hipDNN UHD (local)
One engine, one device, everything on the local machine: the user's GPU, checkout and build,
with hipDNN's own tools (hipdnn_corpus_gen, hipdnn_bench, hipdnn_uhd_features,
uhd_gen) and this skill's scripts/. Read the page for the ask; each names the next.
| Ask |
Page |
| generate, train or refresh an engine's heuristics |
generate.md |
| what data exists, the device key, the store, the slice; convert measurements; retrain or size from stored data |
data.md |
| is a model good enough |
evaluate.md, with reference.md for the report's fields |
| validate, ship, promote or install a model |
validate.md |
| a step failed, hangs, or produced a result that looks wrong |
triage.md |
| an engine that has never been trained |
onboarding.md, with corpus-facets.md |
The lifecycle: onboarding.md once per engine → generate.md
(shape set, slice, collect, store, train; L2 first, then L1 collected under it) →
evaluate.md → validate.md → install on the user's word.
Scripts
Standard library plus uhd_gen; run as python3 $UHD/scripts/<script>, UHD being this
skill's installed directory.
| Script |
Does |
eval_slice.py |
a set's evaluation slice; the union --eval-benchmarks reads; a regime-even sample; the slice's stored rows |
candidate_tree.py |
a real copy of the build's descriptor tree with candidate models installed |
engine_run.py |
used by the three below: one hipdnn_bench call that reads only the tree named for it, with an empty cache of its own |
validate_l1.py |
per graph: measured, incumbent prediction, candidate prediction |
validate_l2.py |
per graph: the engine run under the build's tree and the candidate tree, with the runtime's refusals |
decide.py |
the Stage V gates from those rows |
Rules for every page
- One device. Every set, collection, dataset and model is one product of one arch, keyed
<arch>-<product> (data.md). Never merge two devices' rows; with several
products present, ask which one.
- Nothing changes the checkout until the user says to install a validated model; commit
only when asked, and never push.
- Quote every warning and refusal verbatim in the report.
1---2name: hipdnn-uhd3description: "Generate, evaluate, validate and install hipDNN Universal Heuristic Descriptors (UHDs) for an engine on the local machine: shape sets the local GPU's engine serves (hipdnn_corpus_gen), evaluation slices, collections and stored datasets (uhd_gen), training per role (catalog ranker L2, engine predictor L1), the held-out judgement, Stage V on the local GPU and installation with uhd_gen promote, triage of failed runs, and readying a new engine for training. Use when asked to generate, train, refresh, judge, validate, ship or install heuristics for a hipDNN engine, to manage UHD training data, or to diagnose a UHD run."4argument-hint: "[generate|data|evaluate|validate|triage|onboard] <engine | model dir> [build=<dir>] [store=<dir>]"5allowed-tools: Bash, Read, Write, Grep, Glob6metadata:7 source: https://skillmd.com/skills/rocm/hipdnn-uhd8---910# hipDNN UHD (local)1112One engine, one device, everything on the local machine: the user's GPU, checkout and build,13with hipDNN's own tools (`hipdnn_corpus_gen`, `hipdnn_bench`, `hipdnn_uhd_features`,14`uhd_gen`) and this skill's `scripts/`. Read the page for the ask; each names the next.1516| Ask | Page |17|---|---|18| generate, train or refresh an engine's heuristics | [generate.md](generate.md) |19| what data exists, the device key, the store, the slice; convert measurements; retrain or size from stored data | [data.md](data.md) |20| is a model good enough | [evaluate.md](evaluate.md), with [reference.md](reference.md) for the report's fields |21| validate, ship, promote or install a model | [validate.md](validate.md) |22| a step failed, hangs, or produced a result that looks wrong | [triage.md](triage.md) |23| an engine that has never been trained | [onboarding.md](onboarding.md), with [corpus-facets.md](corpus-facets.md) |2425The lifecycle: [onboarding.md](onboarding.md) once per engine → [generate.md](generate.md)26(shape set, slice, collect, store, train; L2 first, then L1 collected under it) →27[evaluate.md](evaluate.md) → [validate.md](validate.md) → install on the user's word.2829## Scripts3031Standard library plus `uhd_gen`; run as `python3 $UHD/scripts/<script>`, `UHD` being this32skill's installed directory.3334| Script | Does |35|---|---|36| [`eval_slice.py`](scripts/eval_slice.py) | a set's evaluation slice; the union `--eval-benchmarks` reads; a regime-even sample; the slice's stored rows |37| [`candidate_tree.py`](scripts/candidate_tree.py) | a real copy of the build's descriptor tree with candidate models installed |38| [`engine_run.py`](scripts/engine_run.py) | used by the three below: one `hipdnn_bench` call that reads only the tree named for it, with an empty cache of its own |39| [`validate_l1.py`](scripts/validate_l1.py) | per graph: measured, incumbent prediction, candidate prediction |40| [`validate_l2.py`](scripts/validate_l2.py) | per graph: the engine run under the build's tree and the candidate tree, with the runtime's refusals |41| [`decide.py`](scripts/decide.py) | the Stage V gates from those rows |4243## Rules for every page4445- **One device.** Every set, collection, dataset and model is one product of one arch, keyed46 `<arch>-<product>` ([data.md](data.md)). Never merge two devices' rows; with several47 products present, ask which one.48- **Nothing changes the checkout** until the user says to install a validated model; commit49 only when asked, and never push.50- Quote every warning and refusal verbatim in the report.