Skip to content

Ascend/msagent / skills/operator/ops-performance-tuning

ops-performance-tuning

Verified

Compile, profile, diagnose, optimize, and compare Ascend NPU operators across CANN versions and A2/A3/A5 for Ascend C, CATLASS, Triton-Ascend, TileLang-Ascend, PyPTO, and SHMEM/MC2. Use when an operator already has a runnable implementation and the user asks for msOpProf collection, bottleneck analysis, source-level tuning, or a reproducible before/after performance report. Do not use as the primary workflow for operator creation, migration, or unresolved correctness failures.

FileKindSize
SKILL.mdskill md15.4 KB
evals/evals.jsonconfig3.6 KB
references/case-routing.mdmarkdown30.0 KB
references/cases/ascendc/pr_ascendc_allgathermm_scalar_align.mdmarkdown1.7 KB
references/cases/ascendc/pr_ascendc_cann_samples_perf_story.mdmarkdown3.9 KB
references/cases/ascendc/pr_ascendc_cgdr_vf_inverse.mdmarkdown1.6 KB
references/cases/ascendc/pr_ascendc_conv3ddx_tque_tbuf.mdmarkdown1.6 KB
references/cases/ascendc/pr_ascendc_layernorm_areg.mdmarkdown1.6 KB
references/cases/ascendc/pr_ascendc_logit_a5_microapi.mdmarkdown1.9 KB
references/cases/ascendc/pr_ascendc_matmul_case_18x.mdmarkdown1.5 KB
references/cases/ascendc/pr_ascendc_matmul_full_load_wiki.mdmarkdown1.6 KB
references/cases/ascendc/pr_ascendc_matmul_vcv_wiki.mdmarkdown1.8 KB
references/cases/ascendc/pr_ascendc_mc2_fusion.mdmarkdown1.4 KB
references/cases/ascendc/pr_ascendc_mhc_24x.mdmarkdown1.2 KB
references/cases/ascendc/pr_ascendc_msopprof_add_block_imbalance.mdmarkdown1.7 KB
references/cases/ascendc/pr_ascendc_quantize_a2_micro_opt.mdmarkdown2.0 KB
references/cases/ascendc/pr_ascendc_rotate_quant_blaze.mdmarkdown2.8 KB
references/cases/ascendc/pr_ascendc_tiling_l2_official.mdmarkdown1.6 KB
references/cases/catlass/pr_catlass_baseline.mdmarkdown1.3 KB
references/cases/catlass/pr_catlass_fixpipe_dualdst.mdmarkdown1.6 KB
references/cases/catlass/pr_catlass_grouped_matmul_tiling.mdmarkdown1.5 KB
references/cases/catlass/pr_catlass_mstuner_3p8x.mdmarkdown1.9 KB
references/cases/catlass/pr_catlass_swizzle_balance.mdmarkdown1.6 KB
references/cases/pypto/pr_pypto_gqa_decode_vector_merge.mdmarkdown2.4 KB
references/cases/pypto/pr_pypto_l2_hit_layout_s13.mdmarkdown1.9 KB
references/cases/pypto/pr_pypto_per_matmul_tile_shapes.mdmarkdown2.4 KB
references/cases/pypto/pr_pypto_small_shape_matmul.mdmarkdown2.3 KB
references/cases/pypto/pr_pypto_weight_none_cacheable.mdmarkdown2.6 KB
references/cases/shmem/pr_shmem_alltoallv_mr152.mdmarkdown2.4 KB
references/cases/shmem/pr_shmem_reduce_scatter_serial.mdmarkdown2.6 KB
references/cases/tilelang/pr_tilelang_benchmark_vs_ascendc.mdmarkdown1.7 KB
references/cases/tilelang/pr_tilelang_compiler_prs.mdmarkdown2.3 KB
references/cases/tilelang/pr_tilelang_fa_pr698.mdmarkdown1.5 KB
references/cases/tilelang/pr_tilelang_gqa_bwd_pr1494.mdmarkdown1.3 KB
references/cases/tilelang/pr_tilelang_persistent_mqa_decode.mdmarkdown2.2 KB
references/cases/tilelang/pr_tilelang_sfa_pr665.mdmarkdown914 B
references/cases/triton/pr_triton_atomic_simd_24x.mdmarkdown1.8 KB
references/cases/triton/pr_triton_autotune_background.mdmarkdown1.0 KB
references/cases/triton/pr_triton_constexpr_recompile.mdmarkdown1.1 KB
references/cases/triton/pr_triton_lgamma_simt.mdmarkdown1.6 KB
references/compile/compile-ascendc.mdmarkdown9.0 KB
references/compile/compile-catlass.mdmarkdown2.4 KB
references/compile/compile-pypto.mdmarkdown2.0 KB
references/compile/compile-shmem.mdmarkdown2.6 KB
references/compile/compile-tilelang.mdmarkdown2.7 KB
references/compile/compile-triton.mdmarkdown4.4 KB
references/diagnose/diagnose-l2-cache.mdmarkdown2.2 KB
references/diagnose/diagnose-occupancy.mdmarkdown4.0 KB
references/diagnose/diagnose-pipeline.mdmarkdown4.1 KB
references/diagnose/diagnose-roofline.mdmarkdown3.1 KB
references/optimize/optimize-api-usage.mdmarkdown1.7 KB
references/optimize/optimize-ascendc-regbase-vf.mdmarkdown5.9 KB
references/optimize/optimize-ascendc-tiling.mdmarkdown935 B
references/optimize/optimize-catlass.mdmarkdown3.0 KB
references/optimize/optimize-data-copy.mdmarkdown870 B
references/optimize/optimize-index.mdmarkdown3.6 KB
references/optimize/optimize-memory-hierarchy.mdmarkdown599 B
references/optimize/optimize-pipeline.mdmarkdown1.0 KB
references/optimize/optimize-pypto.mdmarkdown1.5 KB
references/optimize/optimize-shmem.mdmarkdown5.1 KB
references/optimize/optimize-tilelang.mdmarkdown1.7 KB
references/optimize/optimize-triton.mdmarkdown3.8 KB
references/profile/benchmark-harness.mdmarkdown2.2 KB
references/profile/profile-msopprof.mdmarkdown7.4 KB
references/troubleshooting.mdmarkdown5.8 KB
scripts/bound_analyzer.pyscript9.0 KB
scripts/delta_report.pyscript4.8 KB
scripts/env_check.shscript5.0 KB
scripts/result_saver.pyscript5.0 KB

scripts/result_saver.py

View on GitHub
#!/usr/bin/env python3
"""结果文件落盘工具 — 生成符合 skill 契约的 <variant>_<board|sim>_<timestamp>.json。

用法:
  python3 scripts/result_saver.py \
    --variant baseline \
    --mode board \
    --soc a5 \
    --precision pass \
    --kernel-avg-us 336.6 \
    --output results/

生成: results/baseline_board_20260808T120000Z.json
"""

from __future__ import annotations

import argparse
import json
import sys
from datetime import datetime, timezone
from pathlib import Path


def main() -> int:
    parser = argparse.ArgumentParser(description="Save tuning result as structured JSON")
    parser.add_argument("--op", required=True, help="算子或性能用例名称")
    parser.add_argument("--variant", required=True, help="源码变体名 (e.g. baseline, round1, round2)")
    parser.add_argument("--mode", required=True, choices=("board", "sim"), help="运行模式")
    parser.add_argument("--soc", required=True, help="芯片代际 (a2/a3/a5)")
    parser.add_argument("--precision", required=True, choices=("pass", "fail"), help="精度校验结果")
    parser.add_argument("--mismatch", type=int, default=0, help="精度不匹配的数量")
    parser.add_argument("--max-abs-err", type=float, default=0.0, help="最大绝对误差")
    parser.add_argument("--max-rel-err", type=float, default=0.0, help="最大相对误差")
    parser.add_argument("--kernel-avg-us", type=float, required=True, help="kernel 平均耗时 (µs)")
    parser.add_argument("--timing-method", default="event",
                        choices=("event", "sim", "msprof", "msprof_task_duration", "profiler_summary"),
                        help="计时方法:event=设备 event 计时(首选);msprof_task_duration=工程无 event 设施且 "
                             "kernel 为长耗时(ms 级)时用 OpBasicInfo Task Duration 作基线;"
                             "profiler_summary=工程自带 profiler 对比框架(如 torch_npu op_summary)")
    parser.add_argument("--status", default="ok", choices=("ok", "partial", "dry_run"),
                        help="结果状态:partial=采集/验证不完整,dry_run=未上板仅离线结论")
    parser.add_argument("--baseline-kind", default="self_before_after",
                        choices=("system", "self_before_after"), help="基线类型")
    parser.add_argument("--cann-version", default=None, help="CANN 完整版本")
    parser.add_argument("--repo-commit", default=None,
                        help="源码 commit;非 git 工程填 n/a 或 sha256:<源码hash>")
    parser.add_argument("--device-id", default=None, help="NPU 设备 ID")
    parser.add_argument("--shape", default=None, help="逻辑/物理 shape;多组时写用例 ID")
    parser.add_argument("--dtype", default=None, help="输入输出 dtype 摘要")
    parser.add_argument("--format", default=None, help="输入输出 format 摘要")
    parser.add_argument("--tiling-key", default=None, help="TilingKey")
    parser.add_argument("--block-dim", type=int, default=None, help="blockDim")
    parser.add_argument("--warmup", type=int, default=None, help="预热次数")
    parser.add_argument("--repeat", type=int, default=None, help="正式采样次数")
    parser.add_argument("--msprof-dir", default=None, help="msprof 采集目录路径(无则 null)")
    parser.add_argument("--exit-code", type=int, default=0, help="运行退出码")
    parser.add_argument("--output", required=True, help="结果输出目录")
    parser.add_argument("--extra", nargs="*", default=[],
                        help="额外键值对 key=value")
    args = parser.parse_args()

    output_dir = Path(args.output)
    output_dir.mkdir(parents=True, exist_ok=True)

    timestamp = datetime.now(timezone.utc).strftime("%Y%m%dT%H%M%SZ")
    filename = f"{args.variant}_{args.mode}_{timestamp}.json"
    filepath = output_dir / filename

    payload = {
        "op": args.op,
        "variant": args.variant,
        "soc": args.soc,
        "mode": args.mode,
        "status": args.status,
        "precision": args.precision,
        "mismatch": args.mismatch,
        "maxAbsErr": args.max_abs_err,
        "maxRelErr": args.max_rel_err,
        "kernel_avg_us": args.kernel_avg_us,
        "timing_method": args.timing_method,
        "baseline_kind": args.baseline_kind,
        "cann_version": args.cann_version,
        "repo_commit": args.repo_commit,
        "device_id": args.device_id,
        "shape": args.shape,
        "dtype": args.dtype,
        "format": args.format,
        "tiling_key": args.tiling_key,
        "block_dim": args.block_dim,
        "warmup": args.warmup,
        "repeat": args.repeat,
        "msprof_dir": args.msprof_dir,
        "exit_code": args.exit_code,
        "timestamp": timestamp,
    }
    for kv in args.extra:
        if "=" in kv:
            k, v = kv.split("=", 1)
            payload[k] = v

    filepath.write_text(json.dumps(payload, indent=2, ensure_ascii=False) + "\n",
                        encoding="utf-8")
    print(f"Result saved: {filepath}")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())
planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies.Stars: 108kUpdated Claude CodeCodexVerifiedInstall
meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要".Stars: 108kUpdated Claude CodeCodexVerifiedInstall