This adds meta_name and bulk_size to the key used to group benchmarks.
Previously it incorrectly compared different benchmarks with each
other, leading to reports with high differences.
(E.g. comparing a benchmark with bulk_size=1000, with a benchmark with
bulk_size=1)
(There seems to be a separate issue in either cr8, or the script
saving the benchmark results into the benchmark table that causes
bulk_size to always be NULL)
This adds meta_name and bulk_size to the key used to group benchmarks. Previously it incorrectly compared different benchmarks with each other, leading to reports with high differences.
(E.g. comparing a benchmark with bulk_size=1000, with a benchmark with bulk_size=1)
(There seems to be a separate issue in either cr8, or the script saving the benchmark results into the benchmark table that causes
bulk_size
to always be NULL)