From: Kai-Lun Hsu [kaimee@cs.washington.edu]
Sent: Tuesday, April 18, 2000 8:56 AM
To: Larry Arnstein
Cc: 'cse466@cs.washington.edu'
Subject: design meeting 4/17 monday (ignore the 1st mail)

RAM implementation vs Streaming issue:
RAM is not necessary easier, you still need to do serial port streaming
stuff.  So serial streaming is the way to go.
 
Goal for lab this week:
 
flow control: receive a byte at a time, and send a byte back for ack. 
the streaming interface won't send a byte til it get a byte back.
 
according to our calculation last week running at 9600 bauds is not a
issue.
 
 
two possible implementation for buffering: 
1. FIFO
2. Double Buffering
 
we chose the 1st, because we don't need to read the data byte more than
once, once the data is read it's thrown away, so double buffering is
not useful in our case.
 
Lab2 outline:
-play a song, put the song in ROM
-start by having a test play sequence from ROM
-play a stream from PC
-convert at least one MIDI song to our CSE466 format
 
our chip:
-wait for command
 	1. start (beginning of the stream)
        2. tempo packet
        3. music packets
        4. stop command 
 
1st byte after the start command byte looks something like this:
 -----------------
| TNE | 0 | 0 | 0 |
 -----------------
 
then it will be followed by the tempo byte.


coninue - 0
stop - 1
tempo - 2