bench_support.H File Reference

Shared harness for the cellular-automata performance-gate anchors. More...

#include <algorithm>
#include <chrono>
#include <cstddef>
#include <cstdint>
#include <cstdio>
#include <cstdlib>
#include <fstream>
#include <iostream>
#include <string>
#include <utility>
#include <vector>

Include dependency graph for bench_support.H:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	Aleph::CA::Bench::Bench_Result
	Outcome of one measured anchor benchmark. More...

Namespaces
namespace	Aleph
	Main namespace for Aleph-w library functions.
namespace	Aleph::CA
namespace	Aleph::CA::Bench

Functions
std::string	Aleph::CA::Bench::compiler_string ()
	Describe the compiler that produced the running binary.
std::string	Aleph::CA::Bench::cpu_string ()
	Read the CPU model name advertised by the host.
std::size_t	Aleph::CA::Bench::resolve_repeats (const std::size_t fallback)
	Resolve the number of timed repetitions for one anchor.
template<typename Work >
Bench_Result	Aleph::CA::Bench::measure (std::string name, const double processed_cells, std::size_t repeats, Work &&work)
	Time a kernel with one warmup pass plus a median of timed passes.
void	Aleph::CA::Bench::emit_json (const Bench_Result &result, std::ostream &out=std::cout)
	Print one benchmark result as a single structured JSON object.

Detailed Description

Shared harness for the cellular-automata performance-gate anchors.

This header is local to benchmarks/: it is perf-gate tooling, not part of the installed Aleph API. Every anchor benchmark times its kernel with the same warmup-plus-median protocol and emits a single line of structured JSON so the perf_gate.py orchestrator can parse it deterministically:

{"name":"gol_1024","wall_ns":123456789,"cells_per_sec":1.234e8,
 "compiler":"gcc 13.2.0","cpu":"Intel(R) Core(TM) ..."}

The median of several timed repetitions (after one untimed warmup) absorbs the run-to-run jitter of shared CI runners, which is why the perf gate can key on a single number per benchmark.

Definition in file bench_support.H.