{
 "file": "gpu-ql3090-24g-nemo-q6-k-control-3090-q6_K-ladder.json",
 "model_tag": "mistral-nemo:12b-instruct-2407-q6_K",
 "arm": "control-3090",
 "box_class": "gpu-ql3090-24g",
 "quantization": "Q6_K",
 "store_gib": 9.37,
 "largest_whole": 131072,
 "topped_out": true,
 "stopped_at": null,
 "errors": [],
 "ups_before": 12.0,
 "ups_after": 11.0,
 "windows": [
  {
   "num_ctx": 4096,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 9829.0
   },
   "runs": 3,
   "decode_median": 65.724,
   "ttft_ms_median": 518.25,
   "board_w_median": 270.81,
   "j_per_1k_median": 4102.0,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 57.0,
   "spread": 0.0067,
   "decode_range": 0.441,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 65.724
  },
  {
   "num_ctx": 8192,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 470.0
   },
   "runs": 3,
   "decode_median": 65.191,
   "ttft_ms_median": 541.14,
   "board_w_median": 257.4,
   "j_per_1k_median": 3948.39,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 61.0,
   "spread": 0.0018,
   "decode_range": 0.117,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 65.191
  },
  {
   "num_ctx": 16384,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 696.0
   },
   "runs": 3,
   "decode_median": 64.764,
   "ttft_ms_median": 479.98,
   "board_w_median": 273.63,
   "j_per_1k_median": 4225.02,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 64.0,
   "spread": 0.008,
   "decode_range": 0.52,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 64.764
  },
  {
   "num_ctx": 32768,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 1392.0
   },
   "runs": 3,
   "decode_median": 64.83,
   "ttft_ms_median": 503.26,
   "board_w_median": 258.17,
   "j_per_1k_median": 3982.29,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 63.0,
   "spread": 0.0043,
   "decode_range": 0.282,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 64.83
  },
  {
   "num_ctx": 49152,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 1342.0
   },
   "runs": 3,
   "decode_median": 64.619,
   "ttft_ms_median": 532.56,
   "board_w_median": 272.05,
   "j_per_1k_median": 4210.26,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 65.0,
   "spread": 0.0053,
   "decode_range": 0.344,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 64.619
  },
  {
   "num_ctx": 65536,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 1440.0
   },
   "runs": 3,
   "decode_median": 64.699,
   "ttft_ms_median": 473.4,
   "board_w_median": 273.5,
   "j_per_1k_median": 4226.9,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 64.0,
   "spread": 0.0066,
   "decode_range": 0.428,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 64.699
  },
  {
   "num_ctx": 98304,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 2880.0
   },
   "runs": 3,
   "decode_median": 64.645,
   "ttft_ms_median": 507.79,
   "board_w_median": 272.56,
   "j_per_1k_median": 4209.01,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 63.0,
   "spread": 0.0037,
   "decode_range": 0.237,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 64.645
  },
  {
   "num_ctx": 131072,
   "whole": true,
   "verdict": "fits: fully resident on the card",
   "vram_delta_mib": {
    "<withheld: GPU UUID of board A>": 2880.0
   },
   "runs": 3,
   "decode_median": 64.739,
   "ttft_ms_median": 544.17,
   "board_w_median": 252.37,
   "j_per_1k_median": 3892.41,
   "sw_power_cap_fraction_median": 1.0,
   "temp_max_c": 64.0,
   "spread": 0.0062,
   "decode_range": 0.401,
   "throttle": [],
   "stop": null,
   "gate_refusals": 0,
   "gates_failed": [],
   "passes": true,
   "headline_decode": 64.739
  }
 ],
 "resident_named_card_growth_mib": 9829.0,
 "resident": true,
 "passing_windows": [
  4096,
  8192,
  16384,
  32768,
  49152,
  65536,
  98304,
  131072
 ]
}