{ "status": "derived_from_audited_measurements", "generated_at": "2026-09-10T05:29:34.293429+00:00", "rows": [ { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_255d8b6427b0b5a79c74a59d751e921af25bbb329fb9011dc0ba8d8b5cc46d94.pt", "input_sha256": "8e60010c4ab22e3ad90412ff00e301e28227cf54db05320e456d91009e843787", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 13516800, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "output_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 13516800, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "shape_parameters": { "M": 2640, "N": 5120, "K": 5120 }, "mac_flops": 138412032000, "bias_additions": 13516800, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 185.344, "dram_read_bytes": 161596672.0, "dram_write_bytes": 24357888.0, "dram_total_bytes": 185954560.0, "nominal_mac_tflops_per_s": 746.7845303867404, "nominal_mac_flops_per_dram_byte": 744.3325509199667, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_255d8b6427b0b5a7_v1", "ncu_metrics_sha256": "43416f95805c211eba76b7145dc8697030d8356b8fd9957b5818538975bfbbe0", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_db47f7eb1f64a850cbdd6262502dc5221a08eb3d9e758d3e5bae5588b6ecf333.pt", "input_sha256": "5486c3201c4910ee4de3cbdc2b72d468ede3117a03f1e731537cf900e60ecc9d", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 13516800, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "output_tensor": { "shape": [ 1, 2640, 15360 ], "stride": [ 40550400, 15360, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 81100800 }, "shape_parameters": { "M": 2640, "N": 15360, "K": 5120 }, "mac_flops": 415236096000, "bias_additions": 40550400, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 518.112, "dram_read_bytes": 451653888.0, "dram_write_bytes": 76100864.0, "dram_total_bytes": 527754752.0, "nominal_mac_tflops_per_s": 801.4408004446916, "nominal_mac_flops_per_dram_byte": 786.7974554969047, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_db47f7eb1f64a850_v1", "ncu_metrics_sha256": "91c8546c250f9ef8e50ba379ebf0cab402c9e366c37d0735baebf50912d4c6d9", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_1b9e93116d23540836f4bb145cbed7f9ee9fb653806d192ee63e4ee58258c639.pt", "input_sha256": "75bf41dc2a7a283f82bdcd0635998b9bc8c93a30b0ebb2b209ce44fd3b9e360c", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 2640, 13824 ], "stride": [ 36495360, 13824, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 72990720 }, "output_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 13516800, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "shape_parameters": { "M": 2640, "N": 5120, "K": 13824 }, "mac_flops": 373712486400, "bias_additions": 13516800, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 472.896, "dram_read_bytes": 436227072.0, "dram_write_bytes": 22518272.0, "dram_total_bytes": 458745344.0, "nominal_mac_tflops_per_s": 790.263580998782, "nominal_mac_flops_per_dram_byte": 814.6403909878156, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_1b9e93116d235408_v1", "ncu_metrics_sha256": "4f8fc65b6645bc2ead7afa5c2da4ecd5b06e5b2753de811bbfb90f26ca6ef6fb", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_9ba227caa2b6dfb28443bedbde400547990f9c556fecfab30d04b1ab02ea72f8.pt", "input_sha256": "c08dc8cdb79c43328ef6849ee7e60049c20bbb243f50823bd9a80415eb91dc94", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 13516800, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "output_tensor": { "shape": [ 1, 2640, 13824 ], "stride": [ 36495360, 13824, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 72990720 }, "shape_parameters": { "M": 2640, "N": 13824, "K": 5120 }, "mac_flops": 373712486400, "bias_additions": 36495360, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 451.03999999999996, "dram_read_bytes": 384699648.0, "dram_write_bytes": 68826368.0, "dram_total_bytes": 453526016.0, "nominal_mac_tflops_per_s": 828.5573040085136, "nominal_mac_flops_per_dram_byte": 824.0155431347956, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_9ba227caa2b6dfb2_v1", "ncu_metrics_sha256": "395668c3b933dfb7929f93b0ae5461ad6886bd1372fac00bb7fc09403084a16f", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_697ad043e70ad8413b17addb52caa1db79ce79f8cedf8b49b4ed075d8e8a9f37.pt", "input_sha256": "984423718cf2b4425d4e6c3c3cc45ee71db6a1b3d36906b61da17257f41a950c", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 54067200, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "output_tensor": { "shape": [ 1, 2640, 5120 ], "stride": [ 13516800, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 27033600 }, "shape_parameters": { "M": 2640, "N": 5120, "K": 5120 }, "mac_flops": 138412032000, "bias_additions": 13516800, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 183.52, "dram_read_bytes": 161599232.0, "dram_write_bytes": 23164672.0, "dram_total_bytes": 184763904.0, "nominal_mac_tflops_per_s": 754.2068003487359, "nominal_mac_flops_per_dram_byte": 749.1291805568256, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_697ad043e70ad841_v1", "ncu_metrics_sha256": "a63f156a08877d29b1133170a0382e77a181045f822abb0b93c8ff78b947a09e", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_0551e0aa26bb720178c9c4337e9d3d1c584a8c49c70d3bcccd459eb0953c470e.pt", "input_sha256": "361110b0df425093397551c589c4c8f8e119f195c1b7c8ce7fe463e598338f6e", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 96, 6, 706, 322 ], "stride": [ 130943232, 1363992, 227332, 322, 1 ], "dtype": "torch.float32", "logical_bytes": 523772928 }, "output_tensor": { "shape": [ 1, 96, 4, 704, 320 ], "stride": [ 86507520, 901120, 225280, 320, 1 ], "dtype": "torch.float32", "logical_bytes": 346030080 }, "shape_parameters": { "input": [ 1, 96, 6, 706, 322 ], "weight": [ 96, 96, 3, 3, 3 ], "output": [ 1, 96, 4, 704, 320 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 448454983680, "bias_additions": 86507520, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 3677.9519999999998, "dram_read_bytes": 2266104000.0, "dram_write_bytes": 1493268480.0, "dram_total_bytes": 3759372480.0, "nominal_mac_tflops_per_s": 121.93062434746294, "nominal_mac_flops_per_dram_byte": 119.28985118282294, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_0551e0aa26bb7201_v1", "ncu_metrics_sha256": "2bc7b21500e77e41becaebf0c80515918ddb5a283e183d46b987e5d33f92a12c", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_1b34e3d0553bbca40db451e4cbfdf5d4bc360830810855ac6cfe66761b05df84.pt", "input_sha256": "836c7ee7ca56334d7997004c7d80dc3a7742322bc7befd6ad4a0f29a4ec8ad0f", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 192, 6, 354, 162 ], "stride": [ 66064896, 344088, 57348, 162, 1 ], "dtype": "torch.float32", "logical_bytes": 264259584 }, "output_tensor": { "shape": [ 1, 192, 4, 352, 160 ], "stride": [ 43253760, 225280, 56320, 160, 1 ], "dtype": "torch.float32", "logical_bytes": 173015040 }, "shape_parameters": { "input": [ 1, 192, 6, 354, 162 ], "weight": [ 192, 192, 3, 3, 3 ], "output": [ 1, 192, 4, 352, 160 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 448454983680, "bias_additions": 43253760, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 2130.8480000000004, "dram_read_bytes": 1223779072.0, "dram_write_bytes": 715595776.0, "dram_total_bytes": 1939374848.0, "nominal_mac_tflops_per_s": 210.4584577032242, "nominal_mac_flops_per_dram_byte": 231.2368772558197, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_1b34e3d0553bbca4_v1", "ncu_metrics_sha256": "b79f1f4706c0c6217b39b35e0b7bc0bfc021033c644d22457e81308e664617cb", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_606cff6747219baca3ee85dedb95903e07e617cd8ab9a51e1cfb1ca479cb34e8.pt", "input_sha256": "70fed7084bfb383625de9935f9cc6ad7e398ccdf921a0e71276f5cf1bdbdcb0d", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 10560, 5120 ], "stride": [ 54067200, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 108134400 }, "output_tensor": { "shape": [ 1, 10560, 5120 ], "stride": [ 54067200, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 108134400 }, "shape_parameters": { "M": 10560, "N": 5120, "K": 5120 }, "mac_flops": 553648128000, "bias_additions": 54067200, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 650.3359999999999, "dram_read_bytes": 569925120.0, "dram_write_bytes": 108107520.0, "dram_total_bytes": 678032640.0, "nominal_mac_tflops_per_s": 851.3262805688138, "nominal_mac_flops_per_dram_byte": 816.5508492334528, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_606cff6747219bac_v1", "ncu_metrics_sha256": "4e58ec02c1d23b861e259aba7b6e394c38455e83002c15868d0b33c63e986707", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_ac3438e4ac0a94854310416adc1640b1bd6b5edf5a633abf80b6aec521127dfc.pt", "input_sha256": "ae178ca2bbf227553643572af23755eb3ed51289f32398e11a61f88fb81ea3a1", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 384, 4, 178, 82 ], "stride": [ 22419456, 58384, 14596, 82, 1 ], "dtype": "torch.float32", "logical_bytes": 89677824 }, "output_tensor": { "shape": [ 1, 384, 2, 176, 80 ], "stride": [ 10813440, 28160, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 43253760 }, "shape_parameters": { "input": [ 1, 384, 4, 178, 82 ], "weight": [ 384, 384, 3, 3, 3 ], "output": [ 1, 384, 2, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 224227491840, "bias_additions": 10813440, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 1129.5360000000003, "dram_read_bytes": 473146368.0, "dram_write_bytes": 152203008.0, "dram_total_bytes": 625349376.0, "nominal_mac_tflops_per_s": 198.5129219785823, "nominal_mac_flops_per_dram_byte": 358.5635493462138, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_ac3438e4ac0a9485_v1", "ncu_metrics_sha256": "47a5b718228e567b8ac7e9550e0916d17be1aa4bb3f7a29b3a31058b0282232f", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_a50d349fe3a8545fe9987302bd3c3111cbad08a76e2a228f1f1d0c0c8fde3d0e.pt", "input_sha256": "fd2ac936cea7b4d5d60612a6045f5c612e27536de2424aa111e4bc689fc1e795", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 384, 3, 90, 42 ], "stride": [ 4354560, 11340, 3780, 42, 1 ], "dtype": "torch.float32", "logical_bytes": 17418240 }, "output_tensor": { "shape": [ 1, 384, 1, 88, 40 ], "stride": [ 1351680, 3520, 3520, 40, 1 ], "dtype": "torch.float32", "logical_bytes": 5406720 }, "shape_parameters": { "input": [ 1, 384, 3, 90, 42 ], "weight": [ 384, 384, 3, 3, 3 ], "output": [ 1, 384, 1, 88, 40 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 28028436480, "bias_additions": 1351680, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 180.32, "dram_read_bytes": 77619200.0, "dram_write_bytes": 6324992.0, "dram_total_bytes": 83944192.0, "nominal_mac_tflops_per_s": 155.43720319432123, "nominal_mac_flops_per_dram_byte": 333.89369546853226, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_a50d349fe3a8545f_v1", "ncu_metrics_sha256": "aff4d277129a49b56a0b330a7043300d5a021b5d482de1d412b8ef46e3ea21c2", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_4cfc56167ab9dd353fa4190c84c2f45332881ef9d5c320f23c49fa58121e2664.pt", "input_sha256": "94f48ff6a98bf87d9ac8185c5db13ca77c238c34547a0c86dd5cac7b7d5b326f", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 10560, 1536 ], "stride": [ 16220160, 1536, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 32440320 }, "output_tensor": { "shape": [ 1, 10560, 5120 ], "stride": [ 54067200, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 108134400 }, "shape_parameters": { "M": 10560, "N": 5120, "K": 1536 }, "mac_flops": 166094438400, "bias_additions": 54067200, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 208.928, "dram_read_bytes": 77009920.0, "dram_write_bytes": 92732928.0, "dram_total_bytes": 169742848.0, "nominal_mac_tflops_per_s": 794.9841017001072, "nominal_mac_flops_per_dram_byte": 978.5062543548227, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_4cfc56167ab9dd35_v1", "ncu_metrics_sha256": "92b9c5fb515f4e9cd1e26b9dfd38bfbd18745ffdfb4c638173a116a9f450416d", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [ { "gpu_id": 0, "pid": 2019660, "user": "unknown", "kind": "foreign", "holder": "zhoutaichang", "holder_account": "zhoutaichang", "command": "[No data]", "memory_mb": 2266, "first_seen": "2026-09-10T05:20:53.219014752Z", "last_seen": "2026-09-10T05:20:54.100960726Z", "sightings": 3, "warn_count": 0, "last_warned": null, "last_signal": null, "end_time": "2026-09-10T05:20:55.549497427Z", "resolution": "process exited" } ] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_053cda5cb0c0dcf6b32fffde86f872cce8fad23b8ca778791892cd7de2f95c12.pt", "input_sha256": "146fb336db70349e18bee5b6332d0ab63940d20d95e88e53d065f0d87a4673e8", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 192, 4, 178, 82 ], "stride": [ 11209728, 58384, 14596, 82, 1 ], "dtype": "torch.float32", "logical_bytes": 44838912 }, "output_tensor": { "shape": [ 1, 384, 2, 176, 80 ], "stride": [ 10813440, 28160, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 43253760 }, "shape_parameters": { "input": [ 1, 192, 4, 178, 82 ], "weight": [ 384, 192, 3, 3, 3 ], "output": [ 1, 384, 2, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 112113745920, "bias_additions": 10813440, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 603.9679999999998, "dram_read_bytes": 222305792.0, "dram_write_bytes": 102874624.0, "dram_total_bytes": 325180416.0, "nominal_mac_tflops_per_s": 185.62861926459684, "nominal_mac_flops_per_dram_byte": 344.7739790086252, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_053cda5cb0c0dcf6_v1", "ncu_metrics_sha256": "2c331c70a4a01e031b488d5080c59a8d50c4cba7b78d20a59b50b420601ecfdf", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_4273bf5ee693569e670cc20ce4617c2c9849254514dbdfb1030015a49ee72ee8.pt", "input_sha256": "05ade70ee900de950adaf23171001399a4d6cc88017bf23770bc74ef28c01d3b", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 384, 4, 176, 80 ], "stride": [ 21626880, 56320, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 86507520 }, "output_tensor": { "shape": [ 1, 768, 2, 176, 80 ], "stride": [ 21626880, 28160, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 86507520 }, "shape_parameters": { "input": [ 1, 384, 4, 176, 80 ], "weight": [ 768, 384, 3, 1, 1 ], "output": [ 1, 768, 2, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 49828331520, "bias_additions": 21626880, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 299.13599999999997, "dram_read_bytes": 397014784.0, "dram_write_bytes": 265400064.0, "dram_total_bytes": 662414848.0, "nominal_mac_tflops_per_s": 166.57417201540437, "nominal_mac_flops_per_dram_byte": 75.22224429365448, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_4273bf5ee693569e_v1", "ncu_metrics_sha256": "ccaeb2e4346e742db251add3dca72c0d6af65604ae2b424566babba0c6ad70a2", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_3d9ee58b5b1e2e00423a0623ac3521a587fe1568c25ec4741e586843ab78e00a.pt", "input_sha256": "31fe6110c4fdd22ab61ff0c539192126b4bc8da9a7355f9916c23de05f620d74", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 10560, 144 ], "stride": [ 144, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 3041280 }, "output_tensor": { "shape": [ 10560, 5120 ], "stride": [ 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 108134400 }, "shape_parameters": { "M": 10560, "N": 5120, "K": 144 }, "mac_flops": 15571353600, "bias_additions": 54067200, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 53.472, "dram_read_bytes": 4755456.0, "dram_write_bytes": 81849600.0, "dram_total_bytes": 86605056.0, "nominal_mac_tflops_per_s": 291.2057450628366, "nominal_mac_flops_per_dram_byte": 179.79728111947645, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_3d9ee58b5b1e2e00_v1", "ncu_metrics_sha256": "637613bd6b7dca12517fab6c0d238c2ec8cf93de069ea6470ca3db5871a719e2", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_1ae7dbe8166a0b6b9f2b1ce0eb32a55d5f296e65051ec6cd88db2f1140324de9.pt", "input_sha256": "85c0ba2d39aa4aa6ffbabaa237b9fa5d7721eefd855682c8cf482c661a5943d9", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 96, 3, 706, 322 ], "stride": [ 65471616, 681996, 227332, 322, 1 ], "dtype": "torch.float32", "logical_bytes": 261886464 }, "output_tensor": { "shape": [ 1, 96, 1, 704, 320 ], "stride": [ 21626880, 225280, 225280, 320, 1 ], "dtype": "torch.float32", "logical_bytes": 86507520 }, "shape_parameters": { "input": [ 1, 96, 3, 706, 322 ], "weight": [ 96, 96, 3, 3, 3 ], "output": [ 1, 96, 1, 704, 320 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 112113745920, "bias_additions": 21626880, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 1006.016, "dram_read_bytes": 698945024.0, "dram_write_bytes": 455985152.0, "dram_total_bytes": 1154930176.0, "nominal_mac_tflops_per_s": 111.44330300909726, "nominal_mac_flops_per_dram_byte": 97.07404676904035, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_1ae7dbe8166a0b6b_v1", "ncu_metrics_sha256": "aa0edd1e76279cf72710425be23ebfba1e4b4f7ffc6a94b0f49d12d954d51f8f", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_996e506f74aef428a26fb27992551214f90bb3cda65c48a58870bca06003c7a4.pt", "input_sha256": "a9625d3e84c044a16224205940b2918538dcbebc7b9e36a51af82b718a774552", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 192, 3, 354, 162 ], "stride": [ 33032448, 172044, 57348, 162, 1 ], "dtype": "torch.float32", "logical_bytes": 132129792 }, "output_tensor": { "shape": [ 1, 192, 1, 352, 160 ], "stride": [ 10813440, 56320, 56320, 160, 1 ], "dtype": "torch.float32", "logical_bytes": 43253760 }, "shape_parameters": { "input": [ 1, 192, 3, 354, 162 ], "weight": [ 192, 192, 3, 3, 3 ], "output": [ 1, 192, 1, 352, 160 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 112113745920, "bias_additions": 10813440, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 562.496, "dram_read_bytes": 361622784.0, "dram_write_bytes": 193293824.0, "dram_total_bytes": 554916608.0, "nominal_mac_tflops_per_s": 199.31474342928658, "nominal_mac_flops_per_dram_byte": 202.0371066637818, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_996e506f74aef428_v1", "ncu_metrics_sha256": "05a6544ddacff75d7aee5bdecf8ab706b391686960b1fb0fa93bc4dcf50575aa", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_e70c8df9d1c257ccdd2614ebd07d193e4fc0aa6634519de3420e63c6ac5517f2.pt", "input_sha256": "9b2cc0033aa13a80bc2dd24c544a6b1097e94f10f8c1f3748cbca111210d5fa1", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 384, 3, 178, 82 ], "stride": [ 16814592, 43788, 14596, 82, 1 ], "dtype": "torch.float32", "logical_bytes": 67258368 }, "output_tensor": { "shape": [ 1, 384, 1, 176, 80 ], "stride": [ 5406720, 14080, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 21626880 }, "shape_parameters": { "input": [ 1, 384, 3, 178, 82 ], "weight": [ 384, 384, 3, 3, 3 ], "output": [ 1, 384, 1, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 112113745920, "bias_additions": 5406720, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 6, "summed_profiled_kernel_us": 585.4079999999999, "dram_read_bytes": 305846272.0, "dram_write_bytes": 82012416.0, "dram_total_bytes": 387858688.0, "nominal_mac_tflops_per_s": 191.5138602820597, "nominal_mac_flops_per_dram_byte": 289.05823019748885, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_e70c8df9d1c257cc_v1", "ncu_metrics_sha256": "322ddd52edb945be7dee84324a3d329dc0d3ca65f88e55dcd3b9ff5a97134901", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_0e33e5013ffbea1c0222ff26c16fc5d6e1847abda1c87e5a62a6c61cc11e6ac5.pt", "input_sha256": "c09536053b980c5e962fadc017011cff0cd6c94409f5c2270984dfec83c493a3", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 96, 6, 706, 322 ], "stride": [ 130943232, 1363992, 227332, 322, 1 ], "dtype": "torch.float32", "logical_bytes": 523772928 }, "output_tensor": { "shape": [ 1, 3, 4, 704, 320 ], "stride": [ 2703360, 901120, 225280, 320, 1 ], "dtype": "torch.float32", "logical_bytes": 10813440 }, "shape_parameters": { "input": [ 1, 96, 6, 706, 322 ], "weight": [ 3, 96, 3, 3, 3 ], "output": [ 1, 3, 4, 704, 320 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 14014218240, "bias_additions": 2703360, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 1515.04, "dram_read_bytes": 1752498000.0, "dram_write_bytes": 519280896.0, "dram_total_bytes": 2271778896.0, "nominal_mac_tflops_per_s": 9.250064843172456, "nominal_mac_flops_per_dram_byte": 6.168830190594393, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_0e33e5013ffbea1c_v1", "ncu_metrics_sha256": "74bab34b00580ebd194e817a92d4213eee0d5e825ff1db2c9c4feb436049855b", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/dit_rank0_5a8bc19ef82213ac39d220cb7cfa551043aa800db8bbe80d36770b99bfe77eb2.pt", "input_sha256": "26ebef86e4633eac87d2d256ccdab318b493bdf4c53c9c855e742e4fe4f4f537", "stage": "dit", "op": "aten.linear.default", "input_tensor": { "shape": [ 1, 10560, 5120 ], "stride": [ 54067200, 5120, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 108134400 }, "output_tensor": { "shape": [ 1, 10560, 64 ], "stride": [ 675840, 64, 1 ], "dtype": "torch.bfloat16", "logical_bytes": 1351680 }, "shape_parameters": { "M": 10560, "N": 64, "K": 5120 }, "mac_flops": 6920601600, "bias_additions": 675840, "formula": "2*M*N*K (multiply and add each count one); bias additions listed separately", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 31.135999999999996, "dram_read_bytes": 108811264.0, "dram_write_bytes": 2653696.0, "dram_total_bytes": 111464960.0, "nominal_mac_tflops_per_s": 222.27009249743065, "nominal_mac_flops_per_dram_byte": 62.08768746698514, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_5a8bc19ef82213ac_v1", "ncu_metrics_sha256": "fc63f48c0abfa5decbde598374626ea692157d29e5189496d4224121807fd167", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_fd5ad4e08e7ff9d878c92949181669828a46f31589e1a281f0bf5b9440cf89dc.pt", "input_sha256": "9dd2254d46cabc12367da79ca54c5220c508380e8b7fd1b985050405f5f81ee3", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 384, 3, 88, 40 ], "stride": [ 4055040, 10560, 3520, 40, 1 ], "dtype": "torch.float32", "logical_bytes": 16220160 }, "output_tensor": { "shape": [ 1, 768, 1, 88, 40 ], "stride": [ 2703360, 3520, 3520, 40, 1 ], "dtype": "torch.float32", "logical_bytes": 10813440 }, "shape_parameters": { "input": [ 1, 384, 3, 88, 40 ], "weight": [ 768, 384, 3, 1, 1 ], "output": [ 1, 768, 1, 88, 40 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 6228541440, "bias_additions": 2703360, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 6, "summed_profiled_kernel_us": 74.752, "dram_read_bytes": 61822720.0, "dram_write_bytes": 4661504.0, "dram_total_bytes": 66484224.0, "nominal_mac_tflops_per_s": 83.32273972602741, "nominal_mac_flops_per_dram_byte": 93.68450235652897, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_fd5ad4e08e7ff9d8_v1", "ncu_metrics_sha256": "2c95712f9ee4b624884277c2fc5b24ed677fc2e372a01bee59bd317528f9aaf0", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_9a2a65226b52659e7d2f8d22162e2b2f0ce6792b117748f4cbfb770eda329377.pt", "input_sha256": "bccd4b59806daff52b45f9c87ba32579441de0ed0adb3dd99356b7ebbcc1f4c7", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 192, 2, 176, 80 ], "stride": [ 5406720, 14080, 2703360, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 21626880 }, "output_tensor": { "shape": [ 1, 384, 2, 176, 80 ], "stride": [ 10813440, 28160, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 43253760 }, "shape_parameters": { "input": [ 1, 192, 2, 176, 80 ], "weight": [ 384, 192, 1, 1, 1 ], "output": [ 1, 384, 2, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 4152360960, "bias_additions": 10813440, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 3, "summed_profiled_kernel_us": 80.96, "dram_read_bytes": 86948352.0, "dram_write_bytes": 41172480.0, "dram_total_bytes": 128120832.0, "nominal_mac_tflops_per_s": 51.28904347826087, "nominal_mac_flops_per_dram_byte": 32.4097252193929, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_9a2a65226b52659e_v1", "ncu_metrics_sha256": "d1057d2c0073e4116247d85f5b4bb8fffccac12d8e5f9ac20970696121831070", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_740d67537eaca4e114dcf7368a69708a8ec9c521e5f7e6aa76576d9b7fd3b90a.pt", "input_sha256": "d56deba2cf2a7667416ca96682c1bcff892969b687097e1b3ea3921ac1a1d6ac", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 192, 3, 178, 82 ], "stride": [ 8407296, 43788, 14596, 82, 1 ], "dtype": "torch.float32", "logical_bytes": 33629184 }, "output_tensor": { "shape": [ 1, 384, 1, 176, 80 ], "stride": [ 5406720, 14080, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 21626880 }, "shape_parameters": { "input": [ 1, 192, 3, 178, 82 ], "weight": [ 384, 192, 3, 3, 3 ], "output": [ 1, 384, 1, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 56056872960, "bias_additions": 5406720, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 6, "summed_profiled_kernel_us": 306.40000000000003, "dram_read_bytes": 151560960.0, "dram_write_bytes": 44080384.0, "dram_total_bytes": 195641344.0, "nominal_mac_tflops_per_s": 182.95324073107048, "nominal_mac_flops_per_dram_byte": 286.52876643497194, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_740d67537eaca4e1_v1", "ncu_metrics_sha256": "085a6ac91102daeaa91c746478b6dfdedd9a64785f2f13970191fb734e2884a6", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_9e24c4a75ab83dea1b42bc8322b881241b1395768618a954f5922c59f55707f2.pt", "input_sha256": "f8fb1466dc2a1a6e997b5eff656e0d892480f377234df948eb6b06009d0b88b2", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 16, 3, 90, 42 ], "stride": [ 181440, 11340, 3780, 42, 1 ], "dtype": "torch.float32", "logical_bytes": 725760 }, "output_tensor": { "shape": [ 1, 384, 1, 88, 40 ], "stride": [ 1351680, 3520, 3520, 40, 1 ], "dtype": "torch.float32", "logical_bytes": 5406720 }, "shape_parameters": { "input": [ 1, 16, 3, 90, 42 ], "weight": [ 384, 16, 3, 3, 3 ], "output": [ 1, 384, 1, 88, 40 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 1167851520, "bias_additions": 1351680, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 4, "summed_profiled_kernel_us": 42.367999999999995, "dram_read_bytes": 8282368.0, "dram_write_bytes": 0.0, "dram_total_bytes": 8282368.0, "nominal_mac_tflops_per_s": 27.564471299093658, "nominal_mac_flops_per_dram_byte": 141.00454362810248, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_9e24c4a75ab83dea_v1", "ncu_metrics_sha256": "d2acd52f6d045bdddb1869c7da2d5c76c226e8b79b9dc45ced407b606751b1f6", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_aa00674f1a72afcda9d547498144743d5db206c81319ed334873cdbea4359a23.pt", "input_sha256": "a2673addbb1775eb79e779d7635c4fbcf9dd34b0d4953bdf81e740bec14aa1ef", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 96, 3, 706, 322 ], "stride": [ 65471616, 681996, 227332, 322, 1 ], "dtype": "torch.float32", "logical_bytes": 261886464 }, "output_tensor": { "shape": [ 1, 3, 1, 704, 320 ], "stride": [ 675840, 225280, 225280, 320, 1 ], "dtype": "torch.float32", "logical_bytes": 2703360 }, "shape_parameters": { "input": [ 1, 96, 3, 706, 322 ], "weight": [ 3, 96, 3, 3, 3 ], "output": [ 1, 3, 1, 704, 320 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 3503554560, "bias_additions": 675840, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 5, "summed_profiled_kernel_us": 502.5599999999999, "dram_read_bytes": 545607680.0, "dram_write_bytes": 248518400.0, "dram_total_bytes": 794126080.0, "nominal_mac_tflops_per_s": 6.971415472779371, "nominal_mac_flops_per_dram_byte": 4.411836669562597, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_aa00674f1a72afcd_v1", "ncu_metrics_sha256": "eae2467302dd677db282998fdded6638d33796da99c1db10c99e5d176d649245", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_2ea2cf3e2f86821c20fcf4be36d4146f39935e68bdd45f4e49aebfbb5f64eb74.pt", "input_sha256": "85c1d12d0656fbd621625fe39cf33eef4823fe7d3338e672b380cbbe7e08c3bc", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 192, 1, 176, 80 ], "stride": [ 2703360, 14080, 2703360, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 10813440 }, "output_tensor": { "shape": [ 1, 384, 1, 176, 80 ], "stride": [ 5406720, 14080, 14080, 80, 1 ], "dtype": "torch.float32", "logical_bytes": 21626880 }, "shape_parameters": { "input": [ 1, 192, 1, 176, 80 ], "weight": [ 384, 192, 1, 1, 1 ], "output": [ 1, 384, 1, 176, 80 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 2076180480, "bias_additions": 5406720, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 2, "summed_profiled_kernel_us": 36.704, "dram_read_bytes": 32779008.0, "dram_write_bytes": 3951616.0, "dram_total_bytes": 36730624.0, "nominal_mac_tflops_per_s": 56.56551002615519, "nominal_mac_flops_per_dram_byte": 56.524508813136414, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_2ea2cf3e2f86821c_v1", "ncu_metrics_sha256": "1126a181566f3b0fb35d34a8a7bb43541344e7684ca954fc2d3d98fb2925609f", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/vae_rank0_b611cd0d7be4ea169fdc4990587d638a3a4d3346571857d9742cc12b6e402b9e.pt", "input_sha256": "28dd946d4a53d1263bf5d552d2aacaa22bc7162bea85950320f6d0e02d5750e7", "stage": "vae", "op": "aten.conv3d.default", "input_tensor": { "shape": [ 1, 16, 3, 88, 160 ], "stride": [ 675840, 42240, 14080, 160, 1 ], "dtype": "torch.float32", "logical_bytes": 2703360 }, "output_tensor": { "shape": [ 1, 16, 3, 88, 160 ], "stride": [ 675840, 42240, 14080, 160, 1 ], "dtype": "torch.float32", "logical_bytes": 2703360 }, "shape_parameters": { "input": [ 1, 16, 3, 88, 160 ], "weight": [ 16, 16, 1, 1, 1 ], "output": [ 1, 16, 3, 88, 160 ], "stride": [ 1, 1, 1 ], "padding": [ 0, 0, 0 ], "dilation": [ 1, 1, 1 ], "groups": 1 }, "mac_flops": 21626880, "bias_additions": 675840, "formula": "2*output_elements*(Cin/groups)*Kd*Kh*Kw; padding already materialized in saved input, implicit padding zero; bias additions separate", "profiled_kernel_count": 2, "summed_profiled_kernel_us": 10.495999999999999, "dram_read_bytes": 5448448.0, "dram_write_bytes": 0.0, "dram_total_bytes": 5448448.0, "nominal_mac_tflops_per_s": 2.0604878048780493, "nominal_mac_flops_per_dram_byte": 3.9693652210684585, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_b611cd0d7be4ea16_v1", "ncu_metrics_sha256": "eca989236210ce6700fea8fb13100fac5b0c0c8624f6fea4d9452260ffc2bd23", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/output_rank0_59ef4c5b540ee17f324e99f46c07bf2544db0e771c768ae3ce80ed65fde6f8a9.pt", "input_sha256": "80d26723a59d26f80a9dfde0de9bf8e28bb650baf110fa77f33200b5bb9e4198", "stage": "output", "op": "aten.to.dtype", "input_tensor": { "shape": [ 1, 3, 9, 704, 1280 ], "stride": [ 24330240, 8110080, 901120, 1280, 1 ], "dtype": "torch.float32", "logical_bytes": 97320960 }, "output_tensor": { "shape": [ 1, 3, 9, 704, 1280 ], "stride": [ 24330240, 8110080, 901120, 1280, 1 ], "dtype": "torch.uint8", "logical_bytes": 24330240 }, "shape_parameters": {}, "mac_flops": null, "bias_additions": 0, "formula": "dtype conversion: no useful floating-point FLOP rate assigned", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 64.63999999999999, "dram_read_bytes": 97357056.0, "dram_write_bytes": 18819328.0, "dram_total_bytes": 116176384.0, "nominal_mac_tflops_per_s": null, "nominal_mac_flops_per_dram_byte": null, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_59ef4c5b540ee17f_v1", "ncu_metrics_sha256": "90f390f74fcc3c51ff2b918a3a653cf5596f4df073c878673bc2b2c0b5bf80ac", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] }, { "input": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/missing_operator_704x1280_tp1sp4_v1/operators/output_rank0_99c188aeb0ac98c3c3458739f715e04b7d4938e0324397737e5d4b2840c5a8bd.pt", "input_sha256": "ff3ca616760a5d842212457937cd00ab117a7e2049cc892ea9847400317b91f6", "stage": "output", "op": "aten.to.dtype", "input_tensor": { "shape": [ 1, 3, 12, 704, 1280 ], "stride": [ 32440320, 10813440, 901120, 1280, 1 ], "dtype": "torch.float32", "logical_bytes": 129761280 }, "output_tensor": { "shape": [ 1, 3, 12, 704, 1280 ], "stride": [ 32440320, 10813440, 901120, 1280, 1 ], "dtype": "torch.uint8", "logical_bytes": 32440320 }, "shape_parameters": {}, "mac_flops": null, "bias_additions": 0, "formula": "dtype conversion: no useful floating-point FLOP rate assigned", "profiled_kernel_count": 1, "summed_profiled_kernel_us": 84.224, "dram_read_bytes": 129798399.99999999, "dram_write_bytes": 27542784.0, "dram_total_bytes": 157341184.0, "nominal_mac_tflops_per_s": null, "nominal_mac_flops_per_dram_byte": null, "ncu_output": "/data/zhoutaichang/feature/Lingbot_world_realtime/runs/analysis_report_20260908/raw/ncu_missing704_99c188aeb0ac98c3_v1", "ncu_metrics_sha256": "dfb3824f64d854f49c2cfa530488add9e055291731235921024a3b93d22bed08", "tensor_policy": { "matmul_precision": "highest", "matmul_allow_tf32": false, "cudnn_allow_tf32": true, "cudnn_benchmark": false, "cudnn_deterministic": false }, "scheduler_note": [] } ], "scope": "28 additional saved operators only. FLOPs count nominal dense multiply-add work including explicitly materialized input padding; not executed instruction FLOPs. Rate divides by summed NCU kernel duration (including layout and bias kernels), not unprofiled operator latency. DRAM traffic sums each profiled kernel, reflects replay cache/writeback behavior, not required logical bytes. Bias additions separate; dtype conversion FLOPs undefined. No local bandwidth ceiling or full-model MFU claimed; no mapping from this isolated op to all integrated calls implied." }