Main measurement: iterations per second
WIP Grouping benchmarks manually
When running multiple runs over multiple benchmarks, randomize the order for each run
Bar-chart for comparing benchamrks to each other
Line-chart for comparing one benchmark across time
Bar-chart for comparing implementations
Port to clisp, cmucl, and ccl
CCL and CMUCL seem happy enough.
A command to save the current implementation under microbench/bin/ for future use, with just enough code to run the benchmarks loaded.