[
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 3,
  "seconds": 2.975518668999939,
  "mem_increment_bytes": 1055813632,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1944.2398157894736,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.094811400999902
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 3,
  "seconds": 16.51123088899999,
  "mem_increment_bytes": 1055674368,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1944.2398157894736,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 19.66855086700002
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 3,
  "seconds": 0.1764341735839844,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1944.2398157894736,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.887881298000025
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 3,
  "seconds": 0.6322238159179687,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1941.0163768421053,
  "n_iter": 0,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 5.194867562000013
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 3,
  "seconds": 0.12555673980712892,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1944.2398157894736,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 8.633253276000005
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 3,
  "seconds": 1.164900390625,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1944.2398157894736,
  "n_iter": 0,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 8.893158813000014
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 3,
  "seconds": 15.087004164000064,
  "mem_increment_bytes": 1046466560,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8681431578948,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 18.21252077200006
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 3,
  "seconds": 95.37065717699988,
  "mem_increment_bytes": 1046892544,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8680757894737,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 97.48805622599957
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 3,
  "seconds": 0.81735986328125,
  "mem_increment_bytes": 973085184,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8681431578948,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.468924751000031
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 3,
  "seconds": 3.3597099609375,
  "mem_increment_bytes": 973085184,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.86848,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.969587861999912
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 3,
  "seconds": 0.9810997924804687,
  "mem_increment_bytes": 972809728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8681431578948,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.8790900649999
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 3,
  "seconds": 3.451068603515625,
  "mem_increment_bytes": 972809728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8685473684211,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 9.38738455500004
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 3,
  "seconds": 17.21629756499999,
  "mem_increment_bytes": 1057370112,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8674021052632,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 19.266664316999936
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 3,
  "seconds": 105.62245664500006,
  "mem_increment_bytes": 1074991104,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.867334736842,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 107.73289491599826
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 3,
  "seconds": 0.9226312255859375,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8672,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.558413864000045
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 3,
  "seconds": 3.980143798828125,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8672,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.541431452999973
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 3,
  "seconds": 1.0120848999023437,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8672,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.016265202999989
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 3,
  "seconds": 4.026781494140625,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8672,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 10.676123966000091
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 3,
  "seconds": 15.37059001099999,
  "mem_increment_bytes": 1055399936,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1865.5249094736841,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 20.174985281999966
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 3,
  "seconds": 2.9691665920000787,
  "mem_increment_bytes": 1055666176,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1895.8974789473684,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.125565556000083
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 3,
  "seconds": 0.650811279296875,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1895.8974789473684,
  "n_iter": 0,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 5.274442044000011
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 3,
  "seconds": 0.16675923156738282,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1895.8974789473684,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.866362419000097
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 3,
  "seconds": 0.5514884033203125,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1895.8974789473684,
  "n_iter": 0,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.518166220000012
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 3,
  "seconds": 0.11008096313476562,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1895.8974789473684,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.153474854000024
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 3,
  "seconds": 86.71334947299965,
  "mem_increment_bytes": 1046683648,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8695578947369,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 88.78655917699962
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 3,
  "seconds": 15.245295252000005,
  "mem_increment_bytes": 1046765568,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.86976,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 17.344309807000172
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 3,
  "seconds": 3.360926513671875,
  "mem_increment_bytes": 973085184,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.869827368421,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.96776804000001
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 3,
  "seconds": 0.7670299072265625,
  "mem_increment_bytes": 973085184,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8698947368421,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.456828932999997
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 3,
  "seconds": 3.452999267578125,
  "mem_increment_bytes": 972809728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.869827368421,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 9.621812940999916
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 3,
  "seconds": 0.8992675170898438,
  "mem_increment_bytes": 972809728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8698947368421,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.909762298000032
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 3,
  "seconds": 107.31864977399891,
  "mem_increment_bytes": 1075077120,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.86592,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 109.40246238399959
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 3,
  "seconds": 17.560029435999922,
  "mem_increment_bytes": 1060794368,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8651115789473,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 19.660202860000027
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 3,
  "seconds": 4.008860107421875,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8655157894736,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.633471060999909
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 3,
  "seconds": 0.9234613647460937,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8653810526316,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.543791704000114
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 3,
  "seconds": 4.00714990234375,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8653810526316,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 9.943488241000068
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 3,
  "seconds": 1.0138573608398438,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8653810526316,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.995354636000002
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 3,
  "seconds": 2.962422012000161,
  "mem_increment_bytes": 1055719424,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1892.3419010526316,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.081399546000057
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 3,
  "seconds": 15.9141066950001,
  "mem_increment_bytes": 1055830016,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1892.3419010526316,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 19.037536746000114
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 3,
  "seconds": 0.16624435424804687,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1896.1248705263158,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.845450567999933
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 3,
  "seconds": 0.6325147094726562,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1896.1248705263158,
  "n_iter": 0,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 5.2261445359999925
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 3,
  "seconds": 0.11023139190673828,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1892.3419010526316,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.191609740000104
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 3,
  "seconds": 0.551878662109375,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1892.3419010526316,
  "n_iter": 0,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.553103973999896
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 3,
  "seconds": 15.006952339999998,
  "mem_increment_bytes": 1046650880,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8663915789474,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 17.096422934999964
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 3,
  "seconds": 88.11016982799993,
  "mem_increment_bytes": 1046716416,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8663242105263,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 90.22467975699965
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 3,
  "seconds": 0.768071533203125,
  "mem_increment_bytes": 973085184,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8664589473684,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.458715459000132
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 3,
  "seconds": 3.358009033203125,
  "mem_increment_bytes": 973085184,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8665263157894,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.91415903699999
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 3,
  "seconds": 0.897775634765625,
  "mem_increment_bytes": 972809728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8664589473684,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 6.894819245000008
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 3,
  "seconds": 3.452896484375,
  "mem_increment_bytes": 972809728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8665936842106,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 9.379505590000008
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 3,
  "seconds": 17.204094941999983,
  "mem_increment_bytes": 1056890880,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8678736842105,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 19.303260964999936
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 3,
  "seconds": 105.96942648000004,
  "mem_increment_bytes": 1075027968,
  "mem_meter": "vmhwm",
  "inertia_per_point": 998.8679410526315,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 108.05118103199857
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 3,
  "seconds": 0.9220197143554687,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8692210526316,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 4.593852807999838
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 3,
  "seconds": 3.980067626953125,
  "mem_increment_bytes": 984478720,
  "mem_meter": "torch-pool",
  "inertia_per_point": 998.8691536842106,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.553677956000001
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 3,
  "seconds": 1.011931884765625,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8680757894737,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 7.027936015000023
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 3,
  "seconds": 4.00462109375,
  "mem_increment_bytes": 984200704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 998.8682778947368,
  "n_iter": 12,
  "tile_n": 11184,
  "tile_k": 3,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3-d1000",
   "n": 950000,
   "d": 1000,
   "k": 3,
   "seed": 101
  },
  "wall_seconds": 9.964689893000013
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 17,
  "seconds": 4.349631005999981,
  "mem_increment_bytes": 1591296000,
  "mem_meter": "vmhwm",
  "inertia_per_point": 291.1385255357143,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 17.00173712900005
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 17,
  "seconds": 54.53316425899993,
  "mem_increment_bytes": 1591586816,
  "mem_meter": "vmhwm",
  "inertia_per_point": 291.1385255357143,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 56.85424712200006
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 17,
  "seconds": 0.2503721008300781,
  "mem_increment_bytes": 1520446976,
  "mem_meter": "torch-pool",
  "inertia_per_point": 291.28440952380953,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.9610781689998475
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 17,
  "seconds": 1.8837750244140625,
  "mem_increment_bytes": 1520446976,
  "mem_meter": "torch-pool",
  "inertia_per_point": 294.728738452381,
  "n_iter": 0,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 5.50902087999998
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 17,
  "seconds": 0.24790220642089844,
  "mem_increment_bytes": 1520400896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 296.1140074404762,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.38652586399985
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 17,
  "seconds": 1.905406982421875,
  "mem_increment_bytes": 1520400896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 296.1140074404762,
  "n_iter": 0,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 5.99937418799982
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 17,
  "seconds": 8.273269051999932,
  "mem_increment_bytes": 2223915008,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.53465904761904,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 10.082271700000092
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 17,
  "seconds": 184.13271705900024,
  "mem_increment_bytes": 2224005120,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.53465904761904,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 185.90612734799925
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 17,
  "seconds": 0.54405322265625,
  "mem_increment_bytes": 2151689216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53468952380953,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.8185790690001795
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 17,
  "seconds": 6.2504208984375,
  "mem_increment_bytes": 2151689216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53468952380953,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.461486121999997
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 17,
  "seconds": 0.6059033203125,
  "mem_increment_bytes": 2150409728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.5346742857143,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.209150031999798
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 17,
  "seconds": 6.35212158203125,
  "mem_increment_bytes": 2150409728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53468952380953,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.866408058999923
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 17,
  "seconds": 11.498862169999484,
  "mem_increment_bytes": 2243452928,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.53334857142858,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 13.163621271000011
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 17,
  "seconds": 227.38568834499893,
  "mem_increment_bytes": 2242211840,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.5333180952381,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 229.22007014200062
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 17,
  "seconds": 0.7546067504882813,
  "mem_increment_bytes": 2151063040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53424761904762,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.958789832999855
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 17,
  "seconds": 8.13638330078125,
  "mem_increment_bytes": 2151063040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53276952380952,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 11.199938912000107
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 17,
  "seconds": 0.841440673828125,
  "mem_increment_bytes": 2150422528,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53321142857143,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.5055909269999574
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 17,
  "seconds": 8.2714716796875,
  "mem_increment_bytes": 2150422528,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.5331657142857,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 11.795710340999904
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 17,
  "seconds": 52.48785859899999,
  "mem_increment_bytes": 1591345152,
  "mem_meter": "vmhwm",
  "inertia_per_point": 290.64847583333335,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 54.836356655000145
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 17,
  "seconds": 4.404536916999859,
  "mem_increment_bytes": 1591427072,
  "mem_meter": "vmhwm",
  "inertia_per_point": 290.64847583333335,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 6.71605438000006
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 17,
  "seconds": 1.886625732421875,
  "mem_increment_bytes": 1520446976,
  "mem_meter": "torch-pool",
  "inertia_per_point": 293.6307726785714,
  "n_iter": 0,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 5.525493048000044
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 17,
  "seconds": 0.23774617004394533,
  "mem_increment_bytes": 1520446976,
  "mem_meter": "torch-pool",
  "inertia_per_point": 294.16653294642856,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.9689598540001043
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 17,
  "seconds": 1.9066767578125,
  "mem_increment_bytes": 1520400896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 292.63916452380954,
  "n_iter": 0,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 5.971368168000026
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 17,
  "seconds": 0.2518814697265625,
  "mem_increment_bytes": 1520400896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 292.63916452380954,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.389707745000123
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 17,
  "seconds": 197.47219092100022,
  "mem_increment_bytes": 2223878144,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.5338057142857,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 199.19151613899885
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 17,
  "seconds": 8.194737789999863,
  "mem_increment_bytes": 2223996928,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.53382095238095,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.946865433000085
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 17,
  "seconds": 6.261611328125,
  "mem_increment_bytes": 2151689216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53382095238095,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.359546096000031
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 17,
  "seconds": 0.5402562866210937,
  "mem_increment_bytes": 2151689216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.5338361904762,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.751096589999861
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 17,
  "seconds": 6.35602294921875,
  "mem_increment_bytes": 2150409728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53382095238095,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.88852642300003
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 17,
  "seconds": 0.606199951171875,
  "mem_increment_bytes": 2150409728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.5338361904762,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.26768057199979
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 17,
  "seconds": 237.0319241010002,
  "mem_increment_bytes": 2242191360,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.5332419047619,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 238.79858384499857
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 17,
  "seconds": 11.531370163000247,
  "mem_increment_bytes": 2243489792,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.53322666666668,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 13.182251393000115
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 17,
  "seconds": 8.12529931640625,
  "mem_increment_bytes": 2151063040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53468952380953,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 11.267519826000125
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 17,
  "seconds": 0.7539080810546875,
  "mem_increment_bytes": 2151063040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.5342780952381,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.931149282999968
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 17,
  "seconds": 8.26768359375,
  "mem_increment_bytes": 2150422528,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53350095238096,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 11.799397880000015
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 17,
  "seconds": 0.8395238647460938,
  "mem_increment_bytes": 2150422528,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53344,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.440711071000123
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 17,
  "seconds": 4.327957163000065,
  "mem_increment_bytes": 1591627776,
  "mem_meter": "vmhwm",
  "inertia_per_point": 293.54299208333333,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 6.6492484509999485
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 17,
  "seconds": 52.63581635200012,
  "mem_increment_bytes": 1591586816,
  "mem_meter": "vmhwm",
  "inertia_per_point": 293.54299208333333,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 54.96700078799995
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 17,
  "seconds": 0.23729458618164062,
  "mem_increment_bytes": 1520446976,
  "mem_meter": "torch-pool",
  "inertia_per_point": 294.0572778869048,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.9688748419998774
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 17,
  "seconds": 1.886396484375,
  "mem_increment_bytes": 1520446976,
  "mem_meter": "torch-pool",
  "inertia_per_point": 289.93418148809525,
  "n_iter": 0,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 5.513349159999962
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 17,
  "seconds": 0.24791859436035157,
  "mem_increment_bytes": 1520400896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 293.59192223214285,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.388332085000002
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 17,
  "seconds": 1.9068897705078125,
  "mem_increment_bytes": 1520400896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 291.2472575892857,
  "n_iter": 0,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 5.969563428999663
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 17,
  "seconds": 8.296059095000146,
  "mem_increment_bytes": 2223898624,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.53433904761906,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 10.075353695000103
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 17,
  "seconds": 188.26507254499847,
  "mem_increment_bytes": 2224119808,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.5344,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 189.97971052399953
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 17,
  "seconds": 0.5274923095703125,
  "mem_increment_bytes": 2151689216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53433904761906,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.7179117069999847
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 17,
  "seconds": 6.2451201171875,
  "mem_increment_bytes": 2151689216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53430857142857,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.35168859400028
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 17,
  "seconds": 0.5929300537109375,
  "mem_increment_bytes": 2150409728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.5343238095238,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.205480684999657
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 17,
  "seconds": 6.3580166015625,
  "mem_increment_bytes": 2150409728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53430857142857,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 9.872939494999628
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 17,
  "seconds": 11.651883257000009,
  "mem_increment_bytes": 2243461120,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.5337904761905,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 13.304393935000917
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 17,
  "seconds": 236.88950374100023,
  "mem_increment_bytes": 2242420736,
  "mem_meter": "vmhwm",
  "inertia_per_point": 176.5338057142857,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 238.62199552499987
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 17,
  "seconds": 0.7525069580078125,
  "mem_increment_bytes": 2151063040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53590857142856,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 3.9460003610001877
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 17,
  "seconds": 8.1277587890625,
  "mem_increment_bytes": 2151063040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 176.53487238095238,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 11.22616149400028
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 17,
  "seconds": 0.8384174194335937,
  "mem_increment_bytes": 2150422528,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53344,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 4.500474428999951
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 17,
  "seconds": 8.2674140625,
  "mem_increment_bytes": 2150422528,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 176.53461333333334,
  "n_iter": 12,
  "tile_n": 10965,
  "tile_k": 17,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k17-d180",
   "n": 2100000,
   "d": 180,
   "k": 17,
   "seed": 102
  },
  "wall_seconds": 11.766128776999722
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 100,
  "seconds": 13.436633625000013,
  "mem_increment_bytes": 309968896,
  "mem_meter": "vmhwm",
  "inertia_per_point": 31.063328184895834,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 16.024201276999975
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 100,
  "seconds": 63.43945243100006,
  "mem_increment_bytes": 456888320,
  "mem_meter": "vmhwm",
  "inertia_per_point": 31.74958788151042,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 65.42471313499937
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 100,
  "seconds": 0.39827972412109375,
  "mem_increment_bytes": 222941696,
  "mem_meter": "torch-pool",
  "inertia_per_point": 31.735357915364585,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.886927676000141
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 100,
  "seconds": 2.272931884765625,
  "mem_increment_bytes": 386633216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 31.01054686328125,
  "n_iter": 0,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 5.736794626000119
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 100,
  "seconds": 0.7000484008789063,
  "mem_increment_bytes": 222005248,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 31.66316535546875,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.1097989500003678
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 100,
  "seconds": 2.50408349609375,
  "mem_increment_bytes": 386633216,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 31.662306958333332,
  "n_iter": 0,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 4.830764750000071
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 100,
  "seconds": 5.702447092000057,
  "mem_increment_bytes": 457281536,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.912869333333333,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 7.315012671000204
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 100,
  "seconds": 139.14029287999983,
  "mem_increment_bytes": 457310208,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.912853333333334,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 140.80610429900116
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 100,
  "seconds": 0.28658743286132815,
  "mem_increment_bytes": 385880064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.912872,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.2304114250000566
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 100,
  "seconds": 4.6357919921875,
  "mem_increment_bytes": 385880064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.912866666666666,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 7.488563200000044
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 100,
  "seconds": 0.3232564392089844,
  "mem_increment_bytes": 384009728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.912864,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 2.1812513200002286
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 100,
  "seconds": 4.53928125,
  "mem_increment_bytes": 384009728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.912872,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 6.257022502000382
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 100,
  "seconds": 17.937159662999875,
  "mem_increment_bytes": 472109056,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.915538666666667,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 19.589511634000246
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 100,
  "seconds": 209.78637198999968,
  "mem_increment_bytes": 465625088,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.91245066666667,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 211.42467124799987
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 100,
  "seconds": 0.6746038208007813,
  "mem_increment_bytes": 385893888,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.917096,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.643276788000094
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 100,
  "seconds": 6.88710888671875,
  "mem_increment_bytes": 386633216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.917864,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 9.753177751000294
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 100,
  "seconds": 1.0190672607421876,
  "mem_increment_bytes": 384023552,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.918442666666667,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 2.868675277999955
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 100,
  "seconds": 7.0250927734375,
  "mem_increment_bytes": 386633216,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.918888,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 8.807021895000162
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 100,
  "seconds": 66.08037796700046,
  "mem_increment_bytes": 457121792,
  "mem_meter": "vmhwm",
  "inertia_per_point": 31.60204926171875,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 68.0449474420002
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 100,
  "seconds": 13.28909882400012,
  "mem_increment_bytes": 310063104,
  "mem_meter": "vmhwm",
  "inertia_per_point": 31.60204926171875,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 15.260966484000164
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 100,
  "seconds": 2.2706953125,
  "mem_increment_bytes": 386633216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 31.530649171875,
  "n_iter": 0,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 5.653799335000258
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 100,
  "seconds": 0.3981875305175781,
  "mem_increment_bytes": 222941696,
  "mem_meter": "torch-pool",
  "inertia_per_point": 31.455729299479167,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.9051284199999827
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 100,
  "seconds": 2.497418212890625,
  "mem_increment_bytes": 386633216,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 31.5677453359375,
  "n_iter": 0,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 4.8384794920002605
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 100,
  "seconds": 0.7001446533203125,
  "mem_increment_bytes": 222005248,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 31.28382440234375,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.1046472619996166
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 100,
  "seconds": 140.3889458930007,
  "mem_increment_bytes": 457453568,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.920504,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 141.99831793200065
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 100,
  "seconds": 5.650380850000147,
  "mem_increment_bytes": 457580544,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.920498666666667,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 7.244110814000123
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 100,
  "seconds": 4.63797509765625,
  "mem_increment_bytes": 385880064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.920506666666668,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 7.468085213999984
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 100,
  "seconds": 0.28554840087890626,
  "mem_increment_bytes": 385880064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.920490666666666,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.242343500000061
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 100,
  "seconds": 4.53791845703125,
  "mem_increment_bytes": 384009728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.920498666666667,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 6.304415643000084
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 100,
  "seconds": 0.32403054809570314,
  "mem_increment_bytes": 384009728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.92048,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 2.1812562129998696
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 100,
  "seconds": 198.32619454999985,
  "mem_increment_bytes": 465428480,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.917082666666666,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 199.91774939899915
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 100,
  "seconds": 18.187127473000146,
  "mem_increment_bytes": 471896064,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.917085333333333,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 19.790152029999717
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 100,
  "seconds": 6.88508935546875,
  "mem_increment_bytes": 386633216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.917496,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 9.706804222999835
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 100,
  "seconds": 0.6709290771484375,
  "mem_increment_bytes": 385893888,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.918973333333334,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.6358986279997225
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 100,
  "seconds": 7.023607421875,
  "mem_increment_bytes": 386633216,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.915976,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 8.803223029000037
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 100,
  "seconds": 1.0237877197265626,
  "mem_increment_bytes": 384023552,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.917829333333334,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 2.9067413880002277
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 100,
  "seconds": 13.299948891999975,
  "mem_increment_bytes": 310030336,
  "mem_meter": "vmhwm",
  "inertia_per_point": 31.874438381510416,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 15.227555935000055
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 100,
  "seconds": 64.9790336570004,
  "mem_increment_bytes": 456986624,
  "mem_meter": "vmhwm",
  "inertia_per_point": 31.405106622395834,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 66.91943278500003
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 100,
  "seconds": 0.400395263671875,
  "mem_increment_bytes": 222941696,
  "mem_meter": "torch-pool",
  "inertia_per_point": 31.73900576171875,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.912107734000074
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 100,
  "seconds": 2.2810234375,
  "mem_increment_bytes": 386633216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 31.710838486979167,
  "n_iter": 0,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 5.634061359000043
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 100,
  "seconds": 0.7029237670898437,
  "mem_increment_bytes": 222005248,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 31.496352192708333,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.0845558269998037
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 100,
  "seconds": 2.507810791015625,
  "mem_increment_bytes": 386633216,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 31.277980334635416,
  "n_iter": 0,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 4.802162582999699
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 100,
  "seconds": 5.685975019000125,
  "mem_increment_bytes": 457367552,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.9172,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 7.313040977000128
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 100,
  "seconds": 149.43517947800137,
  "mem_increment_bytes": 457310208,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.917186666666666,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 151.07001228400077
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 100,
  "seconds": 0.28619406127929686,
  "mem_increment_bytes": 385880064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.917210666666666,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.241822720000073
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 100,
  "seconds": 4.6482431640625,
  "mem_increment_bytes": 385880064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.917194666666667,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 7.499245891999635
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 100,
  "seconds": 0.3252906494140625,
  "mem_increment_bytes": 384009728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.917210666666666,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 2.171656665000228
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 100,
  "seconds": 4.54454833984375,
  "mem_increment_bytes": 384009728,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.917216,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 6.320169927000279
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 100,
  "seconds": 17.541152507000334,
  "mem_increment_bytes": 472166400,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.914136,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 19.1737556879998
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 100,
  "seconds": 218.132385384999,
  "mem_increment_bytes": 465526784,
  "mem_meter": "vmhwm",
  "inertia_per_point": 25.915853333333335,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 219.7659118319989
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 100,
  "seconds": 0.6738767700195313,
  "mem_increment_bytes": 385893888,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.911725333333333,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 3.6097342950001803
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 100,
  "seconds": 6.9066513671875,
  "mem_increment_bytes": 386633216,
  "mem_meter": "torch-pool",
  "inertia_per_point": 25.913325333333333,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 9.774887648000004
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 100,
  "seconds": 1.020650146484375,
  "mem_increment_bytes": 384023552,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.913778666666666,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 2.818388769000194
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 100,
  "seconds": 7.04271044921875,
  "mem_increment_bytes": 386633216,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 25.916149333333333,
  "n_iter": 12,
  "tile_n": 10485,
  "tile_k": 100,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k100-d32",
   "n": 1500000,
   "d": 32,
   "k": 100,
   "seed": 103
  },
  "wall_seconds": 8.823813126999994
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 550,
  "seconds": 14.308090473999982,
  "mem_increment_bytes": 233472000,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.085982373046875,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 16.41461243699996
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 550,
  "seconds": 45.68165783099994,
  "mem_increment_bytes": 367042560,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0868726255580357,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 47.78314455999998
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 550,
  "seconds": 0.652031005859375,
  "mem_increment_bytes": 102904832,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0850312241908482,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.053001270000095
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 550,
  "seconds": 1.4010726318359374,
  "mem_increment_bytes": 270903808,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.087045802641369,
  "n_iter": 0,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.669282625000051
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 550,
  "seconds": 1.3449195556640625,
  "mem_increment_bytes": 102904832,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0856885246930803,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.3710463950001213
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 550,
  "seconds": 2.082630859375,
  "mem_increment_bytes": 270904320,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0862445571754091,
  "n_iter": 0,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.02242078099971
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 550,
  "seconds": 5.276215317000151,
  "mem_increment_bytes": 103530496,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0273471428571428,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 6.881911881000178
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 550,
  "seconds": 73.21192442599931,
  "mem_increment_bytes": 144646144,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0273444047619047,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 74.83387565599878
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 550,
  "seconds": 0.376355224609375,
  "mem_increment_bytes": 77164544,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0273535714285715,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.320180289000291
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 550,
  "seconds": 1.7504088134765625,
  "mem_increment_bytes": 301015040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0273528571428572,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.498381040999902
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 550,
  "seconds": 0.37542373657226563,
  "mem_increment_bytes": 77164544,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0273555952380953,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 1.938818552000157
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 550,
  "seconds": 1.793132080078125,
  "mem_increment_bytes": 300312064,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0273446428571429,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.342676854000274
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 550,
  "seconds": 18.658333959000174,
  "mem_increment_bytes": 233480192,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0246946428571428,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 20.28831047599988
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 550,
  "seconds": 117.09177436200116,
  "mem_increment_bytes": 367091712,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0241085714285714,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 118.71126100400033
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 550,
  "seconds": 1.0133824462890626,
  "mem_increment_bytes": 102904832,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0243005952380952,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.036455160999594
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 550,
  "seconds": 3.14766552734375,
  "mem_increment_bytes": 300333568,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.024240238095238,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 6.152782868000031
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 550,
  "seconds": 1.7099964599609374,
  "mem_increment_bytes": 102904832,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0242966666666666,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.3168087670001114
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 550,
  "seconds": 3.88125634765625,
  "mem_increment_bytes": 300329984,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.024825238095238,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 5.427342744999805
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 550,
  "seconds": 46.07458303000021,
  "mem_increment_bytes": 367104000,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0878030196707589,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 48.175452209000014
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 550,
  "seconds": 14.528232520000074,
  "mem_increment_bytes": 233598976,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0877507915969122,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 16.606920166000236
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 550,
  "seconds": 1.39865087890625,
  "mem_increment_bytes": 270903808,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.085726039922805,
  "n_iter": 0,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.718744422999862
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 550,
  "seconds": 0.6566380615234375,
  "mem_increment_bytes": 102904832,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0864060946800596,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.03574683900024
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 550,
  "seconds": 2.079684814453125,
  "mem_increment_bytes": 270904320,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.085999051455543,
  "n_iter": 0,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.0373954879996745
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 550,
  "seconds": 1.3430906982421875,
  "mem_increment_bytes": 102904832,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0860272533017112,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.4037407290002193
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 550,
  "seconds": 73.38685592900038,
  "mem_increment_bytes": 140414976,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0274103571428572,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 74.99134692299958
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 550,
  "seconds": 5.217243776000032,
  "mem_increment_bytes": 103292928,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0274141666666667,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 6.81104201800008
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 550,
  "seconds": 1.7486815185546876,
  "mem_increment_bytes": 301015040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0274125,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.574768038000002
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 550,
  "seconds": 0.37419732666015626,
  "mem_increment_bytes": 77164544,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0274130952380953,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.2901863349998166
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 550,
  "seconds": 1.7942138671875,
  "mem_increment_bytes": 300312064,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0274142857142856,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.2825342379996982
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 550,
  "seconds": 0.37394644165039065,
  "mem_increment_bytes": 77164544,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0274142857142856,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 1.9425943680003002
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 550,
  "seconds": 118.71923284199875,
  "mem_increment_bytes": 367091712,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0248636904761905,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 120.331911149
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 550,
  "seconds": 18.549372574000245,
  "mem_increment_bytes": 233545728,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0254940476190477,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 20.15956817999995
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 550,
  "seconds": 3.132416259765625,
  "mem_increment_bytes": 300333568,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0247504761904762,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 5.919133246999991
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 550,
  "seconds": 1.00874755859375,
  "mem_increment_bytes": 102904832,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0252188095238095,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.958399689000089
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 550,
  "seconds": 3.87650830078125,
  "mem_increment_bytes": 300329984,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0248086904761904,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 5.384993958999985
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 550,
  "seconds": 1.706610595703125,
  "mem_increment_bytes": 102904832,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0245767857142858,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.2627827959995557
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 550,
  "seconds": 14.393136596999739,
  "mem_increment_bytes": 233578496,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0857637359328498,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 16.481687651000357
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 550,
  "seconds": 47.46237816299981,
  "mem_increment_bytes": 367046656,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0855481146530879,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 49.534178488999714
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 550,
  "seconds": 0.6490829467773438,
  "mem_increment_bytes": 102904832,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.087420523623512,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.9934079469999233
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 550,
  "seconds": 1.399564453125,
  "mem_increment_bytes": 270903808,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0889595169503348,
  "n_iter": 0,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.687817870999879
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 550,
  "seconds": 1.3519893798828124,
  "mem_increment_bytes": 102904832,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0855676035853794,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.398362471999917
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 550,
  "seconds": 2.08317236328125,
  "mem_increment_bytes": 270904320,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0862467882719493,
  "n_iter": 0,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.02739194600008
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 550,
  "seconds": 5.262005100999886,
  "mem_increment_bytes": 103374848,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.026860119047619,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 6.871822410000277
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 550,
  "seconds": 68.70185514600053,
  "mem_increment_bytes": 148467712,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0268577380952382,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 70.33638932000031
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 550,
  "seconds": 0.372901123046875,
  "mem_increment_bytes": 77164544,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0268634523809523,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.3157382949998464
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 550,
  "seconds": 1.750635986328125,
  "mem_increment_bytes": 301015040,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0268580952380952,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 4.5913689760000125
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 550,
  "seconds": 0.3649903564453125,
  "mem_increment_bytes": 77164544,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0268579761904761,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 1.918440591000035
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 550,
  "seconds": 1.7935919189453124,
  "mem_increment_bytes": 300312064,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.026857857142857,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.285148231000221
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 550,
  "seconds": 18.940255208999588,
  "mem_increment_bytes": 233529344,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0249607142857142,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 20.551141531999747
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 550,
  "seconds": 116.27700345300036,
  "mem_increment_bytes": 367075328,
  "mem_meter": "vmhwm",
  "inertia_per_point": 1.0250163095238096,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 117.88684049799849
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 550,
  "seconds": 1.0082467651367188,
  "mem_increment_bytes": 102904832,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.0249811904761905,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.9239358250001715
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 550,
  "seconds": 3.137522705078125,
  "mem_increment_bytes": 300333568,
  "mem_meter": "torch-pool",
  "inertia_per_point": 1.025685,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 5.9080979400000615
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 550,
  "seconds": 1.7165047607421875,
  "mem_increment_bytes": 102904832,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0248336904761906,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 3.331344917000024
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 550,
  "seconds": 3.88836328125,
  "mem_increment_bytes": 300329984,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 1.0249666666666666,
  "n_iter": 12,
  "tile_n": 43690,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k550-d6",
   "n": 525000,
   "d": 6,
   "k": 550,
   "seed": 104
  },
  "wall_seconds": 5.375718493000022
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 3000,
  "seconds": 21.740023334999933,
  "mem_increment_bytes": 126124032,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02150122553507487,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 23.947782338999787
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 3000,
  "seconds": 56.34196206000024,
  "mem_increment_bytes": 125267968,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02150955074451588,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 58.52860308900017
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 3000,
  "seconds": 1.4371256103515626,
  "mem_increment_bytes": 33515008,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.02151625968085395,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.9095498540000335
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 3000,
  "seconds": 1.4242508544921875,
  "mem_increment_bytes": 43743232,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.02142881738168222,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.805890969999837
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 0,
  "k": 3000,
  "seconds": 2.946423828125,
  "mem_increment_bytes": 32943616,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.021443617276792174,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.082959743999709
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 0,
  "k": 3000,
  "seconds": 2.66463037109375,
  "mem_increment_bytes": 43743744,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.021472408012107567,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.714676143000361
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 3000,
  "seconds": 22.046660647999943,
  "mem_increment_bytes": 96743424,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02416318359375,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 23.678713152
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 3000,
  "seconds": 51.3680286040003,
  "mem_increment_bytes": 103600128,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02419941044560185,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 52.97861124200017
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 3000,
  "seconds": 0.5476187744140625,
  "mem_increment_bytes": 70944768,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.024156660517939816,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 3.413689004999469
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 3000,
  "seconds": 1.0080263061523438,
  "mem_increment_bytes": 171608064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.024199399594907408,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 3.7634440410001844
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 0,
  "k": 3000,
  "seconds": 0.5876835327148437,
  "mem_increment_bytes": 70944768,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.024154805049189813,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 2.1500715980000678
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 0,
  "k": 3000,
  "seconds": 1.004025146484375,
  "mem_increment_bytes": 171608576,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.024199412254050926,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 2.4763395460004176
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 3000,
  "seconds": 42.083001091999904,
  "mem_increment_bytes": 126263296,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.01810426070601852,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 43.739714440999705
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 3000,
  "seconds": 71.87257426900032,
  "mem_increment_bytes": 125505536,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.018194907407407406,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 73.4418415429991
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 3000,
  "seconds": 1.8911827392578124,
  "mem_increment_bytes": 70992896,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.018140805844907407,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.838868541000011
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 3000,
  "seconds": 2.353565185546875,
  "mem_increment_bytes": 171656192,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.018063816550925926,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.159816182999748
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 0,
  "k": 3000,
  "seconds": 3.41911767578125,
  "mem_increment_bytes": 70992896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.01814798900462963,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.961431097000059
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 0,
  "k": 3000,
  "seconds": 3.627410400390625,
  "mem_increment_bytes": 171656704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.018133866825810186,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.097672717000023
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 3000,
  "seconds": 45.57760828600021,
  "mem_increment_bytes": 125181952,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.021456989316587096,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 47.78217406600015
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 3000,
  "seconds": 21.48726033499952,
  "mem_increment_bytes": 126300160,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02136390847100152,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 23.67937662300028
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 3000,
  "seconds": 1.3968978271484376,
  "mem_increment_bytes": 43743232,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.021558774538393374,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.776758031000099
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 3000,
  "seconds": 1.4320240478515625,
  "mem_increment_bytes": 33515008,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.021449103207058378,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.947751764999339
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 1,
  "k": 3000,
  "seconds": 2.655341796875,
  "mem_increment_bytes": 43743744,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.021539009405065465,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.75430625899935
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 1,
  "k": 3000,
  "seconds": 2.8863427734375,
  "mem_increment_bytes": 32943616,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.021431095419989693,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.039822197999456
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 3000,
  "seconds": 51.235352996000074,
  "mem_increment_bytes": 103493632,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.024804542824074075,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 52.83519873299974
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 3000,
  "seconds": 21.855867438000132,
  "mem_increment_bytes": 96899072,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02475345413773148,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 23.42379644199991
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 3000,
  "seconds": 1.013929931640625,
  "mem_increment_bytes": 171608064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.024805177589699075,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 3.808161883000139
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 3000,
  "seconds": 0.501158935546875,
  "mem_increment_bytes": 70944768,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.02475267650462963,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 3.4847033699998065
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 1,
  "k": 3000,
  "seconds": 1.0085413818359374,
  "mem_increment_bytes": 171608576,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.024804542824074075,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 2.5024569349998274
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 1,
  "k": 3000,
  "seconds": 0.5164541625976562,
  "mem_increment_bytes": 70944768,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.024751586009837963,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 2.0704675480001242
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 3000,
  "seconds": 80.53766830600034,
  "mem_increment_bytes": 125333504,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.01805990306712963,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 82.10222423200139
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 3000,
  "seconds": 41.324951165999664,
  "mem_increment_bytes": 126263296,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.01811915328414352,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 42.9285410020002
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 3000,
  "seconds": 2.349408203125,
  "mem_increment_bytes": 171656192,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.01810569480613426,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.173706909999964
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 3000,
  "seconds": 1.851029052734375,
  "mem_increment_bytes": 70992896,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.018059678819444445,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.748227451000275
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 1,
  "k": 3000,
  "seconds": 3.674951171875,
  "mem_increment_bytes": 171656704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.018150025318287036,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.156679112999882
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 1,
  "k": 3000,
  "seconds": 3.428280517578125,
  "mem_increment_bytes": 70992896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.018068051938657407,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.993262599000445
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 3000,
  "seconds": 21.48988114599979,
  "mem_increment_bytes": 126197760,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.021480539336027924,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 23.695458717999827
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 3000,
  "seconds": 57.017229838000276,
  "mem_increment_bytes": 125423616,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02153300789727105,
  "n_iter": 0,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 59.22262452200084
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 3000,
  "seconds": 1.4310277099609374,
  "mem_increment_bytes": 33515008,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.021484518277203594,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.919062506000046
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 3000,
  "seconds": 1.397294189453125,
  "mem_increment_bytes": 43743232,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.02150035428647642,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.738260213999638
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "init",
  "seed": 2,
  "k": 3000,
  "seconds": 2.915366943359375,
  "mem_increment_bytes": 32943616,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.02143312355324074,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.056339320000006
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "init",
  "seed": 2,
  "k": 3000,
  "seconds": 2.71364697265625,
  "mem_increment_bytes": 43743744,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.021500798825864437,
  "n_iter": 0,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.827475414999753
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 3000,
  "seconds": 21.924925183000596,
  "mem_increment_bytes": 96976896,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.025534689670138888,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 23.51048524900034
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 3000,
  "seconds": 51.350008357999286,
  "mem_increment_bytes": 101654528,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.02559277705439815,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 52.954762188999666
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 3000,
  "seconds": 0.495962890625,
  "mem_increment_bytes": 70944768,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.025537720630787037,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 3.430602080999961
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 3000,
  "seconds": 1.008218017578125,
  "mem_increment_bytes": 171608064,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.025592775245949075,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 3.7797577840001395
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "lloyd",
  "seed": 2,
  "k": 3000,
  "seconds": 0.5112492980957031,
  "mem_increment_bytes": 70944768,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.02553778392650463,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 2.116575744000329
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "lloyd",
  "seed": 2,
  "k": 3000,
  "seconds": 1.0059411010742187,
  "mem_increment_bytes": 171608576,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.025592775245949075,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 2.4826252330003626
 },
 {
  "backend": "torch-cpu",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 3000,
  "seconds": 42.02860741999939,
  "mem_increment_bytes": 126132224,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.01815677445023148,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 43.63894143699963
 },
 {
  "backend": "torch-cpu",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 3000,
  "seconds": 69.86209783699996,
  "mem_increment_bytes": 125292544,
  "mem_meter": "vmhwm",
  "inertia_per_point": 0.018159398509837964,
  "n_iter": 12,
  "tile_n": 4096,
  "tile_k": 256,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 71.40587942399907
 },
 {
  "backend": "torch-cuda",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 3000,
  "seconds": 1.8704954833984375,
  "mem_increment_bytes": 70992896,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.018111158130787036,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.775966779000555
 },
 {
  "backend": "torch-cuda",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 3000,
  "seconds": 2.366977294921875,
  "mem_increment_bytes": 171656192,
  "mem_meter": "torch-pool",
  "inertia_per_point": 0.0180931640625,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.156740168000397
 },
 {
  "backend": "cupy",
  "method": "gemm",
  "phase": "fit",
  "seed": 2,
  "k": 3000,
  "seconds": 3.40126416015625,
  "mem_increment_bytes": 70992896,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.018072504340277778,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 4.945895197999562
 },
 {
  "backend": "cupy",
  "method": "broadcast",
  "phase": "fit",
  "seed": 2,
  "k": 3000,
  "seconds": 3.584860595703125,
  "mem_increment_bytes": 171656704,
  "mem_meter": "cupy-pool",
  "inertia_per_point": 0.018126573350694446,
  "n_iter": 12,
  "tile_n": 65536,
  "tile_k": 128,
  "broadcast_tile_bytes": 134217728,
  "task": {
   "name": "k3000-d2",
   "n": 135000,
   "d": 2,
   "k": 3000,
   "seed": 105,
   "kind": "blobs",
   "true_k": 300,
   "separation": 4,
   "std": 1,
   "quantize": 0.05
  },
  "wall_seconds": 5.116503406999982
 }
]