{
  "schema_version": 1,
  "protocol": {
    "scope": "complete mixer preparation, recurrence, input and parameter gradients",
    "dtype": "BF16 autocast and FP32 recurrent carry",
    "input_gradient": true,
    "order": "AB/BA counterbalanced",
    "compile_excluded": true,
    "baseline_policy": "strongest previously calibrated legal RG-LRU backend per shape",
    "samu_policy": "strongest previously calibrated legal SAMU backend per shape",
    "resets": "independent sequences reset at first token; no internal packed reset"
  },
  "environment": {
    "torch": "2.1.2+cu118",
    "cuda": "11.8",
    "gpu": "NVIDIA H800 PCIe"
  },
  "rows": [
    {
      "shape": {
        "batch": 1,
        "length": 8192,
        "width": 1024
      },
      "architecture": "rglru",
      "backend": "fused_grouped_prefix32_hierarchical_chunk16",
      "controller": null,
      "forward": {
        "samples_ms": [
          0.619102269411087,
          0.6021782755851746,
          0.562082976102829,
          0.5597881972789764,
          0.5600377917289734,
          0.5622692406177521,
          0.5523748695850372,
          0.5658641457557678,
          0.5492754280567169,
          0.5487613379955292
        ],
        "median_ms": 0.5610603839159012,
        "mean_ms": 0.5681734532117844,
        "std_ms": 0.02224279911379964,
        "minimum_ms": 0.5487613379955292,
        "maximum_ms": 0.619102269411087
      },
      "backward": {
        "samples_ms": [
          1.0051541030406952,
          0.9819567203521729,
          0.9830482304096222,
          0.9888187050819397,
          0.982724130153656,
          0.9694695472717285,
          0.9577721357345581,
          0.9761527180671692,
          0.9677782654762268,
          0.9647868573665619
        ],
        "median_ms": 0.979054719209671,
        "mean_ms": 0.977766141295433,
        "std_ms": 0.012968110990179459,
        "minimum_ms": 0.9577721357345581,
        "maximum_ms": 1.0051541030406952
      },
      "forward_backward": {
        "samples_ms": [
          1.600407063961029,
          1.6071051359176636,
          1.6539357602596283,
          1.6146935522556305,
          1.6069449484348297,
          1.5810951590538025,
          1.5655048191547394,
          1.6330108046531677,
          1.58676877617836,
          1.5816502273082733
        ],
        "median_ms": 1.6036760061979294,
        "mean_ms": 1.6031116247177124,
        "std_ms": 0.02505160428660762,
        "minimum_ms": 1.5655048191547394,
        "maximum_ms": 1.6539357602596283
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 0,
          "forward_ms": [
            0.619102269411087,
            0.6021782755851746,
            0.562082976102829,
            0.5597881972789764,
            0.5600377917289734
          ],
          "backward_ms": [
            1.0051541030406952,
            0.9819567203521729,
            0.9830482304096222,
            0.9888187050819397,
            0.982724130153656
          ],
          "forward_backward_ms": [
            1.600407063961029,
            1.6071051359176636,
            1.6539357602596283,
            1.6146935522556305,
            1.6069449484348297
          ]
        },
        {
          "round": 1,
          "order_index": 1,
          "forward_ms": [
            0.5622692406177521,
            0.5523748695850372,
            0.5658641457557678,
            0.5492754280567169,
            0.5487613379955292
          ],
          "backward_ms": [
            0.9694695472717285,
            0.9577721357345581,
            0.9761527180671692,
            0.9677782654762268,
            0.9647868573665619
          ],
          "forward_backward_ms": [
            1.5810951590538025,
            1.5655048191547394,
            1.6330108046531677,
            1.58676877617836,
            1.5816502273082733
          ]
        }
      ],
      "peak_allocated_delta_bytes": 193410048,
      "peak_reserved_delta_bytes": 272629760,
      "peak_allocated_bytes": 294638080,
      "peak_reserved_bytes": 375390208,
      "baseline_allocated_bytes": 101228032,
      "baseline_reserved_bytes": 102760448,
      "parameters": 134144,
      "parameter_bytes": 536576,
      "dynamic_control_values_per_token": 2048,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 8192,
        "width": 1024
      },
      "architecture": "samu",
      "backend": "fused_output_fused_write_shared_sfu_grouped_prefix64_hierarchical_chunk32",
      "controller": "fp32_fused_coords",
      "forward": {
        "samples_ms": [
          0.8977875113487244,
          0.9003616869449615,
          0.8605122566223145,
          0.8677802979946136,
          0.8776746690273285,
          0.8482486009597778,
          0.8236952126026154,
          0.828232616186142,
          0.8277110755443573,
          0.834248960018158
        ],
        "median_ms": 0.8543804287910461,
        "mean_ms": 0.8566252887248993,
        "std_ms": 0.027362848423410278,
        "minimum_ms": 0.8236952126026154,
        "maximum_ms": 0.9003616869449615
      },
      "backward": {
        "samples_ms": [
          1.1330805718898773,
          1.1113286018371582,
          1.123320311307907,
          1.1089593172073364,
          1.1703334748744965,
          6.927937269210815,
          1.1605210602283478,
          1.1214092373847961,
          1.1219456791877747,
          1.0889805853366852
        ],
        "median_ms": 1.1226329952478409,
        "mean_ms": 1.7067816108465195,
        "std_ms": 1.7405324283442962,
        "minimum_ms": 1.0889805853366852,
        "maximum_ms": 6.927937269210815
      },
      "forward_backward": {
        "samples_ms": [
          2.0499490201473236,
          2.0485147833824158,
          2.001706510782242,
          2.0784102380275726,
          2.0475871860980988,
          2.0267926156520844,
          2.015169709920883,
          2.036750316619873,
          2.077259123325348,
          2.0110569894313812
        ],
        "median_ms": 2.042168751358986,
        "mean_ms": 2.0393196493387222,
        "std_ms": 0.024887020197197626,
        "minimum_ms": 2.001706510782242,
        "maximum_ms": 2.0784102380275726
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 1,
          "forward_ms": [
            0.8977875113487244,
            0.9003616869449615,
            0.8605122566223145,
            0.8677802979946136,
            0.8776746690273285
          ],
          "backward_ms": [
            1.1330805718898773,
            1.1113286018371582,
            1.123320311307907,
            1.1089593172073364,
            1.1703334748744965
          ],
          "forward_backward_ms": [
            2.0499490201473236,
            2.0485147833824158,
            2.001706510782242,
            2.0784102380275726,
            2.0475871860980988
          ]
        },
        {
          "round": 1,
          "order_index": 0,
          "forward_ms": [
            0.8482486009597778,
            0.8236952126026154,
            0.828232616186142,
            0.8277110755443573,
            0.834248960018158
          ],
          "backward_ms": [
            6.927937269210815,
            1.1605210602283478,
            1.1214092373847961,
            1.1219456791877747,
            1.0889805853366852
          ],
          "forward_backward_ms": [
            2.0267926156520844,
            2.015169709920883,
            2.036750316619873,
            2.077259123325348,
            2.0110569894313812
          ]
        }
      ],
      "peak_allocated_delta_bytes": 151240192,
      "peak_reserved_delta_bytes": 155189248,
      "peak_allocated_bytes": 252990976,
      "peak_reserved_bytes": 257949696,
      "baseline_allocated_bytes": 101750784,
      "baseline_reserved_bytes": 102760448,
      "parameters": 3076,
      "parameter_bytes": 12304,
      "dynamic_control_values_per_token": 2,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 32768,
        "width": 1024
      },
      "architecture": "rglru",
      "backend": "fused_grouped_prefix32_hierarchical_chunk16",
      "controller": null,
      "forward": {
        "samples_ms": [
          1.6945600509643555,
          1.6846656799316406,
          1.6853995621204376,
          1.690242439508438,
          1.6876570880413055,
          1.6996599733829498,
          1.695837825536728,
          1.6842745244503021,
          1.6848817467689514,
          1.6896426677703857
        ],
        "median_ms": 1.6886498779058456,
        "mean_ms": 1.6896821558475494,
        "std_ms": 0.005113231794747119,
        "minimum_ms": 1.6842745244503021,
        "maximum_ms": 1.6996599733829498
      },
      "backward": {
        "samples_ms": [
          3.075346350669861,
          3.0732937157154083,
          3.069743514060974,
          3.078687936067581,
          3.085114061832428,
          3.0810460448265076,
          3.0779801309108734,
          3.0757635831832886,
          3.077305853366852,
          3.077518194913864
        ],
        "median_ms": 3.077412024140358,
        "mean_ms": 3.077179938554764,
        "std_ms": 0.003958698907195855,
        "minimum_ms": 3.069743514060974,
        "maximum_ms": 3.085114061832428
      },
      "forward_backward": {
        "samples_ms": [
          6.919998675584793,
          4.829574376344681,
          4.829678684473038,
          4.854585975408554,
          4.962272942066193,
          4.850246012210846,
          4.842046648263931,
          4.8285722732543945,
          4.829026758670807,
          4.846550524234772
        ],
        "median_ms": 4.8442985862493515,
        "mean_ms": 5.059255287051201,
        "std_ms": 0.6214054563266509,
        "minimum_ms": 4.8285722732543945,
        "maximum_ms": 6.919998675584793
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 1,
          "forward_ms": [
            1.6945600509643555,
            1.6846656799316406,
            1.6853995621204376,
            1.690242439508438,
            1.6876570880413055
          ],
          "backward_ms": [
            3.075346350669861,
            3.0732937157154083,
            3.069743514060974,
            3.078687936067581,
            3.085114061832428
          ],
          "forward_backward_ms": [
            6.919998675584793,
            4.829574376344681,
            4.829678684473038,
            4.854585975408554,
            4.962272942066193
          ]
        },
        {
          "round": 1,
          "order_index": 0,
          "forward_ms": [
            1.6996599733829498,
            1.695837825536728,
            1.6842745244503021,
            1.6848817467689514,
            1.6896426677703857
          ],
          "backward_ms": [
            3.0810460448265076,
            3.0779801309108734,
            3.0757635831832886,
            3.077305853366852,
            3.077518194913864
          ],
          "forward_backward_ms": [
            4.850246012210846,
            4.842046648263931,
            4.8285722732543945,
            4.829026758670807,
            4.846550524234772
          ]
        }
      ],
      "peak_allocated_delta_bytes": 789591552,
      "peak_reserved_delta_bytes": 796917760,
      "peak_allocated_bytes": 991482880,
      "peak_reserved_bytes": 1000341504,
      "baseline_allocated_bytes": 201891328,
      "baseline_reserved_bytes": 203423744,
      "parameters": 134144,
      "parameter_bytes": 536576,
      "dynamic_control_values_per_token": 2048,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 32768,
        "width": 1024
      },
      "architecture": "samu",
      "backend": "fused_output_fused_write_shared_sfu_grouped_prefix64_hierarchical_chunk32",
      "controller": "fp32_fused_coords",
      "forward": {
        "samples_ms": [
          0.9703077375888824,
          0.9042061865329742,
          0.8833073079586029,
          0.888437032699585,
          0.8989609777927399,
          0.8740685880184174,
          0.8772425353527069,
          0.902228057384491,
          0.9026601910591125,
          0.8751340210437775
        ],
        "median_ms": 0.8936990052461624,
        "mean_ms": 0.897655263543129,
        "std_ms": 0.026720811958442295,
        "minimum_ms": 0.8740685880184174,
        "maximum_ms": 0.9703077375888824
      },
      "backward": {
        "samples_ms": [
          2.051983028650284,
          2.069879323244095,
          2.0554102957248688,
          2.046409994363785,
          2.0464174449443817,
          2.045895904302597,
          2.0662955939769745,
          2.0527131855487823,
          2.074003219604492,
          2.0466819405555725
        ],
        "median_ms": 2.052348107099533,
        "mean_ms": 2.0555689930915833,
        "std_ms": 0.010103923129118538,
        "minimum_ms": 2.045895904302597,
        "maximum_ms": 2.074003219604492
      },
      "forward_backward": {
        "samples_ms": [
          3.0341893434524536,
          3.046572208404541,
          3.0417032539844513,
          3.0197985470294952,
          3.052830696105957,
          3.0572041869163513,
          3.049209713935852,
          3.05301696062088,
          3.3432766795158386,
          3.069218248128891
        ],
        "median_ms": 3.0510202050209045,
        "mean_ms": 3.076701983809471,
        "std_ms": 0.08975685073577983,
        "minimum_ms": 3.0197985470294952,
        "maximum_ms": 3.3432766795158386
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 0,
          "forward_ms": [
            0.9703077375888824,
            0.9042061865329742,
            0.8833073079586029,
            0.888437032699585,
            0.8989609777927399
          ],
          "backward_ms": [
            2.051983028650284,
            2.069879323244095,
            2.0554102957248688,
            2.046409994363785,
            2.0464174449443817
          ],
          "forward_backward_ms": [
            3.0341893434524536,
            3.046572208404541,
            3.0417032539844513,
            3.0197985470294952,
            3.052830696105957
          ]
        },
        {
          "round": 1,
          "order_index": 1,
          "forward_ms": [
            0.8740685880184174,
            0.8772425353527069,
            0.902228057384491,
            0.9026601910591125,
            0.8751340210437775
          ],
          "backward_ms": [
            2.045895904302597,
            2.0662955939769745,
            2.0527131855487823,
            2.074003219604492,
            2.0466819405555725
          ],
          "forward_backward_ms": [
            3.0572041869163513,
            3.049209713935852,
            3.05301696062088,
            3.3432766795158386,
            3.069218248128891
          ]
        }
      ],
      "peak_allocated_delta_bytes": 604913152,
      "peak_reserved_delta_bytes": 673185792,
      "peak_allocated_bytes": 807327232,
      "peak_reserved_bytes": 876609536,
      "baseline_allocated_bytes": 202414080,
      "baseline_reserved_bytes": 203423744,
      "parameters": 3076,
      "parameter_bytes": 12304,
      "dynamic_control_values_per_token": 2,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 65536,
        "width": 1024
      },
      "architecture": "rglru",
      "backend": "fused_grouped_prefix32_hierarchical_chunk16",
      "controller": null,
      "forward": {
        "samples_ms": [
          3.1721219420433044,
          3.1619668006896973,
          3.1569115817546844,
          3.157738596200943,
          3.1581223011016846,
          3.1683817505836487,
          3.149278461933136,
          3.154739737510681,
          3.152359277009964,
          3.147970885038376
        ],
        "median_ms": 3.1573250889778137,
        "mean_ms": 3.157959133386612,
        "std_ms": 0.007370835661331802,
        "minimum_ms": 3.147970885038376,
        "maximum_ms": 3.1721219420433044
      },
      "backward": {
        "samples_ms": [
          5.878131836652756,
          5.8758072555065155,
          5.88347390294075,
          5.876760929822922,
          5.888588726520538,
          5.8945901691913605,
          5.892902612686157,
          5.8878399431705475,
          5.890857428312302,
          5.889147520065308
        ],
        "median_ms": 5.888214334845543,
        "mean_ms": 5.885810032486916,
        "std_ms": 0.00649773960057475,
        "minimum_ms": 5.8758072555065155,
        "maximum_ms": 5.8945901691913605
      },
      "forward_backward": {
        "samples_ms": [
          9.106967598199844,
          9.107712656259537,
          9.136814624071121,
          9.142652153968811,
          9.120255708694458,
          9.135577827692032,
          9.115461260080338,
          9.10554826259613,
          9.107507765293121,
          9.098947048187256
        ],
        "median_ms": 9.111586958169937,
        "mean_ms": 9.117744490504265,
        "std_ms": 0.014619406378826076,
        "minimum_ms": 9.098947048187256,
        "maximum_ms": 9.142652153968811
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 0,
          "forward_ms": [
            3.1721219420433044,
            3.1619668006896973,
            3.1569115817546844,
            3.157738596200943,
            3.1581223011016846
          ],
          "backward_ms": [
            5.878131836652756,
            5.8758072555065155,
            5.88347390294075,
            5.876760929822922,
            5.888588726520538
          ],
          "forward_backward_ms": [
            9.106967598199844,
            9.107712656259537,
            9.136814624071121,
            9.142652153968811,
            9.120255708694458
          ]
        },
        {
          "round": 1,
          "order_index": 1,
          "forward_ms": [
            3.1683817505836487,
            3.149278461933136,
            3.154739737510681,
            3.152359277009964,
            3.147970885038376
          ],
          "backward_ms": [
            5.8945901691913605,
            5.892902612686157,
            5.8878399431705475,
            5.890857428312302,
            5.889147520065308
          ],
          "forward_backward_ms": [
            9.135577827692032,
            9.115461260080338,
            9.10554826259613,
            9.107507765293121,
            9.098947048187256
          ]
        }
      ],
      "peak_allocated_delta_bytes": 1578907136,
      "peak_reserved_delta_bytes": 1579155456,
      "peak_allocated_bytes": 1915016192,
      "peak_reserved_bytes": 1916796928,
      "baseline_allocated_bytes": 336109056,
      "baseline_reserved_bytes": 337641472,
      "parameters": 134144,
      "parameter_bytes": 536576,
      "dynamic_control_values_per_token": 2048,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 65536,
        "width": 1024
      },
      "architecture": "samu",
      "backend": "fused_output_fused_write_shared_sfu_grouped_prefix64_hierarchical_chunk32",
      "controller": "fp32_fused_coords",
      "forward": {
        "samples_ms": [
          1.3426020741462708,
          1.3428069651126862,
          1.3662204146385193,
          1.3196654617786407,
          1.360628753900528,
          1.3418234884738922,
          1.3256184756755829,
          1.3261735439300537,
          1.324877142906189,
          1.3015680015087128
        ],
        "median_ms": 1.333998516201973,
        "mean_ms": 1.3351984322071075,
        "std_ms": 0.01848700067768149,
        "minimum_ms": 1.3015680015087128,
        "maximum_ms": 1.3662204146385193
      },
      "backward": {
        "samples_ms": [
          3.8796737790107727,
          3.8156546652317047,
          3.8766153156757355,
          3.815535455942154,
          3.83077934384346,
          3.8307905197143555,
          3.8740746676921844,
          3.8700103759765625,
          3.8322247564792633,
          3.811948001384735
        ],
        "median_ms": 3.8315076380968094,
        "mean_ms": 3.8437306880950928,
        "std_ms": 0.026543603178794718,
        "minimum_ms": 3.811948001384735,
        "maximum_ms": 3.8796737790107727
      },
      "forward_backward": {
        "samples_ms": [
          5.245201289653778,
          5.234971642494202,
          5.22313266992569,
          5.175970494747162,
          5.169250071048737,
          5.187083035707474,
          5.241919308900833,
          5.243472754955292,
          5.209174007177353,
          5.192488431930542
        ],
        "median_ms": 5.216153338551521,
        "mean_ms": 5.212266370654106,
        "std_ms": 0.027882203028411507,
        "minimum_ms": 5.169250071048737,
        "maximum_ms": 5.245201289653778
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 1,
          "forward_ms": [
            1.3426020741462708,
            1.3428069651126862,
            1.3662204146385193,
            1.3196654617786407,
            1.360628753900528
          ],
          "backward_ms": [
            3.8796737790107727,
            3.8156546652317047,
            3.8766153156757355,
            3.815535455942154,
            3.83077934384346
          ],
          "forward_backward_ms": [
            5.245201289653778,
            5.234971642494202,
            5.22313266992569,
            5.175970494747162,
            5.169250071048737
          ]
        },
        {
          "round": 1,
          "order_index": 0,
          "forward_ms": [
            1.3418234884738922,
            1.3256184756755829,
            1.3261735439300537,
            1.324877142906189,
            1.3015680015087128
          ],
          "backward_ms": [
            3.8307905197143555,
            3.8740746676921844,
            3.8700103759765625,
            3.8322247564792633,
            3.811948001384735
          ],
          "forward_backward_ms": [
            5.187083035707474,
            5.241919308900833,
            5.243472754955292,
            5.209174007177353,
            5.192488431930542
          ]
        }
      ],
      "peak_allocated_delta_bytes": 1209810432,
      "peak_reserved_delta_bytes": 1346371584,
      "peak_allocated_bytes": 1546442240,
      "peak_reserved_bytes": 1684013056,
      "baseline_allocated_bytes": 336631808,
      "baseline_reserved_bytes": 337641472,
      "parameters": 3076,
      "parameter_bytes": 12304,
      "dynamic_control_values_per_token": 2,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 131072,
        "width": 1024
      },
      "architecture": "rglru",
      "backend": "fused_grouped_prefix32_hierarchical_chunk16",
      "controller": null,
      "forward": {
        "samples_ms": [
          7.324062287807465,
          6.1409808695316315,
          6.101291626691818,
          6.102666258811951,
          6.106160581111908,
          6.144531071186066,
          6.121303886175156,
          6.127659231424332,
          6.116405129432678,
          6.114482879638672
        ],
        "median_ms": 6.118854507803917,
        "mean_ms": 6.239954382181168,
        "std_ms": 0.3616439052718223,
        "minimum_ms": 6.101291626691818,
        "maximum_ms": 7.324062287807465
      },
      "backward": {
        "samples_ms": [
          11.526592075824738,
          11.523239314556122,
          11.510655283927917,
          11.51467114686966,
          11.523108929395676,
          11.531662195920944,
          11.533614248037338,
          11.526327580213547,
          11.517435312271118,
          11.515967547893524
        ],
        "median_ms": 11.523174121975899,
        "mean_ms": 11.522327363491058,
        "std_ms": 0.007137221996229179,
        "minimum_ms": 11.510655283927917,
        "maximum_ms": 11.533614248037338
      },
      "forward_backward": {
        "samples_ms": [
          17.793815582990646,
          17.736852169036865,
          17.733514308929443,
          17.731565982103348,
          17.73640885949135,
          17.729133367538452,
          17.747819423675537,
          17.725195735692978,
          17.723672091960907,
          17.714228481054306
        ],
        "median_ms": 17.732540145516396,
        "mean_ms": 17.737220600247383,
        "std_ms": 0.02069790670161889,
        "minimum_ms": 17.714228481054306,
        "maximum_ms": 17.793815582990646
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 1,
          "forward_ms": [
            7.324062287807465,
            6.1409808695316315,
            6.101291626691818,
            6.102666258811951,
            6.106160581111908
          ],
          "backward_ms": [
            11.526592075824738,
            11.523239314556122,
            11.510655283927917,
            11.51467114686966,
            11.523108929395676
          ],
          "forward_backward_ms": [
            17.793815582990646,
            17.736852169036865,
            17.733514308929443,
            17.731565982103348,
            17.73640885949135
          ]
        },
        {
          "round": 1,
          "order_index": 0,
          "forward_ms": [
            6.144531071186066,
            6.121303886175156,
            6.127659231424332,
            6.116405129432678,
            6.114482879638672
          ],
          "backward_ms": [
            11.531662195920944,
            11.533614248037338,
            11.526327580213547,
            11.517435312271118,
            11.515967547893524
          ],
          "forward_backward_ms": [
            17.729133367538452,
            17.747819423675537,
            17.725195735692978,
            17.723672091960907,
            17.714228481054306
          ]
        }
      ],
      "peak_allocated_delta_bytes": 3157538304,
      "peak_reserved_delta_bytes": 3156213760,
      "peak_allocated_bytes": 3762082816,
      "peak_reserved_bytes": 3762290688,
      "baseline_allocated_bytes": 604544512,
      "baseline_reserved_bytes": 606076928,
      "parameters": 134144,
      "parameter_bytes": 536576,
      "dynamic_control_values_per_token": 2048,
      "materialized_state_sized_transition": false
    },
    {
      "shape": {
        "batch": 1,
        "length": 131072,
        "width": 1024
      },
      "architecture": "samu",
      "backend": "fused_output_fused_write_shared_sfu_grouped_prefix64_hierarchical_chunk32",
      "controller": "fp32_fused_coords",
      "forward": {
        "samples_ms": [
          2.6107728481292725,
          2.5836192071437836,
          2.571612596511841,
          2.551998943090439,
          2.5486014783382416,
          2.5838054716587067,
          2.5471337139606476,
          2.5761276483535767,
          2.5488995015621185,
          2.5652609765529633
        ],
        "median_ms": 2.568436786532402,
        "mean_ms": 2.568783238530159,
        "std_ms": 0.01958656081753415,
        "minimum_ms": 2.5471337139606476,
        "maximum_ms": 2.6107728481292725
      },
      "backward": {
        "samples_ms": [
          7.477611303329468,
          7.464997470378876,
          7.45711475610733,
          7.449500262737274,
          7.456157356500626,
          7.458783686161041,
          7.472489029169083,
          7.471505552530289,
          7.470764219760895,
          7.460597902536392
        ],
        "median_ms": 7.462797686457634,
        "mean_ms": 7.463952153921127,
        "std_ms": 0.008475443678234915,
        "minimum_ms": 7.449500262737274,
        "maximum_ms": 7.477611303329468
      },
      "forward_backward": {
        "samples_ms": [
          10.057486593723297,
          10.039575397968292,
          10.0686214864254,
          10.035242885351181,
          10.04805788397789,
          10.082703083753586,
          10.061293840408325,
          10.059371590614319,
          10.072864592075348,
          10.064393281936646
        ],
        "median_ms": 10.060332715511322,
        "mean_ms": 10.058961063623428,
        "std_ms": 0.013945766988745893,
        "minimum_ms": 10.035242885351181,
        "maximum_ms": 10.082703083753586
      },
      "rounds": [
        {
          "round": 0,
          "order_index": 0,
          "forward_ms": [
            2.6107728481292725,
            2.5836192071437836,
            2.571612596511841,
            2.551998943090439,
            2.5486014783382416
          ],
          "backward_ms": [
            7.477611303329468,
            7.464997470378876,
            7.45711475610733,
            7.449500262737274,
            7.456157356500626
          ],
          "forward_backward_ms": [
            10.057486593723297,
            10.039575397968292,
            10.0686214864254,
            10.035242885351181,
            10.04805788397789
          ]
        },
        {
          "round": 1,
          "order_index": 1,
          "forward_ms": [
            2.5838054716587067,
            2.5471337139606476,
            2.5761276483535767,
            2.5488995015621185,
            2.5652609765529633
          ],
          "backward_ms": [
            7.458783686161041,
            7.472489029169083,
            7.471505552530289,
            7.470764219760895,
            7.460597902536392
          ],
          "forward_backward_ms": [
            10.082703083753586,
            10.061293840408325,
            10.059371590614319,
            10.072864592075348,
            10.064393281936646
          ]
        }
      ],
      "peak_allocated_delta_bytes": 2419604992,
      "peak_reserved_delta_bytes": 2692743168,
      "peak_allocated_bytes": 3024672256,
      "peak_reserved_bytes": 3298820096,
      "baseline_allocated_bytes": 605067264,
      "baseline_reserved_bytes": 606076928,
      "parameters": 3076,
      "parameter_bytes": 12304,
      "dynamic_control_values_per_token": 2,
      "materialized_state_sized_transition": false
    }
  ]
}
