- Updated autograders to compute and summarize all autograded scores
on the gradesheets.
- Fixed some typos in the writeup and the slides.
- In the writeup, updated references to problem numbers for iaddl
and leave to include the corresponding problem numbers for the
This is a major update of the Arch Lab that reflects the changes in Chapter 4 of CS:APP2e.
- Students can now use the conditional move instructions to avoid
the performance problems of conditional jumps.
- The benchmark test has a random selection of positive
vs. negative numbers. So, conditional moves are really the way to go.
- The correctness check is more robust, looking for things like overshooting the array bounds.
- Added a 1000-byte limit on the size of the object code for the ncopy function.
- Added Perl scripts in src/ptest that provide comprehensive
regression testing of the different Y86 simulators:
- Tests each individual instruction type.
- Tests all of the jump types under different conditions.
- Tests different pipeline control combinations.
- Tests many different hazard possibilities.
- Incorporated the regresssion tests in src/ptest into the autograding
of the student solutions for parts B and C.
- Fixed problem that caused some newer Linux distributions to
get tk.h from /usr/local/include instead of /usr/include.
Thanks to Prof. Dr. Gerd Doeben-Henisch, Fachhochschule Frankfurt.
- Fixed minor bug that caused some compilers to complain.
Changed type of return value from getopt() from char
Thanks to Morgan Harvey, Portland State University.