processoranalyzer
Usage
script.py <input_file_name> <output_file_name> <test_name>
"test_name" refers to the value that will be attached to every cycle. This is intended so that you can distinguish individual "test" sets when analyzing the data in a larger database.
Behavior
Given a debug log from Chipyard, this code will use regular expressions and stream each cycle to the script to extract the value of individual fields (flags, addresses, etc.) into a Dictionary before writing results into a CSV.
The script outputs CSV as that is the default format of Apache Superset, which the team planned to use for analysis and storage purposes.
Known Issues and Possible Improvements
In addition to existing @TODOS in source code:
- Prepend arbitrary date in front of "cycle_count" or create an arbitrary "date" field so that platforms like Apache Superset can natively plot this field as a time axis.
- Skip the CSV format and simply output a JSON format. May not work with
- Work with the Scala interface of Chipyard. This could provide the advantage of feeding data directly into other platforms.