NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/sys/atari/st/tech/4640

Internet Message Format | 1992-08-31 | 1.8 KB

Path: sparky!uunet!van-bc!jonh.wimsey.bc.ca!jhenders
From: [e-mail] ( John Henders )
Newsgroups: comp.sys.atari.st.tech
Message-ID: <[e-mail]>
Organization:  One of these days for sure
Subject: Re: DSP in the Falcon
References: <[e-mail]> <[e-mail]>
Reply-To: [e-mail]
X-Software: HERMES GUS 1.03 Rev. Apr 14 1992
Date: Mon, 31 Aug 1992 18:20:24 -0700
Lines: 28

In <[e-mail]>, Steven Eker writes:
>
>Things which the 56001 should be useful for include software sound synthesis
>& matrix*vector multiplcation (the basis of 3D vector games).
>
>If the DSP DMA is implemented right (does anyone have a programmer eye view of
>this yet?) is should be possible to use it as an extremely fast graphics
>processor for polygons/phong shading/texture mapping (I wonder if anyone
>will bother to write an "Ultima Underworld" type game for the Falcon?).
>I worry about this though - the DSP is 24 bits and the 68030 is 32 bits
>so there's plenty of room for Atari to screw up...
>
    For anyone who hasn't seen the latest AEO, the Falcon appears to
have been design as a 24 bit computer, partly to maintain compatability
with existing software. This doesn't mean third party boards etc can't
extent that, but it explains the 14 meg memory limit. The other captures
from Jim Allen posts to GEnie say that the Falcon chips all seem to talk
at 16mhz. Another poster mentioned here that there's an 8bit processor to
dsp channel, I guess for uploading code into the dsp, though I don't
know if that means you can't tell the dsp to load a program from a memory
location itself, which would be faster, I'd guess.
    Motorola says the top speed the 56001 is clocked at is 33mhz, so the
ad/da converters must be the parts clocked at 50mhz.


--
[e-mail]