compare between targets

[drdozer]

Is it possible to run criterion with a range of targets, and then get it to compare the performances? I need to collect benchmarks for:

• cargo bench
• RUSTFLAGS="-C target-cpu=native" cargo bench
• O3 vs O2
• possibly other things

I think all the raw data is being collected when I run with -- --save-baseline default, for example, to put each independent run into a different folder. But while I see lots of json files, I can't see any command in either criterion or bench to generate reports comparing them. I would ideally like to avoid pulling this all into a python session.

I'm happy to use different command-line combinations. Or even some other cargo tool. I'd just rather avoid this becoming a python workbooks horror show.