Dsv41 Scheduler Benchmark

Run and compare DeepSeek V4.1 Flash decode scheduler benchmarks on Ascend A5. Use for pypto-lib dev branches or daily CI performance cases, Simpler tests/fdwic/dsv41_scheduler, A5 HBG no-graph baselines, runtime A/B or stacked-PR comparisons, and L1/L3 traces for these cases.

49 stars 102 forksGit & Code Review

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add hw-native-sys/simpler --skill dsv41-scheduler-benchmark
Manual install
git clone --depth 1 https://github.com/hw-native-sys/simpler.git /tmp/simpler
mkdir -p ~/.claude/skills
cp -r /tmp/simpler/.claude/skills/dsv41-scheduler-benchmark ~/.claude/skills/dsv41-scheduler-benchmark

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# DeepSeek V4.1 scheduler benchmarks on A5

## Choose the workload

- Use [pypto-lib decode](references/pypto-lib.md) for the live dev branch, pypto-lib daily CI cases, or a no-graph A/B with live model kernels. Read the case manifest in the chosen lib checkout.
- Use [Simpler FDWIC](references/simpler-fdwic.md) for vendored scheduler cases, the 18-case matrix, or reproducible scheduler traces with captured inputs. Use the suite README in the chosen Simpler checkout.
- Use both when checking whether a runtime change generalizes. Report each suite with its own case inputs, timed region, and rank aggregation.

- For [percounter L1 and Perfetto](references/percounter-l1.md), capture the matching dependency graph and check that the timing records cover the executable tasks.

## Run a matched comparison

1. Pin the workload and each runtime arm. Record the resolved commits and any local diff for the relevant repos, CANN and toolchain versions, the loaded runtime build and scheduler variant, device IDs, graph mode, input shape and routing, warmup, rounds, and the metric and its aggregation. Verify the loaded artifacts for each arm. For a runtime-only comparison, hold the generated business kernels, inputs, graph mode, and build flags fixed.
Read the full skill on GitHub

Details

Skill name
dsv41-scheduler-benchmark
Path
.claude/skills/dsv41-scheduler-benchmark/SKILL.md
Repo updated
18 November 2026

FAQ

What does the Dsv41 Scheduler Benchmark skill do?

Run and compare DeepSeek V4.1 Flash decode scheduler benchmarks on Ascend A5. Use for pypto-lib dev branches or daily CI performance cases, Simpler tests/fdwic/dsv41_scheduler, A5 HBG no-graph baselines, runtime A/B or stacked-PR comparisons, and L1/L3 traces for these cases.

How do I install the Dsv41 Scheduler Benchmark skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add hw-native-sys/simpler --skill dsv41-scheduler-benchmark. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Dsv41 Scheduler Benchmark popular?

It lives in hw-native-sys/simpler, which has 49 GitHub stars and 102 forks.

Related skills

Benchmark

49

hw-native-sys/simpler

Benchmark runtime performance on hardware. If the current branch has commits ahead of upstream/main or uncommitted changes, compares against the fork point (merge-base). Otherwise benchmarks current state only. Use when the user asks to benchmark, measure performance, or compare latency.

Git & Code Review102

Checkout PR

49

hw-native-sys/simpler

Add a remote for a given PR's head repository and checkout a local pr-xxx-work branch. Use when the user wants to work on someone else's PR locally.

Git & Code Review102

Clean Branches

49

hw-native-sys/simpler

Remove stale local and remote git branches that have been merged into main. Detects regular merges and squash merges. Cleans up fork remote branches and prunes tracking refs. Use when the user asks to clean up branches, remove merged branches, or tidy up git.

Git & Code Review102

Fix Issue

49

hw-native-sys/simpler

Implement a GitHub Issue as its assignee or delegated contributor, selecting direct repair or develop-task planning and following through to a reviewed PR. Use when asked to fix or work on an Issue; use process-issue for maintainer triage.

Git & Code Review102

Dispatch

49

lllyys/vreader

Parallel lane dispatch — the ONLY sanctioned multi-item fan-out (feature #130, rule 55). Use when the user asks to fix several bugs at once ('fix #a #b #c'), implement independent feature WIs in parallel, or run a dispatch batch. Orchestrates worktree-isolated implementer lanes on leased simulators, validates HANDOFFs, and runs the serial integration tail (rebase → re-test → tracker/docs → version-at-slot → PR → merge-from-worktree → tag). A single item degrades to the inline /fix-issue or /feature-workflow flow.

Git & Code Review21

Fix PR

49

hw-native-sys/simpler

Fix GitHub PR issues — address all PR feedback (inline review threads, review summary bodies, and conversation comments) and resolve CI failures in a loop until the PR is fully clean. Fetches CI errors online and triages review feedback. Use when fixing PR problems, addressing review comments, or resolving CI failures.

Git & Code Review102

Browse the most popular Claude skills, all git & code review skills, or write your own with the skills generator.