These scripts render state transition sequences from snapshot traces. In the sequence plots, filled circles represent stable states that repeated 2 or more times. Unfilled circles represent unstable states that did not repeat.
Run install.sh
python3 ./tpprof.py <input_data>
- Parses the snapshot trace in input_data to generate intermediate results and plots in <NAME>.{cluster,subsequence,pdf}
- NAME is automatically set to the concatenation of 'tmp/' and the text between the last '/' and '.' in input_data, e.g.,
input_data = data/alexnet.raw
=>result_prefix = tmp/alexnet.*
- tpprof automatically uses existing intermediate results. If you need to regenerate clustering or sequencing, delete the intermediate files.
cd snapGrep
make
python3 updateData.py <input_data> <new_format_data>
./snapGrep <num_switches> <signatures> <pattern> <new_format_data>
- Outputs matches and score on command line.
- The data formar is different than the one data in folder. We have a simple script,
snapGrep/updateData.py
, to change format accordingly. - For detail, please refer to the readme file in
snapGrep
folder or section 6 of the paper.