possible useful memory trick

Graham Stabler

This seemed quite interesting:

elm-chan.org/docs/avr/avrdma_e.html

Graham

Mike Huselton

I found this pseudo DMA technique intriguing. Maybe this can go in the Prop bag of tricks.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
JMH

deSilva

Assume we use I/O 8..23 as address register, and some other I/Os 0...7 as (masks!) MEMRD and DMAWR

   MOV rep, #blcksize
   MOV addr, startaddr
   SHL addr, #8
   OR addr, #<MEMRD
loop
   ANDN OUTA, addrRdWrMask  ' $FFFF03  e.g.
   OR OUTA, addr
   ADD addr, #256 
   OR OUTA, #|<DMAWR         ' give 100 ns to read memory
   DJNZ rep, #loop                 ' give 100 ns to send it

Will be 250 ns = 4 MByte/s

Ken Peterson

Wow....I have all kinds of lightbulbs going on in my head! Perhaps using an external SRAM for a graphics controller!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


The more I know, the more I know I don't know.· Is this what they call Wisdom?

hippy

It's a useful trick; I had a 32KB SRAM scheme using 2-bits to generate a video stream before the Propeller made that redundant.

If one wants to minimise I/O usage, RAM addressing can be done through an external counter, read via an 8-to-1 multiplexor and output via a 1-to-8 bit settable latch. That slows down access to RAM but the raw DMA can run just as quickly. With a bit of logic on the clocking line ( another counter and some gates ), it should be possible to blat the data across at very high speed by using the internal CTRA/CTRB. Even closer to DMA with the Propeller just providing the clock rate.