riscy
[Top] [All Lists]

Re: Video subsystem...

To: mackinla@cs.curtin.edu.au (Patrick Mackinlay)
Subject: Re: Video subsystem...
From: jeremy@sw.oz.au (Jeremy Fitzhardinge)
Date: Mon, 28 Jun 1993 13:00:38 +1000 (EST)
Cc: riscy@pyramid.com
In-reply-to: <199306251447.AA26313@vincent.cs.curtin.edu.au> from "Patrick Mackinlay" at Jun 25, 93 10:47:52 pm
Organization: Softway Pty Ltd
Patrick Mackinlay bubbles:

Just a comment:  My preference for onboard video would be a simple frame
buffer using vram.  It would be nice if we could get the chip described
in the "the interesting dec paper" (the thesis was that smart video chips
are an anachonism, and are effectivly useless when tied to a fast risc
chip -- mainly because they are always a generation of speed behind
the cpu, and the cpu can saturate the memory bandwidth anyway).

In any case, it is much more important to have the video memory directly
and completely accessable by the CPU than have limited access and have a
"smart" chip.

> Any comments on this? Anyone have pricing info on the 34010?

The TMS34010 is *slow*.  It only has a 16 bit memory data bus, which has
a max transfer of about 6MB/s at 40MHz.  This is because it is a very
CISCy chip, with a not particularly fast memory interface (partly because
all addressess are bit addresses, and the hardware is more complex).
The 34020 is 32bit and probably much faster, but I think it is an
unnecessary complication.  It would make the video inaccessable from
the cpu (unless you want shared memory, ugh), making it much harder
to program.  Neil almost certainly has his own views on this :-).

Neither chip would require another rom on the board.

 

<Prev in Thread] Current Thread [Next in Thread>