How to run:
- Make sure you have a recent
nasm; check that
$ git clone https://github.com/davidad/asm_concurrency.git $ cd asm_concurrency $ make
- See results.
- If you want to verify deterministic behavior, run
You can also supply a command-line argument, which will be opened as a binary file.
$ make concurrency-noprint-x3 concurrency-noprint-x4 $ ./concurrency-noprint-x4 foo $ ./concurrency-noprint-x3 bar $ diff foo bar $ # same results, yay!
The number after the x indicates how many processes will be forked by that executable. You can even run more than one instance of "concurrency" concurrently (shocking!).
$ make clean $ make size=10000000 concurrency-noprint-x3 $ ./concurrency-noprint-x3 quux & # this will take a little while $ ./concurrency-noprint-x3 quux $ ./concurrency-noprint-x3 quuux & $ ./concurrency-noprint-x3 quuux & $ ./concurrency-noprint-x3 quuux $ diff quux quuux $ # hoo-ah!
make options you can tweak (with their default values):
$ make size=1000000 "nprocs=1 2 3 4 5 6 7 8 9 10 11 12 13 14 15"
- The output file will have size
(size+1)*8. (This is due to keeping a bookkeeping quad-word in the file and assuming for simplicity that the file is an integer number of quad-words.)
nprocsspecifies the set of executables to build and test (
-x2, etc.). This should have just been a command-line argument, but eh, this way works.