Noel Chiappa wrote:
From: Jerome
H. Fine
both DEC and DSD needed a bounce buffer managed
by software
Love that term, "bounce buffer" (I wrote a whole package to support them in a
packet switch I did) - I'm officially adopting it, right now! :-)
Noel
Hey - anything that anyone writes is automatically copyrighted.
So first you need permission to use that! I will try and figure out
who the person was that first used that phrase so we can both use it.
I did not mention that the concept worked quite well with the DEC
RX02 and the DSD RX03 when a PDP-11/73 was used. But when
version 1.0 of that device driver was used with a PDP-11/23, the
transfer rate was painfully slow because the interleave gap was
not long enough relative to the time needed to bounce the buffer.
Since the DMA silo had already been emptied into the bounce buffer,
the solution was to immediately initiate the next READ into the silo
and then bounce the buffer (for a READ request, of course). That
allowed the READ of the next sector on the floppy media to be
performed by the controller while the CPU was performing a
transfer out of the bounce buffer into the user buffer one word
at a time. I don't need to test the timing on any slower CPU since,
as far as I know, none support an MMU which would be required
to use a Mapped RT-11 monitor.
I may have the exact details and terminology incorrect - it was about
20 years ago.
Jerome Fine
