[sdiy] Re: PCSYNTH Web Page Up

[Magnus Danielson]

From: Helen Austin <gwenhwyfaer at gmail.com>
Subject: [sdiy] Re: PCSYNTH Web Page Up
Date: Wed, 8 Jun 2005 09:47:32 -0700
Message-ID: <df5b05ea05060809471f719a0b at mail.gmail.com>

> On 6/8/05, Scott Gravenhorst <music.maker at gte.net> wrote:
> > There is also an explanation as to why I designed an 8 bit memory cycle ISA
> > card instead of a 16 bit memory cycle ISA card.
> 
> Just a quick comment. You plan to decode the addresses so that
> $B8000 and $B8002 are the low and high byte of the DAC. Why not decode
> them so that $B8000 is the low byte and $B8001 is the high? That way
> you can use a single 16-bit MOV to write to the DAC; the ISA bus
> management should translate them automagically. Then the code
> becomes shorter, and you can get more of it into the CPU's cache.
> (It also becomes simpler and more direct to write.)
> 
> Helen (following this thread avidly!)

Good point. Since the introduction of the IBM AT and its 8 MHz 80286, the
IBM A bus (later named ISA) is a 16-bit bus and since then doing 16 bit
transfers makes sense. The instructions supports it from the 8086, but the
IBM PC bus was only 8 bits (and hence only an 8088 was used).

Remember, the x86 have least significant first so Helens address map is the one
to use.

I have been following this thread too, and I'd like to point out that DMA is
fairly simple to do, but with todays speeds it not as much of an issue. But
more importantly a little buffer for a few words is a good thing. Also, I got
myself a bunch of full-length lab-boards for about 15 USD each, so I got a pack
of 5! I gave away one to a friend in need. I wonder whatever happend to that
project. You guys have me wondered for the first time in years!

Cheers,
Magnus