Sean McBride
31226d5bdd
|
3 years ago | |
|---|---|---|
| .. | ||
| .gitignore | 3 years ago | |
| README.md | 3 years ago | |
| edf_nopreemption.env | 3 years ago | |
| edf_preemption.env | 3 years ago | |
| fifo_nopreemption.env | 3 years ago | |
| fifo_preemption.env | 3 years ago | |
| initial_state.dat | 3 years ago | |
| one_iteration.dat | 3 years ago | |
| run.sh | 3 years ago | |
| spec.json | 3 years ago | |
| three_iterations.dat | 3 years ago | |
| two_iterations.dat | 3 years ago | |
README.md
EKF
Executes TinyEKF as shows by You Chong's GPS example
This test executes multiple runs of three iterations (the output of a previous iteration is refed as input), comparing the binary result against a known memoized result stored in initial_state.dat, one_iteration.dat, two_iterations.dat, and three_iterations.dat.
The rust.sh script stores per-run results to temporary files suffixed with *.res.dat and diff -s because I had trouble using cURL to pass a bash string of binary input.
In order to be compatible with the stdin/stdout model of serverless, the input and output files are binary concatenations of various C structs.
See main() in applications/TinyEKF/extras/c/gps_ekf_fn.c for specifics.
Useful parsing of the log
What is the average execution time of the first iteration? cat log.csv | grep ekf_first_iter | awk -F , '{ sum+=$5 } END{ print sum/NR }'