Index of values

f %% t extends t by applying f within the resource bracket.

p1 ++ p2 the probe that measures both the counters in p1 and p2.

t1 ++ t2 is the right-biased union of t1 and t2.

add_json buf run writes a JSON text representing run to buf.

bench ?attr name f is a named benchmark measuring f.

Wired-in counters that are always collected.

counters p are the counters that p measures.

ctr ?unit ?desc name is a description of counter named name.

Empty set of attributes.

eval t runs t.

filter ?q run is run with only the benchmarks that match q.

Benchmark eliminator.

A probe for the GC subsystem, using Gc.counters.

A probe for the GC subsystem, using Gc.quick_stat.

group name benchmarks is a named group of benchmarks.

group_f ~init ~fini name f is a group of benchmarks that depend on a temporarily acquired resource.

is_empty t is true iff t is empty.

key pp ~name v is a set of attributes that contains only the mapping name -> v.

Log source for this module.

Benchmark entry point.

Entry point with a hook for command-line arguments.

matches ?group ~g name checks if name matches the query q in the sense above.

measure ~probe ~min_t ~min_s r performs a series of measurements of r.

nothing is the probe that measures nothing.

of_events evts is a probe that uses PAPI to count events evts.

of_json string is Ok (run, rest) when run is encoded as JSON text in some prefix of string, where rest is the remainder of string; or, Error e otherwise.

Ordinary Least Squares is the usual linear regression that minimizes average squared residuals (and maximizes R²).

pp ppf p pretty-prints p on ppf.

pp ppf t pretty-prints t on ppf.

pp_ctr ppf c pretty-prints c on ppf.

Pretty-prints e.

Default measurement probe.

probe ~counters f is a probe p that collects counters counters using the function f.

Reads the timestamp counter, using x86 RDTSC instruction.

run ?probe ?min_t ?min_s ?q ?note ~suite benchmarks runs the benchmark suite benchmarks named suite.

sample ~probe ~iters r performs a single measurement of r and returns a sample s.

Theil–Sen Estimator is an order-based linear regression.

Gives an interpretation of e's validity.

warmup ~seconds busy-loops the CPU to force it into full-power mode.