDE (common
device driver environment) to (and this I kind of am tickled by)
someone saying that SunLABS has done a Java Class library for
writing device drivers.

What will be more interesting is what the producers of hardware
will go with- they're the ones who will probably drive the standards-
not the other way around. Having been part of efforts in the past to come
up with grand scheme solutions, I am somewhat doubtful of the ability
of hierarchical software models for abstracting hardware, no matter how
well designed and enticingly offered, to ultimately win.

Note, btw, the inbound/outbound message queues- I suspect that DEC
might have offered that one (it fits *very* nicely with the I/O hose
stuff they have for their high end machines).