Files
marqdo-cli/docs/performance.md
T
cflmy a3f3ecf9d0
CI / fmt + clippy + test (Linux x86_64) (push) Waiting to run
CI / schema instance validation + consumers (push) Waiting to run
CI / runtime contract (macos-latest) (push) Waiting to run
CI / runtime contract (windows-latest) (push) Waiting to run
Linux strict security lane (G-C) / strict isolation escape matrix (push) Waiting to run
ci/woodpecker/manual/performance Pipeline failed
ci/woodpecker/push/main Pipeline was successful
ci(perf): add the fixed Woodpecker G4 workflow
Move the push/PR pipeline to .woodpecker/main.yml so the repository can use
Woodpecker's multi-workflow directory. Add .woodpecker/performance.yml: manual
only, master only, on the single fixed Woodpecker agent (platform_id
woodpecker-agent-1). It builds one release binary, pins the first CPU in the
container cpuset, runs 5 warmups + 20 samples with --approved-node, and prints
the full marqdo.perf.v2 report plus a compact G4 summary for extraction.

No PR code is executed on the performance agent; the workflow is manual.
2026-11-21 19:51:09 +08:00

48 lines
2.3 KiB
Markdown

# Performance measurements (G4)
Status: the sampler and correctness corpus are implemented; **no release
performance baseline exists yet**. Do not cite a development-container or
TCG/QEMU number as a release gate.
## Three separate measurements
| Class | Goal | Host |
|---|---|---|
| Language core throughput | interpreter, expressions, memory/containers/stdlib algorithms | fixed G4 Linux node |
| CLI experience | cold `check`/`run`/`inspect`, parsing, stdlib loading | fixed G4 Linux node |
| Strict overhead | probe, re-exec, verification and strict run time vs pure | separate isolated performance node |
The G3 security VM can report correctness, not release performance: TCG time is
dominated by emulation and is not comparable with native or KVM.
## Tool
`tools/perf/runner.py` (schema `marqdo.perf.v2`) builds one release binary,
records its SHA256/Cargo.lock/rustc/source SHA, verifies every workload's exact
wire digest **before** timing, then reports warmups/samples, p50, p95, MAD,
CPU user/sys, peak RSS and the environment manifest. See
[`tools/perf/README.md`](../tools/perf/README.md).
Rules enforced by the tool:
* a wrong workload output is `CORRECTNESS_FAIL`; its timing is never reported;
* fewer than 5 warmups / 20 samples is `SMOKE`, not release evidence;
* `--approved-node` is required for `MEASURED`;
* a baseline from another `platform_id` or profile is refused;
* p50 +15% / RSS +20% are investigation thresholds, not automatic rollbacks.
## Current evidence
* Development-container smoke run (2026-11-21, `--warmups 1 --samples 3
--allow-smoke`): all nine workloads `VERIFIED`/`SMOKE`, no correctness
failures. Stored as a smoke artifact, **not** a release baseline.
* Remote Woodpecker G4: the repository now has a manual
[`.woodpecker/performance.yml`](../.woodpecker/performance.yml) workflow on
the designated fixed agent (`platform_id: woodpecker-agent-1`). It runs
5 warmups / 20 samples, pins the first CPU in the container's
`Cpus_allowed_list`, and records the hardware/toolchain manifest. Trigger it
with the Woodpecker API using `"workflow": "performance"`; do not run it on
PR code. Its report is the G4 baseline for this node.
* GitHub G4: `.github/workflows/performance-linux.yml` remains available for a
physically separate node.