home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.amiga.hardware
- Path: news-1.csn.net!ub!dsinc!scala!news
- From: dave.haynie@scala.com (Dave Haynie)
- Subject: Re: A3000+ Questions
- Sender: news@scala.scala.com (Usenet administrator)
- Message-ID: <1996Mar7.232627.10991@scala.scala.com>
- Date: Thu, 7 Mar 1996 23:26:27 GMT
- Reply-To: dave.haynie@scala.com (Dave Haynie)
- References: <4hjtj1$au8@news.cityscape.co.uk>
- Nntp-Posting-Host: gator
- Organization: Scala Computer Television, US Research Center
-
- In <4hjtj1$au8@news.cityscape.co.uk>, bs05@cityscape.co.uk (NEW COLLEGE) writes:
- >At The amiga home page (http://ww.omnipresnece.com/amiga/) I was reading
- >herstory page, looking at all the stuff by david haynie. Apparently the
- >C= engineers had an amiga 3000 with aga and dsp integerated into the
- >OS that could emulate a modem etc.
-
- That's true. The DSP subsystem was never 100% functional, due to a bug
- in the DSP control logic added to the DMAC and a problem with bus
- retries I discovered in Gary. Both would have been revised, of course,
- had the A3000+ continued on to production.
-
- >There were even some working prototypes built,
-
- We had about 50 units made, most of which worked fine as AA
- machines. I hacked in some basic logic to work around the DMAC bugs,
- enough to get two DSP systems up and running, so the VCOS/VCAS port
- could be completed (that's the AT&T DSP operating kernel and system
- interface module).
-
- >my question is did AT inherit these designs and prototypes,
-
- They certainly have the designs. I think most of the prototypes
- remained with the developers who received them from Commodore. A small
- number of A3000+ systems were sold off during the West Chester
- auction, including all of the non-functional PCBs.
-
- >if so why dont they use this as the 1200+ or ehatever,
-
- The DSP subsystem was a reasonable thing to include in an A3000 class
- '040 machine (the A3000+ prototypes were '030-based, but we has
- planned to go to the '030 in a later revision -- AA and the DSP
- subsystem were the gating factors for software development). It's too
- expensive for an A1200-class machine, unfortunately.
-
- >the improved processor would be an immediate improvement and the dsp
- >would improve things no end.
-
- The DSP was a very nice idea at the time. Today, I think AT's efforts
- would be better spent working on PowerPC systems. One problem with the
- DSP subsystem is the fact you wind up with two separate OSs and two
- separate programming environments. They work pretty well together,
- sure, but you can't, for example, trade spare DSP time for spare CPU
- time. A better solution is to incorporate DSP-like instructions into
- your main CPU (many RISCs do at least a bit of this, including the
- PowerPC), and support multiprocessing in your main system OS. Then
- you're using standard hardware, rather than specialized hardware, for
- your DSP algorithms. As history keeps showing us every now and then,
- commodity parts have a tendency to improve much faster than specialize
- parts, at least when you're talking evolutionary
- improvements. Specialized parts tend to be revolutionary, but unless
- they're incorporated into the commodity architecture, they're soon
- outclassed by the regular stuff.
-
- >Have AT ever approached you since the buyout?,
-
- Wouldn't you like to know.
-
- >Did you ever write some sort of full history of your experiances at C=,
- >if so where can I get it?
-
- Nope, I haven't done anything like this. It would be a cool thing to
- see done, but I don't expect to have the time, anytime soon.
-
- Dave Haynie | ex-Commodore Engineering | for DiskSalv 3 &
- Sr. Systems Engineer | Hardwired Media Company | "The Deathbed Vigil"
- Scala Inc., US R&D | Ki No Kawa Aikido | info@iam.com
-
- "Feeling ... Pretty ... Psyched" -R.E.M.
-
-