{
  "generated_at": "2026-09-10T00:25:09.269118+00:00",
  "source": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/analysis/nsys_worker_flush_v2_physical/kernel_invocations.jsonl.gz",
  "source_sha256": "b93ccf0d00c664d889b6c4561d862e1ffc9b74ccfc7ee21ee62da6d514f618cb",
  "scanned_calls": 1091749,
  "aggregate_reconciliation": "passed for non-NCCL, NCCL and intersection on each physical GPU",
  "scope": "Four profiled CPU-ready delivery intervals only. Non-NCCL includes non-compute kernels. Kernel gaps exclude memcpy/memset/CPU records and are not proven idle or waiting. Temporal overlap is not hardware concurrency or speedup. Boundary-crossing kernels counted in each intersecting window.",
  "rows": [
    {
      "chunk": 5,
      "physical_gpu": 0,
      "window_ms": 847.493771,
      "kernel_calls_intersecting": 11081,
      "non_nccl_union_ms": 672.786462,
      "nccl_union_ms": 88.813929,
      "non_nccl_nccl_overlap_ms": 21.212455,
      "kernel_union_ms": 740.387936,
      "without_recorded_kernel_ms": 107.105835,
      "longest_without_recorded_kernel_ms": 40.557143
    },
    {
      "chunk": 5,
      "physical_gpu": 1,
      "window_ms": 847.493771,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 672.822567,
      "nccl_union_ms": 76.603894,
      "non_nccl_nccl_overlap_ms": 19.479262,
      "kernel_union_ms": 729.947199,
      "without_recorded_kernel_ms": 117.546572,
      "longest_without_recorded_kernel_ms": 40.768828
    },
    {
      "chunk": 5,
      "physical_gpu": 2,
      "window_ms": 847.493771,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 674.870597,
      "nccl_union_ms": 84.763244,
      "non_nccl_nccl_overlap_ms": 20.748557,
      "kernel_union_ms": 738.885284,
      "without_recorded_kernel_ms": 108.608487,
      "longest_without_recorded_kernel_ms": 40.746487
    },
    {
      "chunk": 5,
      "physical_gpu": 3,
      "window_ms": 847.493771,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 681.34997,
      "nccl_union_ms": 72.69876,
      "non_nccl_nccl_overlap_ms": 15.424485,
      "kernel_union_ms": 738.624245,
      "without_recorded_kernel_ms": 108.869526,
      "longest_without_recorded_kernel_ms": 40.805944
    },
    {
      "chunk": 6,
      "physical_gpu": 0,
      "window_ms": 853.145076,
      "kernel_calls_intersecting": 11081,
      "non_nccl_union_ms": 673.0043,
      "nccl_union_ms": 76.4586,
      "non_nccl_nccl_overlap_ms": 19.191627,
      "kernel_union_ms": 730.271273,
      "without_recorded_kernel_ms": 122.873803,
      "longest_without_recorded_kernel_ms": 48.94991
    },
    {
      "chunk": 6,
      "physical_gpu": 1,
      "window_ms": 853.145076,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 670.476316,
      "nccl_union_ms": 84.899495,
      "non_nccl_nccl_overlap_ms": 21.249108,
      "kernel_union_ms": 734.126703,
      "without_recorded_kernel_ms": 119.018373,
      "longest_without_recorded_kernel_ms": 49.125901
    },
    {
      "chunk": 6,
      "physical_gpu": 2,
      "window_ms": 853.145076,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 673.051902,
      "nccl_union_ms": 76.541392,
      "non_nccl_nccl_overlap_ms": 19.646698,
      "kernel_union_ms": 729.946596,
      "without_recorded_kernel_ms": 123.19848,
      "longest_without_recorded_kernel_ms": 49.133769
    },
    {
      "chunk": 6,
      "physical_gpu": 3,
      "window_ms": 853.145076,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 677.783353,
      "nccl_union_ms": 74.12971,
      "non_nccl_nccl_overlap_ms": 16.278225,
      "kernel_union_ms": 735.634838,
      "without_recorded_kernel_ms": 117.510238,
      "longest_without_recorded_kernel_ms": 49.153333
    },
    {
      "chunk": 7,
      "physical_gpu": 0,
      "window_ms": 853.727596,
      "kernel_calls_intersecting": 11081,
      "non_nccl_union_ms": 669.86607,
      "nccl_union_ms": 80.790155,
      "non_nccl_nccl_overlap_ms": 19.667247,
      "kernel_union_ms": 730.988978,
      "without_recorded_kernel_ms": 122.738618,
      "longest_without_recorded_kernel_ms": 47.675986
    },
    {
      "chunk": 7,
      "physical_gpu": 1,
      "window_ms": 853.727596,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 670.583,
      "nccl_union_ms": 78.106613,
      "non_nccl_nccl_overlap_ms": 20.147244,
      "kernel_union_ms": 728.542369,
      "without_recorded_kernel_ms": 125.185227,
      "longest_without_recorded_kernel_ms": 47.190355
    },
    {
      "chunk": 7,
      "physical_gpu": 2,
      "window_ms": 853.727596,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 674.273854,
      "nccl_union_ms": 75.545737,
      "non_nccl_nccl_overlap_ms": 18.831785,
      "kernel_union_ms": 730.987806,
      "without_recorded_kernel_ms": 122.73979,
      "longest_without_recorded_kernel_ms": 47.190654
    },
    {
      "chunk": 7,
      "physical_gpu": 3,
      "window_ms": 853.727596,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 679.622264,
      "nccl_union_ms": 68.339076,
      "non_nccl_nccl_overlap_ms": 15.369264,
      "kernel_union_ms": 732.592076,
      "without_recorded_kernel_ms": 121.13552,
      "longest_without_recorded_kernel_ms": 47.191068
    },
    {
      "chunk": 8,
      "physical_gpu": 0,
      "window_ms": 856.183643,
      "kernel_calls_intersecting": 11081,
      "non_nccl_union_ms": 670.027501,
      "nccl_union_ms": 92.971412,
      "non_nccl_nccl_overlap_ms": 21.929839,
      "kernel_union_ms": 741.069074,
      "without_recorded_kernel_ms": 115.114569,
      "longest_without_recorded_kernel_ms": 49.730648
    },
    {
      "chunk": 8,
      "physical_gpu": 1,
      "window_ms": 856.183643,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 671.052943,
      "nccl_union_ms": 88.775992,
      "non_nccl_nccl_overlap_ms": 21.304834,
      "kernel_union_ms": 738.524101,
      "without_recorded_kernel_ms": 117.659542,
      "longest_without_recorded_kernel_ms": 50.411854
    },
    {
      "chunk": 8,
      "physical_gpu": 2,
      "window_ms": 856.183643,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 676.033294,
      "nccl_union_ms": 85.223517,
      "non_nccl_nccl_overlap_ms": 20.260222,
      "kernel_union_ms": 740.996589,
      "without_recorded_kernel_ms": 115.187054,
      "longest_without_recorded_kernel_ms": 49.969824
    },
    {
      "chunk": 8,
      "physical_gpu": 3,
      "window_ms": 856.183643,
      "kernel_calls_intersecting": 11055,
      "non_nccl_union_ms": 682.199375,
      "nccl_union_ms": 62.48545,
      "non_nccl_nccl_overlap_ms": 14.744423,
      "kernel_union_ms": 729.940402,
      "without_recorded_kernel_ms": 126.243241,
      "longest_without_recorded_kernel_ms": 50.387669
    }
  ]
}
