Epic H graph-capture closeout record¶
Scope and current disposition¶
Plan disposition: Shipped by maintainer decision on 2026-09-27. E8 and E8-F1 through E8-F8 are closed based on the maintainer's completion report. The decision supersedes the original requirement to keep plan metadata open until the designated-device evidence gate passes. It does not supply or certify missing measurements.
Measured-evidence disposition: UNSHIPPED/BLOCKED. E8-F7/T7 owns profiling and machine-bounded recommendations. F8-P3 records the evidence ledger; missing evidence remains unavailable and unmeasured, not zero or passing. The original 2026-09-07 evidence snapshot below is retained for follow-up.
Release sequence: merge this closeout, then publish 0.2.13 with both facades and containers. The planned 0.3.0 migration removes the facades and updates dependent workflows and all affected documentation and examples.
The committed record has no populated final revision, designated qualified CUDA device, measurement, or reviewed artifact. Measurement fields remain unavailable, and reviewed artifact fields remain unavailable, until qualifying evidence is collected and reviewed.
Closeout metadata¶
| Field | Value |
|---|---|
| Record date | 2026-09-07 |
| Final source revision | UNAVAILABLE — evidence not collected |
| Python | UNAVAILABLE — evidence not collected |
| Warp | UNAVAILABLE — evidence not collected |
| CUDA driver/runtime | UNAVAILABLE — evidence not collected |
| Warp availability | UNAVAILABLE — evidence not collected |
| Designated qualified CUDA device | UNAVAILABLE — no designated qualified CUDA device |
| Supplemental devices | UNAVAILABLE — evidence not collected |
Target-derivation ledger¶
Future producers must inspect the E8-F1 through E8-F7 implementation records
and final executable diff, exclude tests, docs, .opencode/plans/, and
.artifacts/, then freeze a sorted, deduplicated repository-relative
production-module list. These static refresh inputs are not a live dependency
of this record or its test:
| Track | Inline repository plan path |
|---|---|
| E8-F1 | .opencode/plans/sections/features/E8-F1/implementation_tasks.md |
| E8-F2 | .opencode/plans/sections/features/E8-F2/implementation_tasks.md |
| E8-F3 | .opencode/plans/sections/features/E8-F3/implementation_tasks.md |
| E8-F4 | .opencode/plans/sections/features/E8-F4/implementation_tasks.md |
| E8-F5 | .opencode/plans/sections/features/E8-F5/implementation_tasks.md |
| E8-F6 | .opencode/plans/sections/features/E8-F6/implementation_tasks.md |
| E8-F7 | .opencode/plans/sections/features/E8-F7/implementation_tasks.md |
Final executable diff: UNAVAILABLE — final executable diff not collected.
Candidate production modules¶
UNAVAILABLE — final executable diff not collected.
Frozen production targets¶
UNAVAILABLE — final executable diff not collected.
Future coverage rows use exactly this format:
<repository-relative module> | <literal pytest --cov module command> |
<term-missing output> | PASS|FAILED|UNAVAILABLE
| Coverage record | Requirement | Status |
|---|---|---|
| Target list | Frozen production targets | UNAVAILABLE |
| Aggregate changed-module coverage | >=80% |
UNAVAILABLE |
The P3 closeout document and its test are excluded from production targets.
Required evidence matrix¶
Only PASS with a populated final revision, designated device, required command
output, and applicable valid provenance satisfies a row. FAILED, STALE,
INFERRED, UNAVAILABLE, and CLEAN-SKIP do not. Supplemental devices are
metadata only.
| ID | Required evidence | Final revision | Designated device | Status | Evidence / blocker |
|---|---|---|---|---|---|
| H1 | Qualified capture/replay with no replay-time allocation, host transfer, or bulk synchronization | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C2, C7; E8-F2 incomplete; A1 |
| H2 | Deterministic rejection or explicit recapture for structural changes | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C2; E8-F2 incomplete; A1 |
| H3 | CPU, uncaptured Warp, and captured CUDA parity plus tight concentration-weighted conservation | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C2, C7; E8-F2 incomplete; A1 |
| H4 | Persistent nonaliasing RNG continuation, reset, and restart behavior | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C2; E8-F2 and E8-F3 incomplete; A1 |
| H5 | Co-located tests and configured >=80% coverage |
UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C4; frozen targets and aggregate gate unavailable |
| H6 | 1/10/100/1000-box scaling evidence or explicit unavailable rows | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C8; F6 artifact unavailable; A1 |
| H7 | Small/medium captured-versus-uncaptured launch-overhead provenance | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C8; F6 artifact unavailable; A1 |
| H8 | Logical memory model and observed peak-memory comparison | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C8; F6 artifact unavailable; A1 |
| H9 | Nsight occupancy/memory-access evidence and bounded follow-up | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C9; F7 artifact unavailable; A2 |
| H10 | Example/runbook/limitations and strict MkDocs result | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C5, C6; final closeout output unavailable |
| H11 | Exact closeout command matrix and literal results before promotion | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | C1–C9; literal results unavailable |
Command ledger¶
Historical command snapshot (2026-09-07): C1 and C6 refer to work-specific documentation tests removed on 2026-09-27 at maintainer request because they pinned mutable prose and plan status. They are retained below only as historical entries, not runnable instructions or future gates. Use strict MkDocs (C5) for documentation validation and the plan-schema validator for plan metadata. Future evidence collection must revise this ledger and its H-row mappings before applying the measured-evidence promotion rule.
Rows C1–C6 are validation commands. Focused rows are assertion-only, not coverage evidence; C4 is this record's sole repository coverage command, and C5 is its strict MkDocs documentation-rendering command. Rows C7–C9 are opt-in, availability-only native-CUDA evidence: pass-or-clean-skip is not required measured evidence and cannot discharge H1–H9.
C1¶
pytest particula/tests/gpu_graph_capture_closeout_docs_test.py -q --no-cov
NOT RUN — output must be pasted verbatim after execution
C2¶
pytest particula/execution/tests/graph_capture_test.py particula/execution/tests/captured_full_loop_test.py particula/execution/tests/rng_invariance_test.py particula/execution/tests/checkpoint_test.py -q --no-cov
NOT RUN — output must be pasted verbatim after execution
C3¶
.opencode/tools/run_linters.py
NOT RUN — output must be pasted verbatim after execution
C4¶
.opencode/tools/run_pytest.py
NOT RUN — output must be pasted verbatim after execution
C5¶
mkdocs build --strict
NOT RUN — output must be pasted verbatim after execution
C6¶
pytest particula/tests/gpu_graph_capture_runbook_docs_test.py -q --no-cov
NOT RUN — output must be pasted verbatim after execution
C7¶
pytest particula/execution/tests/captured_full_loop_test.py -q -m "warp and cuda" --no-cov
NOT RUN — output must be pasted verbatim after execution
C8¶
pytest particula/gpu/tests/benchmark_test.py --benchmark -k resident -v -s --no-cov
NOT RUN — output must be pasted verbatim after execution
C9¶
pytest particula/gpu/tests/profiling_smoke_test.py --benchmark -q --no-cov
NOT RUN — output must be pasted verbatim after execution
Artifact provenance rules¶
Each artifact/manifest ledger record must contain record ID, manifest pointer, schema/version, final source revision, workload ID, machine/device provenance, contained relative raw filename, byte size, and lowercase SHA-256. Reject absolute or traversal paths, latest-path selection, symlinks, raw local reports, and copied summaries.
| Record ID | Manifest pointer | Schema/version | Final revision | Workload | Provenance | Raw filename | Byte size | SHA-256 |
|---|---|---|---|---|---|---|---|---|
| A1 | UNAVAILABLE — absent reviewed F6 CUDA artifact | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE |
| A2 | UNAVAILABLE — absent reviewed F7 CUDA artifact | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE | UNAVAILABLE |
Blockers and promotion rule¶
The original evidence snapshot cited outstanding E8-F2 and E8-F3 work.
Those plan statuses are superseded by the 2026-09-27 maintainer closeout;
historical matrix references are not current implementation status.
Absent reviewed F6/F7 CUDA artifacts still block measured-evidence promotion.
Measured-evidence promotion requires H1–H11 PASS on one final revision and designated qualified
CUDA device, reviewed safe artifacts where applicable, all literal command
outputs, and frozen coverage targets/gate. Otherwise retain the measured-evidence
disposition UNSHIPPED/BLOCKED, independently of Shipped plan metadata.
A later evidence update must preserve row and command ordering, validate manifest consistency before changing statuses, and remain blocked if any required input is absent or fails. This is a retry-safe, auditable recovery path that does not mutate runtime state.