What to submit
- demonstration of functionality and verification of part D during office hours
- hard copies of
- lab writeup
- .s55 file of part D
- Matlab plots of part E
- Submit in the drop box soft copies of
- .s55 file of part D
- .cmd file of part D
- Matlab verification .m file of part D
- uniform.m of part E
- mulaw.m of part E
- any other .m files you wrote
Goals
- Correct functionality of part D
- At most 230 cycles for part D will get you full credit for the "execution cycles" component of your grade.
- The grading will be mostly based on how well the write-up questions are answered and how efficiently the difference of two quantization schemes has been shown using the plots.
Tips
Part D Matlab verification
Use int16 to convert Matlab numbers into 16-bit signed integers.
The src address is different every iteration, so you need to re-fetch it every iteration. Use the two-breakpoint method: one at a linenum before the copy and one at a linenum after the copy.
There's no easy way to fetch the BUFSIZE value because it's not an actual address. Just hard-code the value in Matlab.
If you get an error when it loads the program, try increasing the pause period in the line between the build and the load (say, pause(10)).
Make sure you set the breakpoints at linenums that make sense.