Path: utzoo!attcan!uunet!lll-winken!brutus.cs.uiuc.edu!wuarchive!udel!rochester!rutgers!bpa!cbmvax!daveh
From: daveh@cbmvax.commodore.com (Dave Haynie)
Newsgroups: comp.sys.amiga.hardware
Subject: Re: (B) 68040 vs. gfx coprocessor
Message-ID: <9662@cbmvax.commodore.com>
Date: 13 Feb 90 21:52:14 GMT
References: <633@xdos.UUCP> <3046@pur-phy> <9586@cbmvax.commodore.com> <1123@savax.UUCP>
Reply-To: daveh@cbmvax.cbm.commodore.com (Dave Haynie)
Organization: Commodore, West Chester, PA
Lines: 50

In article <1123@savax.UUCP> thompson@savax.UUCP (thompson mark) writes:
>In article <9586@cbmvax.commodore.com> daveh@cbmvax.cbm.commodore.com (Dave Haynie) writes:

>I believe you mean the TI 340 family....specifically, the 34010

Um, er, yeah, that one.  Home of the 34010 and 34020, etc.

>This is the TI 320 family (some very nice parts I might add).

There are very nice parts, already available in floating point.  And
they even seem to keep some lines pin compatible, a nice trick for
any manufacturer.

>>You wouldn't pick a DSP as a blitter replacement, most are probably
>>less well equipped at handling bit-aligned operations than some general
>>purpose CPU.  But if you did lots of image or sound processing on that nice
>>graphics display, a good DSP would certainly be something nice to have around.

>Especially for a 3D rendering accellerator. A TI 320C30 would do very well.
>However, I think such an application would work much better if the 320C30
>was an entirely separate processor (not a coprocessor) working out of its
>own memory which the Amiga could DMA display lists into.

All the DSP chips have their own memory, far as I know.  It's not an option,
it's generally required for any performance at all.  Some even have separate
buses, for separate I and D, or possibly two data paths (most DSP operations 
are of the form A op B, so it makes sense to fetch two data sets in parallel).

>On a slightly different note....I heard someone state that the ULowell
>board will probably never come to market because of some very expensive
>processors on the board. I think they might be confused with the ULowell
>image processing board for the Amiga. This board used several (I think 5)
>NEC pipelined image processor chips. These are relatively expensive and
>esoteric parts. The graphics board uses a TI 34010. These are plentiful
>and cheap (not to mention rather low performance).

That's very likely the case.  I played with the Dataflow board way back at
the Washington, D.C. DevCon.  It worked, doing DMA straight to Chip memory,
real-time edge detection of images, etc.  Pretty cool, but way expensive 
and hard to program, I gather.  The Video board would probably cost more
in the range of an A2620 type device -- nothing really blue-sky.  That's just 
a guess, but there certainly aren't any wacky parts on it.

>|      Mark Thompson                                                     |


-- 
Dave Haynie Commodore-Amiga (Systems Engineering) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
                    Too much of everything is just enough