mirror of
https://github.com/azaion/gps-denied-onboard.git
synced 2026-06-21 18:41:13 +00:00
Oleksandr Bezdieniezhnykh
65ad2168ed
AZ-300 mandatory simple-baseline InferenceRuntime (eager FP16 PyTorch).
Implements the AZ-297 Protocol; current_runtime_label returns
"pytorch_fp16". Numerical reference every fancier C7 strategy (AZ-298
TRT, AZ-299 ORT) is measured against, and the only viable runtime for
Tier-1 workstation Docker where TRT is non-trivial to install.
Production code (new):
- components/c7_inference/pytorch_fp16_runtime.py — runtime +
PytorchEngineHandle + output-shape adapter
- components/c7_inference/architecture_registry.py — torch-free
register_architecture / default_registry / ArchitectureFactory
(Risk-1 mitigation: no L2->L3 back-edge from C7 into per-backbone
code)
- components/c7_inference/__init__.py — re-exports the registry
mechanism. Still does NOT import the concrete strategy module
(Invariant I-5)
- components/c7_inference/config.py — adds per_frame_debug_log bool
field (gates the DEBUG per-frame latency log)
Tests (new): tests/unit/c7_inference/test_pytorch_fp16_runtime.py
covers AC-1..AC-8 + NFRs. AC-1/2/6/7 + thermal/release/registry
guards run unconditionally (17 tests); AC-3/4/5/8 +
NFR-perf-deserialize + NFR-reliability-eval-mode require CUDA and
skip on Tier-1 CI / macOS dev.
Tests (modified):
- test_protocol_conformance.py — narrowed
test_ac5_build_inference_runtime_flag_on_but_module_missing
parametrisation to exclude pytorch_fp16 (now-built); TRT / ORT
still covered until AZ-298 / AZ-299 ship.
CI: .github/workflows/ci.yml lint + unit jobs now install
'-e .[dev,inference]' because mypy + pytest need torch + torchvision +
onnxruntime on the runner.
Three task-spec -> as-built deltas documented in
_docs/02_tasks/done/AZ-300_c7_pytorch_baseline.md Implementation Notes:
1. Constructor conforms to AZ-297 factory shape (config positional;
thermal_publisher + registry + clock keyword-only optionals).
AZ-302 will update the factory to thread thermal_publisher.
2. Architecture registry uses extras["model_name"] as lookup key
(avoids touching the frozen BuildConfig / EngineCacheEntry DTOs).
3. Warm-up forward deferred to AZ-300 tier-2 follow-up — the zero-arg
registry has no per-backbone input-shape metadata.
Suite: 1120 passed / 10 skipped (CUDA + Tier-2 + cmake / actionlint
environment gates). No regressions in non-c7_inference areas.
Co-authored-by: Cursor <cursoragent@cursor.com>
105 lines
3.5 KiB
YAML
105 lines
3.5 KiB
YAML
name: ci-tier1
|
|
|
|
on:
|
|
push:
|
|
branches: [dev, stage, main]
|
|
pull_request:
|
|
branches: [dev, stage, main]
|
|
|
|
jobs:
|
|
lint:
|
|
runs-on: ubuntu-22.04
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- uses: actions/setup-python@v5
|
|
with:
|
|
python-version: "3.10"
|
|
# AZ-300 — `[inference]` (torch + torchvision + onnxruntime) is now
|
|
# required for `mypy src` to type-check `c7_inference.pytorch_fp16_runtime`
|
|
# and for `pytest` to collect `test_pytorch_fp16_runtime.py`. Tier-1
|
|
# CI uses the CPU-only torch wheel; CUDA-gated tests skip themselves
|
|
# via `pytest.mark.skipif(not torch.cuda.is_available(), ...)`.
|
|
- run: pip install -e ".[dev,inference]"
|
|
- run: ruff check src tests
|
|
- run: mypy src
|
|
|
|
unit:
|
|
runs-on: ubuntu-22.04
|
|
needs: lint
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- uses: actions/setup-python@v5
|
|
with:
|
|
python-version: "3.10"
|
|
- run: pip install -e ".[dev,inference]"
|
|
- name: pytest unit (per-component coverage gate)
|
|
run: pytest -q --cov=gps_denied_onboard --cov-fail-under=75 tests/unit
|
|
|
|
integration:
|
|
runs-on: ubuntu-22.04
|
|
needs: unit
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- name: docker compose up
|
|
run: docker compose -f docker-compose.test.yml up --abort-on-container-exit --exit-code-from e2e-runner --build
|
|
|
|
build:
|
|
name: build-${{ matrix.kind }}
|
|
runs-on: ubuntu-22.04
|
|
needs: lint
|
|
strategy:
|
|
fail-fast: false
|
|
matrix:
|
|
kind: [deployment, research]
|
|
include:
|
|
# AZ-332 — BUILD_OKVIS2 forced OFF in Tier-1 CI until the tier2
|
|
# follow-up wires `okvis::ThreadedKFVio` end-to-end. The C++
|
|
# binding skeleton + CMake glue still ship in this build; full
|
|
# OKVIS2 native compile is gated on installing Ceres-solver +
|
|
# OKVIS2 vendored submodules (BRISK, DBoW2) via apt, plus
|
|
# `submodules: recursive` checkout. That CI lift is the
|
|
# tier2 task's surface, not AZ-332's.
|
|
- kind: deployment
|
|
cmake_flags: >-
|
|
-DBUILD_OKVIS2=OFF -DBUILD_VINS_MONO=OFF
|
|
-DBUILD_VPR_SALAD=OFF -DBUILD_C11_TILE_MANAGER=OFF
|
|
- kind: research
|
|
cmake_flags: >-
|
|
-DBUILD_OKVIS2=OFF -DBUILD_VINS_MONO=ON -DBUILD_VPR_SALAD=ON
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- run: cmake -S . -B build ${{ matrix.cmake_flags }}
|
|
- run: cmake --build build --parallel
|
|
|
|
sbom-diff:
|
|
runs-on: ubuntu-22.04
|
|
needs: build
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- uses: actions/setup-python@v5
|
|
with:
|
|
python-version: "3.10"
|
|
- name: SBOM diff (ADR-002 enforcement)
|
|
run: python ci/sbom_diff.py --deployment build-deployment-sbom.json --research build-research-sbom.json
|
|
|
|
security:
|
|
runs-on: ubuntu-22.04
|
|
needs: build
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- uses: actions/setup-python@v5
|
|
with:
|
|
python-version: "3.10"
|
|
- run: pip install pip-audit
|
|
- run: pip-audit -r pyproject.toml || true
|
|
- name: OpenCV pin gate (D-CROSS-CVE-1)
|
|
run: python ci/opencv_pin_gate.py --pyproject pyproject.toml
|
|
|
|
push-images:
|
|
runs-on: ubuntu-22.04
|
|
if: github.event_name == 'push' && contains(fromJson('["refs/heads/dev","refs/heads/stage","refs/heads/main"]'), github.ref)
|
|
needs: [unit, integration, build, sbom-diff, security]
|
|
steps:
|
|
- uses: actions/checkout@v4
|
|
- run: echo "push images to GHCR (deployment + research) — wiring lands per release task"
|