-
Add support single file csv output (combine all df's from df_dict)
-
add loop delta analysis between events
-
Add some optimizations to reduce loading time
- maybe use python's csv module? pandas maybe an overkill
-
perform ANOVA to observe change significance.
-
Optimize loop stats calculation
-
Add support to filter by pid
-
add config file to support flexible pattern suppliment to the code
-
filter pattern at the time of grep'g and sed'g the raw data, to reduce load time in csv produced later
-
Better use of reason part of
perf scriptoutput to provide detailed statistics -
Implement submission, completion latencies from conf file pattern
-
Separate all latencies in different files.
-
Get visualizations
-
Better UI
-
in bin/perf_script_processor, use dirname instead of complex stripping method (duh)