{
"_metadata": {
"description": "CANN baseline 性能数据(950PR 平台,经验证)",
"hardware": "950pr",
"source": "950pr1.json + verify_run1/2_dev3.json (verified mean)",
"note": "t_hw_us 为 hardware-anchored 性能下界,亚微秒结果按 1us 取下限;baseline_perf_us 优先采用 950PR 实测均值,缺失项按 max(t_hw_us × 10, 10us) 回退。",
"t_hw_source": "hardware-anchored 性能下界 (roofline)"
},
"level1": {
"exp": {
"1": {
"baseline_perf_us": 3.75,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 16.25,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 32.08,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 173.62,
"t_hw_us": 83.89
},
"5": {
"baseline_perf_us": 388.95,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 4.7,
"t_hw_us": 1.31
},
"7": {
"baseline_perf_us": 4.6,
"t_hw_us": 1.29
},
"8": {
"baseline_perf_us": 5.99,
"t_hw_us": 2.96
},
"9": {
"baseline_perf_us": 115.655,
"t_hw_us": 62.11
},
"10": {
"baseline_perf_us": 4.68,
"t_hw_us": 1.31
},
"11": {
"baseline_perf_us": 5.59,
"t_hw_us": 2.73
},
"12": {
"baseline_perf_us": 4.51,
"t_hw_us": 1.25
},
"13": {
"baseline_perf_us": 44.76,
"t_hw_us": 27.28
},
"14": {
"baseline_perf_us": 8.1,
"t_hw_us": 3.8
},
"15": {
"baseline_perf_us": 5.52,
"t_hw_us": 2.62
},
"16": {
"baseline_perf_us": 6.32,
"t_hw_us": 2.61
},
"17": {
"baseline_perf_us": 6.36,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 8.61,
"t_hw_us": 5.24
},
"19": {
"baseline_perf_us": 6.39,
"t_hw_us": 2.61
},
"20": {
"baseline_perf_us": 20.2,
"t_hw_us": 13.19
}
},
"foreach_addcdiv_scalar": {
"1": {
"baseline_perf_us": 23.96,
"t_hw_us": 15.73
},
"2": {
"baseline_perf_us": 91.63,
"t_hw_us": 47.19
},
"3": {
"baseline_perf_us": 119.67,
"t_hw_us": 62.91
},
"4": {
"baseline_perf_us": 859.945,
"t_hw_us": 503.32
},
"5": {
"baseline_perf_us": 478.87,
"t_hw_us": 251.66
},
"6": {
"baseline_perf_us": 12.26,
"t_hw_us": 3.92
},
"7": {
"baseline_perf_us": 13.025,
"t_hw_us": 7.73
},
"8": {
"baseline_perf_us": 12.47,
"t_hw_us": 4.43
},
"9": {
"baseline_perf_us": 746.28,
"t_hw_us": 372.69
},
"10": {
"baseline_perf_us": 13.295,
"t_hw_us": 7.88
},
"11": {
"baseline_perf_us": 20.84,
"t_hw_us": 8.19
},
"12": {
"baseline_perf_us": 12.55,
"t_hw_us": 7.5
},
"13": {
"baseline_perf_us": 80.76,
"t_hw_us": 40.92
},
"14": {
"baseline_perf_us": 31.665,
"t_hw_us": 22.82
},
"15": {
"baseline_perf_us": 23.57,
"t_hw_us": 15.74
},
"16": {
"baseline_perf_us": 62.805,
"t_hw_us": 31.34
},
"17": {
"baseline_perf_us": 19.2,
"t_hw_us": 7.85
},
"18": {
"baseline_perf_us": 44.03,
"t_hw_us": 31.41
},
"19": {
"baseline_perf_us": 20.105,
"t_hw_us": 7.84
},
"20": {
"baseline_perf_us": 116.175,
"t_hw_us": 79.12
}
},
"foreach_norm": {
"1": {
"baseline_perf_us": 10.225,
"t_hw_us": 2.62
},
"2": {
"baseline_perf_us": 43.365,
"t_hw_us": 31.46
},
"3": {
"baseline_perf_us": 31.59,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 14.595,
"t_hw_us": 5.24
},
"5": {
"baseline_perf_us": 92.105,
"t_hw_us": 62.91
},
"6": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"7": {
"baseline_perf_us": 12.9,
"t_hw_us": 1.29
},
"8": {
"baseline_perf_us": 22.625,
"t_hw_us": 2.95
},
"9": {
"baseline_perf_us": 152.215,
"t_hw_us": 124.23
},
"10": {
"baseline_perf_us": 8.19,
"t_hw_us": 1.31
},
"11": {
"baseline_perf_us": 17.345,
"t_hw_us": 8.19
},
"12": {
"baseline_perf_us": 8.68,
"t_hw_us": 1.25
},
"13": {
"baseline_perf_us": 49.14,
"t_hw_us": 27.28
},
"14": {
"baseline_perf_us": 12.805,
"t_hw_us": 3.79
},
"15": {
"baseline_perf_us": 13.385,
"t_hw_us": 5.25
},
"16": {
"baseline_perf_us": 18.78,
"t_hw_us": 10.45
},
"17": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 19.405,
"t_hw_us": 10.47
},
"19": {
"baseline_perf_us": 21.44,
"t_hw_us": 5.22
},
"20": {
"baseline_perf_us": 147.84,
"t_hw_us": 105.49
}
},
"gelu": {
"1": {
"baseline_perf_us": 4.49,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 15.37,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 30.14,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 172.93,
"t_hw_us": 83.89
},
"5": {
"baseline_perf_us": 387.265,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 4.46,
"t_hw_us": 1.31
},
"7": {
"baseline_perf_us": 4.45,
"t_hw_us": 1.29
},
"8": {
"baseline_perf_us": 6.11,
"t_hw_us": 2.95
},
"9": {
"baseline_perf_us": 117.53,
"t_hw_us": 62.11
},
"10": {
"baseline_perf_us": 4.56,
"t_hw_us": 1.31
},
"11": {
"baseline_perf_us": 6.05,
"t_hw_us": 2.73
},
"12": {
"baseline_perf_us": 4.38,
"t_hw_us": 1.25
},
"13": {
"baseline_perf_us": 37.455,
"t_hw_us": 27.28
},
"14": {
"baseline_perf_us": 7.78,
"t_hw_us": 3.79
},
"15": {
"baseline_perf_us": 5.98,
"t_hw_us": 2.62
},
"16": {
"baseline_perf_us": 6.23,
"t_hw_us": 2.61
},
"17": {
"baseline_perf_us": 6.31,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 8.86,
"t_hw_us": 5.24
},
"19": {
"baseline_perf_us": 6.26,
"t_hw_us": 2.61
},
"20": {
"baseline_perf_us": 36.195,
"t_hw_us": 26.37
}
},
"masked_scale": {
"1": {
"baseline_perf_us": 11.495,
"t_hw_us": 1.97
},
"2": {
"baseline_perf_us": 31.83,
"t_hw_us": 13.11
},
"3": {
"baseline_perf_us": 87.98,
"t_hw_us": 41.94
},
"4": {
"baseline_perf_us": 817.845,
"t_hw_us": 251.66
},
"5": {
"baseline_perf_us": 1353.855,
"t_hw_us": 209.72
},
"6": {
"baseline_perf_us": 11.485,
"t_hw_us": 1.96
},
"7": {
"baseline_perf_us": 11.59,
"t_hw_us": 1.93
},
"8": {
"baseline_perf_us": 14.5,
"t_hw_us": 4.43
},
"9": {
"baseline_perf_us": 588.285,
"t_hw_us": 186.34
},
"10": {
"baseline_perf_us": 11.46,
"t_hw_us": 1.97
},
"11": {
"baseline_perf_us": 13.24,
"t_hw_us": 3.42
},
"12": {
"baseline_perf_us": 9.21,
"t_hw_us": 2.5
},
"13": {
"baseline_perf_us": 87.7,
"t_hw_us": 54.56
},
"14": {
"baseline_perf_us": 17.95,
"t_hw_us": 5.7
},
"15": {
"baseline_perf_us": 13.07,
"t_hw_us": 3.28
},
"16": {
"baseline_perf_us": 14.21,
"t_hw_us": 3.92
},
"17": {
"baseline_perf_us": 13.92,
"t_hw_us": 3.93
},
"18": {
"baseline_perf_us": 19.52,
"t_hw_us": 7.85
},
"19": {
"baseline_perf_us": 17.62,
"t_hw_us": 7.84
},
"20": {
"baseline_perf_us": 69.12,
"t_hw_us": 32.97
}
},
"mish": {
"1": {
"baseline_perf_us": 5.88,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 16.4,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 52.37,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 206.72,
"t_hw_us": 83.89
},
"5": {
"baseline_perf_us": 388.81,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 6.03,
"t_hw_us": 1.31
},
"7": {
"baseline_perf_us": 5.645,
"t_hw_us": 1.29
},
"8": {
"baseline_perf_us": 7.58,
"t_hw_us": 2.96
},
"9": {
"baseline_perf_us": 148.8,
"t_hw_us": 62.11
},
"10": {
"baseline_perf_us": 5.88,
"t_hw_us": 1.31
},
"11": {
"baseline_perf_us": 7.19,
"t_hw_us": 2.73
},
"12": {
"baseline_perf_us": 5.81,
"t_hw_us": 1.25
},
"13": {
"baseline_perf_us": 39.905,
"t_hw_us": 27.28
},
"14": {
"baseline_perf_us": 11.95,
"t_hw_us": 3.79
},
"15": {
"baseline_perf_us": 7.09,
"t_hw_us": 2.62
},
"16": {
"baseline_perf_us": 9.37,
"t_hw_us": 2.61
},
"17": {
"baseline_perf_us": 9.3,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 10.59,
"t_hw_us": 5.24
},
"19": {
"baseline_perf_us": 9.5,
"t_hw_us": 2.61
},
"20": {
"baseline_perf_us": 37.59,
"t_hw_us": 26.37
}
},
"sigmoid": {
"1": {
"baseline_perf_us": 4.25,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 15.23,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 29.635,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 168.32,
"t_hw_us": 83.89
},
"5": {
"baseline_perf_us": 387.715,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 4.51,
"t_hw_us": 1.31
},
"7": {
"baseline_perf_us": 4.41,
"t_hw_us": 1.29
},
"8": {
"baseline_perf_us": 6.42,
"t_hw_us": 2.95
},
"9": {
"baseline_perf_us": 113.67,
"t_hw_us": 62.11
},
"10": {
"baseline_perf_us": 4.41,
"t_hw_us": 1.31
},
"11": {
"baseline_perf_us": 6.16,
"t_hw_us": 2.73
},
"12": {
"baseline_perf_us": 4.36,
"t_hw_us": 1.25
},
"13": {
"baseline_perf_us": 37.825,
"t_hw_us": 27.28
},
"14": {
"baseline_perf_us": 8.32,
"t_hw_us": 3.79
},
"15": {
"baseline_perf_us": 6.04,
"t_hw_us": 2.62
},
"16": {
"baseline_perf_us": 6.83,
"t_hw_us": 2.61
},
"17": {
"baseline_perf_us": 6.66,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 8.97,
"t_hw_us": 5.24
},
"19": {
"baseline_perf_us": 6.79,
"t_hw_us": 2.61
},
"20": {
"baseline_perf_us": 36.74,
"t_hw_us": 26.37
}
},
"swi_glu": {
"1": {
"baseline_perf_us": 6.75,
"t_hw_us": 2.62
},
"2": {
"baseline_perf_us": 27.47,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 57.455,
"t_hw_us": 41.94
},
"4": {
"baseline_perf_us": 294.38,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 620.68,
"t_hw_us": 333.99
},
"6": {
"baseline_perf_us": 6.3,
"t_hw_us": 2.62
},
"7": {
"baseline_perf_us": 6.985,
"t_hw_us": 2.54
},
"8": {
"baseline_perf_us": 9.55,
"t_hw_us": 5.91
},
"9": {
"baseline_perf_us": 21.21,
"t_hw_us": 2.33
},
"10": {
"baseline_perf_us": 6.725,
"t_hw_us": 2.62
},
"11": {
"baseline_perf_us": 3.83,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 25.0,
"t_hw_us": 2.5
},
"13": {
"baseline_perf_us": 3.52,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 9.06,
"t_hw_us": 3.8
},
"15": {
"baseline_perf_us": 8.73,
"t_hw_us": 5.24
},
"16": {
"baseline_perf_us": 9.47,
"t_hw_us": 5.2
},
"17": {
"baseline_perf_us": 9.72,
"t_hw_us": 5.24
},
"18": {
"baseline_perf_us": 27.47,
"t_hw_us": 20.95
},
"19": {
"baseline_perf_us": 16.15,
"t_hw_us": 10.45
},
"20": {
"baseline_perf_us": 21.225,
"t_hw_us": 13.06
}
}
},
"level2": {
"apply_adam_w": {
"1": {
"baseline_perf_us": 16.46,
"t_hw_us": 10.49
},
"2": {
"baseline_perf_us": 30.18,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 128.83,
"t_hw_us": 83.89
},
"4": {
"baseline_perf_us": 312.435,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 320.605,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 11.58,
"t_hw_us": 5.23
},
"7": {
"baseline_perf_us": 16.08,
"t_hw_us": 10.3
},
"8": {
"baseline_perf_us": 11.91,
"t_hw_us": 5.91
},
"9": {
"baseline_perf_us": 503.455,
"t_hw_us": 248.46
},
"10": {
"baseline_perf_us": 16.545,
"t_hw_us": 10.51
},
"11": {
"baseline_perf_us": 11.51,
"t_hw_us": 5.46
},
"12": {
"baseline_perf_us": 10.91,
"t_hw_us": 5.0
},
"13": {
"baseline_perf_us": 180.335,
"t_hw_us": 109.13
},
"14": {
"baseline_perf_us": 22.53,
"t_hw_us": 15.21
},
"15": {
"baseline_perf_us": 16.345,
"t_hw_us": 10.49
},
"16": {
"baseline_perf_us": 17.35,
"t_hw_us": 10.45
},
"17": {
"baseline_perf_us": 17.06,
"t_hw_us": 10.47
},
"18": {
"baseline_perf_us": 29.125,
"t_hw_us": 20.94
},
"19": {
"baseline_perf_us": 17.75,
"t_hw_us": 10.45
},
"20": {
"baseline_perf_us": 172.065,
"t_hw_us": 105.49
}
},
"apply_rotary_pos_emb": {
"1": {
"baseline_perf_us": 79.215,
"t_hw_us": 42.02
},
"2": {
"baseline_perf_us": 113.24,
"t_hw_us": 71.06
},
"3": {
"baseline_perf_us": 39.725,
"t_hw_us": 17.47
},
"4": {
"baseline_perf_us": 180.82,
"t_hw_us": 39.33
},
"5": {
"baseline_perf_us": 391.07,
"t_hw_us": 168.43
},
"6": {
"baseline_perf_us": 364.325,
"t_hw_us": 41.74
},
"7": {
"baseline_perf_us": 57.25,
"t_hw_us": 27.72
},
"8": {
"baseline_perf_us": 60.48,
"t_hw_us": 31.97
},
"9": {
"baseline_perf_us": 102.77,
"t_hw_us": 23.77
},
"10": {
"baseline_perf_us": 68.54,
"t_hw_us": 19.56
},
"11": {
"baseline_perf_us": 186.7,
"t_hw_us": 118.34
},
"12": {
"baseline_perf_us": 57.6,
"t_hw_us": 5.76
},
"13": {
"baseline_perf_us": 423.87,
"t_hw_us": 120.0
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 26.1,
"t_hw_us": 2.61
},
"16": {
"baseline_perf_us": 22.06,
"t_hw_us": 5.01
},
"17": {
"baseline_perf_us": 2489.31,
"t_hw_us": 314.6
},
"18": {
"baseline_perf_us": 23.28,
"t_hw_us": 5.26
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 146.54,
"t_hw_us": 64.85
}
},
"arg_max": {
"1": {
"baseline_perf_us": 6.09,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 16.63,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 31.25,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 235.02,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 200.98,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 234.14,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 4.3,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 12.32,
"t_hw_us": 2.58
},
"9": {
"baseline_perf_us": 5.47,
"t_hw_us": 1.48
},
"10": {
"baseline_perf_us": 168.61,
"t_hw_us": 124.23
},
"11": {
"baseline_perf_us": 4.3,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 6.79,
"t_hw_us": 2.73
},
"13": {
"baseline_perf_us": 5.18,
"t_hw_us": 1.12
},
"14": {
"baseline_perf_us": 6.79,
"t_hw_us": 2.62
},
"15": {
"baseline_perf_us": 19.93,
"t_hw_us": 13.64
},
"16": {
"baseline_perf_us": 21.19,
"t_hw_us": 15.21
},
"17": {
"baseline_perf_us": 6.42,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 9.59,
"t_hw_us": 2.61
},
"19": {
"baseline_perf_us": 9.59,
"t_hw_us": 5.23
},
"20": {
"baseline_perf_us": 9.08,
"t_hw_us": 2.62
}
},
"cross_entropy_loss": {
"1": {
"baseline_perf_us": 18.535,
"t_hw_us": 2.63
},
"2": {
"baseline_perf_us": 40.355,
"t_hw_us": 20.98
},
"3": {
"baseline_perf_us": 78.92,
"t_hw_us": 41.96
},
"4": {
"baseline_perf_us": 414.49,
"t_hw_us": 167.81
},
"5": {
"baseline_perf_us": 410.93,
"t_hw_us": 167.79
},
"6": {
"baseline_perf_us": 18.635,
"t_hw_us": 2.62
},
"7": {
"baseline_perf_us": 12.9,
"t_hw_us": 1.29
},
"8": {
"baseline_perf_us": 18.475,
"t_hw_us": 2.96
},
"9": {
"baseline_perf_us": 270.385,
"t_hw_us": 62.78
},
"10": {
"baseline_perf_us": 13.2,
"t_hw_us": 1.32
},
"11": {
"baseline_perf_us": 27.3,
"t_hw_us": 2.73
},
"12": {
"baseline_perf_us": 75.0,
"t_hw_us": 7.5
},
"13": {
"baseline_perf_us": 60.79,
"t_hw_us": 14.46
},
"14": {
"baseline_perf_us": 39.08,
"t_hw_us": 7.62
},
"15": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"16": {
"baseline_perf_us": 15.41,
"t_hw_us": 2.61
},
"17": {
"baseline_perf_us": 17.14,
"t_hw_us": 5.25
},
"18": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"19": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"20": {
"baseline_perf_us": 68.94,
"t_hw_us": 26.9
}
},
"cummin": {
"1": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 209.7,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 3355.4,
"t_hw_us": 335.54
},
"6": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 12.9,
"t_hw_us": 1.29
},
"9": {
"baseline_perf_us": 29.5,
"t_hw_us": 2.95
},
"10": {
"baseline_perf_us": 496.705,
"t_hw_us": 62.11
},
"11": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 27.3,
"t_hw_us": 2.73
},
"13": {
"baseline_perf_us": 12.5,
"t_hw_us": 1.25
},
"14": {
"baseline_perf_us": 90.805,
"t_hw_us": 13.64
},
"15": {
"baseline_perf_us": 76.1,
"t_hw_us": 7.61
},
"16": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"17": {
"baseline_perf_us": 20.94,
"t_hw_us": 2.61
},
"18": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"19": {
"baseline_perf_us": 27.15,
"t_hw_us": 5.24
},
"20": {
"baseline_perf_us": 26.1,
"t_hw_us": 2.61
}
},
"dynamic_quant": {
"1": {
"baseline_perf_us": 5.35,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 10.37,
"t_hw_us": 5.24
},
"3": {
"baseline_perf_us": 29.87,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 136.98,
"t_hw_us": 83.89
},
"5": {
"baseline_perf_us": 136.46,
"t_hw_us": 83.89
},
"6": {
"baseline_perf_us": 340.025,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 5.715,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 5.705,
"t_hw_us": 1.29
},
"9": {
"baseline_perf_us": 6.18,
"t_hw_us": 1.48
},
"10": {
"baseline_perf_us": 110.065,
"t_hw_us": 62.11
},
"11": {
"baseline_perf_us": 5.97,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 5.83,
"t_hw_us": 1.37
},
"13": {
"baseline_perf_us": 5.655,
"t_hw_us": 1.12
},
"14": {
"baseline_perf_us": 39.16,
"t_hw_us": 13.64
},
"15": {
"baseline_perf_us": 5.755,
"t_hw_us": 1.31
},
"16": {
"baseline_perf_us": 9.495,
"t_hw_us": 3.8
},
"17": {
"baseline_perf_us": 5.48,
"t_hw_us": 1.31
},
"18": {
"baseline_perf_us": 7.57,
"t_hw_us": 2.61
},
"19": {
"baseline_perf_us": 7.22,
"t_hw_us": 2.62
},
"20": {
"baseline_perf_us": 7.2,
"t_hw_us": 2.62
}
},
"gather": {
"1": {
"baseline_perf_us": 19.7,
"t_hw_us": 1.97
},
"2": {
"baseline_perf_us": 78.01,
"t_hw_us": 15.73
},
"3": {
"baseline_perf_us": 311.32,
"t_hw_us": 31.46
},
"4": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 2097.2,
"t_hw_us": 209.72
},
"7": {
"baseline_perf_us": 9.12,
"t_hw_us": 3.27
},
"8": {
"baseline_perf_us": 19.86,
"t_hw_us": 2.58
},
"9": {
"baseline_perf_us": 24.73,
"t_hw_us": 3.7
},
"10": {
"baseline_perf_us": 528.045,
"t_hw_us": 92.91
},
"11": {
"baseline_perf_us": 22.12,
"t_hw_us": 3.94
},
"12": {
"baseline_perf_us": 10.31,
"t_hw_us": 2.73
},
"13": {
"baseline_perf_us": 2.41,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 98.515,
"t_hw_us": 26.04
},
"15": {
"baseline_perf_us": 28.485,
"t_hw_us": 15.21
},
"16": {
"baseline_perf_us": 19.7,
"t_hw_us": 1.97
},
"17": {
"baseline_perf_us": 40.41,
"t_hw_us": 6.5
},
"18": {
"baseline_perf_us": 37.42,
"t_hw_us": 5.23
},
"19": {
"baseline_perf_us": 40.89,
"t_hw_us": 6.53
},
"20": {
"baseline_perf_us": 21.0,
"t_hw_us": 5.2
}
},
"gcd": {
"1": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"2": {
"baseline_perf_us": 84.67,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"4": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 1122.29,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 72.04,
"t_hw_us": 10.47
},
"7": {
"baseline_perf_us": 13.1,
"t_hw_us": 5.15
},
"8": {
"baseline_perf_us": 29.5,
"t_hw_us": 2.95
},
"9": {
"baseline_perf_us": 2001.86,
"t_hw_us": 247.79
},
"10": {
"baseline_perf_us": 13.94,
"t_hw_us": 5.26
},
"11": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"12": {
"baseline_perf_us": 121.1,
"t_hw_us": 12.11
},
"13": {
"baseline_perf_us": 393.3,
"t_hw_us": 39.33
},
"14": {
"baseline_perf_us": 9.04,
"t_hw_us": 5.24
},
"15": {
"baseline_perf_us": 81.445,
"t_hw_us": 54.56
},
"16": {
"baseline_perf_us": 38.0,
"t_hw_us": 3.8
},
"17": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"18": {
"baseline_perf_us": 38.79,
"t_hw_us": 5.24
},
"19": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"20": {
"baseline_perf_us": 527.5,
"t_hw_us": 52.75
}
},
"grid_sampler_3d": {
"1": {
"baseline_perf_us": 28.7,
"t_hw_us": 2.87
},
"2": {
"baseline_perf_us": 27.4,
"t_hw_us": 2.74
},
"3": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"4": {
"baseline_perf_us": 219.5,
"t_hw_us": 21.95
},
"5": {
"baseline_perf_us": 858.5,
"t_hw_us": 85.85
},
"6": {
"baseline_perf_us": 848.7,
"t_hw_us": 84.87
},
"7": {
"baseline_perf_us": 38.0,
"t_hw_us": 3.8
},
"8": {
"baseline_perf_us": 11.1,
"t_hw_us": 1.11
},
"9": {
"baseline_perf_us": 22.8,
"t_hw_us": 2.28
},
"10": {
"baseline_perf_us": 993.8,
"t_hw_us": 99.38
},
"11": {
"baseline_perf_us": 32.3,
"t_hw_us": 3.23
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 25.3,
"t_hw_us": 2.53
},
"14": {
"baseline_perf_us": 7.01,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 98.3,
"t_hw_us": 9.83
},
"16": {
"baseline_perf_us": 209.6,
"t_hw_us": 20.96
},
"17": {
"baseline_perf_us": 2514.9,
"t_hw_us": 251.49
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 50.6,
"t_hw_us": 5.06
},
"20": {
"baseline_perf_us": 3276.4,
"t_hw_us": 327.64
}
},
"group_norm": {
"1": {
"baseline_perf_us": 9.48,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 29.18,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 60.16,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 28.96,
"t_hw_us": 5.1
},
"5": {
"baseline_perf_us": 13.54,
"t_hw_us": 2.61
},
"6": {
"baseline_perf_us": 6.48,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 31.68,
"t_hw_us": 10.49
},
"8": {
"baseline_perf_us": 20.35,
"t_hw_us": 5.2
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 10.99,
"t_hw_us": 1.15
},
"11": {
"baseline_perf_us": 24.36,
"t_hw_us": 7.86
},
"12": {
"baseline_perf_us": 7.81,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 7.19,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 34.19,
"t_hw_us": 5.04
},
"15": {
"baseline_perf_us": 97.7,
"t_hw_us": 9.77
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 19.23,
"t_hw_us": 2.6
},
"18": {
"baseline_perf_us": 30.81,
"t_hw_us": 10.49
},
"19": {
"baseline_perf_us": 103.03,
"t_hw_us": 41.94
},
"20": {
"baseline_perf_us": 50.9,
"t_hw_us": 5.09
}
},
"maximum": {
"1": {
"baseline_perf_us": 4.87,
"t_hw_us": 2.62
},
"2": {
"baseline_perf_us": 27.285,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 55.09,
"t_hw_us": 41.94
},
"4": {
"baseline_perf_us": 613.58,
"t_hw_us": 335.54
},
"5": {
"baseline_perf_us": 290.1,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 289.01,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 4.86,
"t_hw_us": 2.62
},
"8": {
"baseline_perf_us": 7.82,
"t_hw_us": 5.15
},
"9": {
"baseline_perf_us": 5.23,
"t_hw_us": 2.95
},
"10": {
"baseline_perf_us": 205.47,
"t_hw_us": 124.23
},
"11": {
"baseline_perf_us": 6.39,
"t_hw_us": 2.63
},
"12": {
"baseline_perf_us": 35.06,
"t_hw_us": 24.21
},
"13": {
"baseline_perf_us": 27.12,
"t_hw_us": 19.66
},
"14": {
"baseline_perf_us": 7.62,
"t_hw_us": 5.0
},
"15": {
"baseline_perf_us": 35.03,
"t_hw_us": 27.28
},
"16": {
"baseline_perf_us": 20.03,
"t_hw_us": 15.21
},
"17": {
"baseline_perf_us": 4.86,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 6.0,
"t_hw_us": 2.62
},
"19": {
"baseline_perf_us": 15.46,
"t_hw_us": 10.47
},
"20": {
"baseline_perf_us": 37.365,
"t_hw_us": 26.38
}
},
"resize_bilinear": {
"1": {
"baseline_perf_us": 22.05,
"t_hw_us": 5.24
},
"2": {
"baseline_perf_us": 88.66,
"t_hw_us": 41.94
},
"3": {
"baseline_perf_us": 238.62,
"t_hw_us": 83.89
},
"4": {
"baseline_perf_us": 32.3,
"t_hw_us": 3.23
},
"5": {
"baseline_perf_us": 491.3,
"t_hw_us": 112.37
},
"6": {
"baseline_perf_us": 78.79,
"t_hw_us": 30.03
},
"7": {
"baseline_perf_us": 32.3,
"t_hw_us": 3.23
},
"8": {
"baseline_perf_us": 19.65,
"t_hw_us": 3.44
},
"9": {
"baseline_perf_us": 40.135,
"t_hw_us": 7.43
},
"10": {
"baseline_perf_us": 111.53,
"t_hw_us": 41.94
},
"11": {
"baseline_perf_us": 18.6,
"t_hw_us": 7.86
},
"12": {
"baseline_perf_us": 210.14,
"t_hw_us": 80.16
},
"13": {
"baseline_perf_us": 72.1,
"t_hw_us": 7.21
},
"14": {
"baseline_perf_us": 11.88,
"t_hw_us": 1.37
},
"15": {
"baseline_perf_us": 408.155,
"t_hw_us": 167.77
},
"16": {
"baseline_perf_us": 489.375,
"t_hw_us": 166.46
},
"17": {
"baseline_perf_us": 52.25,
"t_hw_us": 20.94
},
"18": {
"baseline_perf_us": 16.1,
"t_hw_us": 1.61
},
"19": {
"baseline_perf_us": 16.1,
"t_hw_us": 1.61
},
"20": {
"baseline_perf_us": 31.66,
"t_hw_us": 5.9
}
},
"rms_norm": {
"1": {
"baseline_perf_us": 9.885,
"t_hw_us": 3.93
},
"2": {
"baseline_perf_us": 15.61,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 17.06,
"t_hw_us": 10.49
},
"4": {
"baseline_perf_us": 30.04,
"t_hw_us": 20.98
},
"5": {
"baseline_perf_us": 170.84,
"t_hw_us": 83.91
},
"6": {
"baseline_perf_us": 31.145,
"t_hw_us": 20.96
},
"7": {
"baseline_perf_us": 11.455,
"t_hw_us": 5.4
},
"8": {
"baseline_perf_us": 64.78,
"t_hw_us": 41.89
},
"9": {
"baseline_perf_us": 67.08,
"t_hw_us": 41.86
},
"10": {
"baseline_perf_us": 10.18,
"t_hw_us": 4.03
},
"11": {
"baseline_perf_us": 28.335,
"t_hw_us": 20.49
},
"12": {
"baseline_perf_us": 32.62,
"t_hw_us": 22.21
},
"13": {
"baseline_perf_us": 15.53,
"t_hw_us": 9.02
},
"14": {
"baseline_perf_us": 8.0,
"t_hw_us": 3.77
},
"15": {
"baseline_perf_us": 25.0,
"t_hw_us": 2.5
},
"16": {
"baseline_perf_us": 3.3,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 5.83,
"t_hw_us": 2.38
},
"18": {
"baseline_perf_us": 16.56,
"t_hw_us": 10.48
},
"19": {
"baseline_perf_us": 10.19,
"t_hw_us": 5.23
},
"20": {
"baseline_perf_us": 47.89,
"t_hw_us": 33.42
}
},
"scatter": {
"1": {
"baseline_perf_us": 32.8,
"t_hw_us": 3.28
},
"2": {
"baseline_perf_us": 142.0,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 734.0,
"t_hw_us": 73.4
},
"4": {
"baseline_perf_us": 2097.2,
"t_hw_us": 209.72
},
"5": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 2516.6,
"t_hw_us": 251.66
},
"7": {
"baseline_perf_us": 32.7,
"t_hw_us": 3.27
},
"8": {
"baseline_perf_us": 32.2,
"t_hw_us": 3.22
},
"9": {
"baseline_perf_us": 59.1,
"t_hw_us": 5.91
},
"10": {
"baseline_perf_us": 1550.3,
"t_hw_us": 155.03
},
"11": {
"baseline_perf_us": 32.8,
"t_hw_us": 3.28
},
"12": {
"baseline_perf_us": 81.9,
"t_hw_us": 8.19
},
"13": {
"baseline_perf_us": 22.65,
"t_hw_us": 3.05
},
"14": {
"baseline_perf_us": 13.45,
"t_hw_us": 2.63
},
"15": {
"baseline_perf_us": 269.2,
"t_hw_us": 57.04
},
"16": {
"baseline_perf_us": 54.1,
"t_hw_us": 17.75
},
"17": {
"baseline_perf_us": 32.8,
"t_hw_us": 3.28
},
"18": {
"baseline_perf_us": 87.15,
"t_hw_us": 9.11
},
"19": {
"baseline_perf_us": 82.5,
"t_hw_us": 20.93
},
"20": {
"baseline_perf_us": 65.4,
"t_hw_us": 6.54
}
},
"softmax": {
"1": {
"baseline_perf_us": 5.51,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 15.89,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 30.5,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 171.99,
"t_hw_us": 83.89
},
"5": {
"baseline_perf_us": 389.945,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 207.12,
"t_hw_us": 84.74
},
"7": {
"baseline_perf_us": 8.28,
"t_hw_us": 2.62
},
"8": {
"baseline_perf_us": 29.56,
"t_hw_us": 20.99
},
"9": {
"baseline_perf_us": 74.47,
"t_hw_us": 41.74
},
"10": {
"baseline_perf_us": 5.66,
"t_hw_us": 1.29
},
"11": {
"baseline_perf_us": 297.98,
"t_hw_us": 129.7
},
"12": {
"baseline_perf_us": 19.49,
"t_hw_us": 12.16
},
"13": {
"baseline_perf_us": 20.23,
"t_hw_us": 2.5
},
"14": {
"baseline_perf_us": 51.865,
"t_hw_us": 27.28
},
"15": {
"baseline_perf_us": 9.9,
"t_hw_us": 3.8
},
"16": {
"baseline_perf_us": 5.93,
"t_hw_us": 1.31
},
"17": {
"baseline_perf_us": 9.12,
"t_hw_us": 5.22
},
"18": {
"baseline_perf_us": 8.19,
"t_hw_us": 2.62
},
"19": {
"baseline_perf_us": 8.3,
"t_hw_us": 2.61
},
"20": {
"baseline_perf_us": 51.3,
"t_hw_us": 26.37
}
},
"unsorted_segment_sum": {
"1": {
"baseline_perf_us": 9.42,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 27.08,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 54.21,
"t_hw_us": 20.99
},
"4": {
"baseline_perf_us": 648.86,
"t_hw_us": 167.79
},
"5": {
"baseline_perf_us": 6157.0,
"t_hw_us": 615.7
},
"6": {
"baseline_perf_us": 10.46,
"t_hw_us": 1.31
},
"7": {
"baseline_perf_us": 11.72,
"t_hw_us": 2.58
},
"8": {
"baseline_perf_us": 13.5,
"t_hw_us": 1.48
},
"9": {
"baseline_perf_us": 137.86,
"t_hw_us": 62.12
},
"10": {
"baseline_perf_us": 13.17,
"t_hw_us": 2.64
},
"11": {
"baseline_perf_us": 6.92,
"t_hw_us": 1.37
},
"12": {
"baseline_perf_us": 39.3,
"t_hw_us": 3.93
},
"13": {
"baseline_perf_us": 61.96,
"t_hw_us": 27.28
},
"14": {
"baseline_perf_us": 14.25,
"t_hw_us": 3.8
},
"15": {
"baseline_perf_us": 52.4,
"t_hw_us": 5.24
},
"16": {
"baseline_perf_us": 78.6,
"t_hw_us": 7.86
},
"17": {
"baseline_perf_us": 6.7,
"t_hw_us": 1.31
},
"18": {
"baseline_perf_us": 10.14,
"t_hw_us": 2.61
},
"19": {
"baseline_perf_us": 22.425,
"t_hw_us": 5.24
},
"20": {
"baseline_perf_us": 9.99,
"t_hw_us": 2.62
}
}
},
"level3": {
"adaptive_avg_pool_3d": {
"1": {
"baseline_perf_us": 195.94,
"t_hw_us": 20.97
},
"2": {
"baseline_perf_us": 31.16,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"4": {
"baseline_perf_us": 32.3,
"t_hw_us": 3.23
},
"5": {
"baseline_perf_us": 353.19,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 899.31,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 207.595,
"t_hw_us": 29.07
},
"8": {
"baseline_perf_us": 93.8,
"t_hw_us": 9.38
},
"9": {
"baseline_perf_us": 21.4,
"t_hw_us": 2.14
},
"10": {
"baseline_perf_us": 286.21,
"t_hw_us": 48.2
},
"11": {
"baseline_perf_us": 1169.335,
"t_hw_us": 176.45
},
"12": {
"baseline_perf_us": 68.2,
"t_hw_us": 6.82
},
"13": {
"baseline_perf_us": 29.41,
"t_hw_us": 4.62
},
"14": {
"baseline_perf_us": 7.01,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 93.8,
"t_hw_us": 9.38
},
"16": {
"baseline_perf_us": 101.69,
"t_hw_us": 20.0
},
"17": {
"baseline_perf_us": 582.385,
"t_hw_us": 238.17
},
"18": {
"baseline_perf_us": 10.8,
"t_hw_us": 1.08
},
"19": {
"baseline_perf_us": 11.02,
"t_hw_us": 4.62
},
"20": {
"baseline_perf_us": 981.96,
"t_hw_us": 320.08
}
},
"conv_2d": {
"1": {
"baseline_perf_us": 7.61,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 42.95,
"t_hw_us": 22.37
},
"3": {
"baseline_perf_us": 316.855,
"t_hw_us": 248.55
},
"4": {
"baseline_perf_us": 197.1,
"t_hw_us": 89.48
},
"5": {
"baseline_perf_us": 764.83,
"t_hw_us": 357.91
},
"6": {
"baseline_perf_us": 22.84,
"t_hw_us": 9.94
},
"7": {
"baseline_perf_us": 8.6,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 32.625,
"t_hw_us": 10.0
},
"9": {
"baseline_perf_us": 7.24,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 33.0,
"t_hw_us": 3.3
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 5.8,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 9.43,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 6.34,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 9.06,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 7.61,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 46.105,
"t_hw_us": 11.47
},
"18": {
"baseline_perf_us": 176.22,
"t_hw_us": 81.98
},
"19": {
"baseline_perf_us": 6.81,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 96.13,
"t_hw_us": 42.0
}
},
"depthwise_conv_2d": {
"1": {
"baseline_perf_us": 7.91,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 29.325,
"t_hw_us": 5.25
},
"3": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"4": {
"baseline_perf_us": 73.205,
"t_hw_us": 20.98
},
"5": {
"baseline_perf_us": 77.035,
"t_hw_us": 10.51
},
"6": {
"baseline_perf_us": 13.2,
"t_hw_us": 1.32
},
"7": {
"baseline_perf_us": 9.085,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 16.62,
"t_hw_us": 2.37
},
"9": {
"baseline_perf_us": 7.18,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 11.2,
"t_hw_us": 1.12
},
"11": {
"baseline_perf_us": 7.61,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 6.98,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 8.13,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 7.195,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 7.97,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 17.47,
"t_hw_us": 3.79
},
"18": {
"baseline_perf_us": 56.95,
"t_hw_us": 10.29
},
"19": {
"baseline_perf_us": 8.38,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 35.72,
"t_hw_us": 10.25
}
},
"dequant_swiglu_quant": {
"1": {
"baseline_perf_us": 8.84,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 17.23,
"t_hw_us": 5.24
},
"3": {
"baseline_perf_us": 47.87,
"t_hw_us": 20.98
},
"4": {
"baseline_perf_us": 47.39,
"t_hw_us": 20.98
},
"5": {
"baseline_perf_us": 8.44,
"t_hw_us": 1.0
},
"6": {
"baseline_perf_us": 27.62,
"t_hw_us": 10.49
},
"7": {
"baseline_perf_us": 17.55,
"t_hw_us": 5.24
},
"8": {
"baseline_perf_us": 11.38,
"t_hw_us": 2.61
},
"9": {
"baseline_perf_us": 8.775,
"t_hw_us": 2.63
},
"10": {
"baseline_perf_us": 19.63,
"t_hw_us": 10.5
},
"11": {
"baseline_perf_us": 56.57,
"t_hw_us": 41.98
},
"12": {
"baseline_perf_us": 55.94,
"t_hw_us": 41.97
},
"13": {
"baseline_perf_us": 6.605,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 33.35,
"t_hw_us": 21.0
},
"15": {
"baseline_perf_us": 19.88,
"t_hw_us": 10.49
},
"16": {
"baseline_perf_us": 13.985,
"t_hw_us": 5.25
},
"17": {
"baseline_perf_us": 12.24,
"t_hw_us": 1.63
},
"18": {
"baseline_perf_us": 33.865,
"t_hw_us": 21.05
},
"19": {
"baseline_perf_us": 32.385,
"t_hw_us": 20.5
},
"20": {
"baseline_perf_us": 32.95,
"t_hw_us": 21.01
}
},
"grouped_matmul": {
"1": {
"baseline_perf_us": 6.3,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 11.945,
"t_hw_us": 5.41
},
"3": {
"baseline_perf_us": 49.97,
"t_hw_us": 23.61
},
"4": {
"baseline_perf_us": 17.46,
"t_hw_us": 6.54
},
"5": {
"baseline_perf_us": 14.42,
"t_hw_us": 5.99
},
"6": {
"baseline_perf_us": 14.96,
"t_hw_us": 6.07
},
"7": {
"baseline_perf_us": 7.84,
"t_hw_us": 2.06
},
"8": {
"baseline_perf_us": 7.68,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 15.39,
"t_hw_us": 3.27
},
"10": {
"baseline_perf_us": 54.975,
"t_hw_us": 23.56
},
"11": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 5.68,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 68.295,
"t_hw_us": 39.77
},
"14": {
"baseline_perf_us": 9.67,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 29.13,
"t_hw_us": 11.78
},
"16": {
"baseline_perf_us": 6.2,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 7.615,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 5.035,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 3.7,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 9.78,
"t_hw_us": 1.27
}
},
"moe_finalize_routing": {
"1": {
"baseline_perf_us": 40.405,
"t_hw_us": 6.62
},
"2": {
"baseline_perf_us": 181.08,
"t_hw_us": 126.16
},
"3": {
"baseline_perf_us": 55.57,
"t_hw_us": 42.01
},
"4": {
"baseline_perf_us": 42.87,
"t_hw_us": 6.62
},
"5": {
"baseline_perf_us": 26.945,
"t_hw_us": 9.89
},
"6": {
"baseline_perf_us": 7.105,
"t_hw_us": 1.98
},
"7": {
"baseline_perf_us": 13.4,
"t_hw_us": 1.34
},
"8": {
"baseline_perf_us": 38.155,
"t_hw_us": 6.6
},
"9": {
"baseline_perf_us": 121.2,
"t_hw_us": 12.12
},
"10": {
"baseline_perf_us": 5.11,
"t_hw_us": 1.0
},
"11": {
"baseline_perf_us": 6.48,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 43.375,
"t_hw_us": 7.94
},
"13": {
"baseline_perf_us": 8.705,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 5.58,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 1059.085,
"t_hw_us": 211.39
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.545,
"t_hw_us": 3.95
},
"18": {
"baseline_perf_us": 20.08,
"t_hw_us": 7.77
},
"19": {
"baseline_perf_us": 8.3,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 5.25,
"t_hw_us": 1.0
}
},
"moe_gating_top_k_softmax": {
"1": {
"baseline_perf_us": 4.505,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"3": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"4": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"5": {
"baseline_perf_us": 52.4,
"t_hw_us": 5.24
},
"6": {
"baseline_perf_us": 4.06,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 7.015,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 52.8,
"t_hw_us": 5.28
},
"9": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"10": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"11": {
"baseline_perf_us": 4.85,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"13": {
"baseline_perf_us": 12.8,
"t_hw_us": 1.28
},
"14": {
"baseline_perf_us": 4.435,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 4.79,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 4.35,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"19": {
"baseline_perf_us": 4.545,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 5.65,
"t_hw_us": 1.0
}
},
"moe_re_routing": {
"1": {
"baseline_perf_us": 7.73,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 46.08,
"t_hw_us": 21.01
},
"3": {
"baseline_perf_us": 834.285,
"t_hw_us": 335.87
},
"4": {
"baseline_perf_us": 10.89,
"t_hw_us": 2.62
},
"5": {
"baseline_perf_us": 22.9,
"t_hw_us": 10.49
},
"6": {
"baseline_perf_us": 5.385,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 13.305,
"t_hw_us": 5.24
},
"8": {
"baseline_perf_us": 7.235,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 6.575,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 43.69,
"t_hw_us": 10.57
},
"11": {
"baseline_perf_us": 38.1,
"t_hw_us": 21.14
},
"12": {
"baseline_perf_us": 9.045,
"t_hw_us": 1.31
},
"13": {
"baseline_perf_us": 14.205,
"t_hw_us": 5.24
},
"14": {
"baseline_perf_us": 7.99,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 7.275,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 399.29,
"t_hw_us": 168.43
},
"19": {
"baseline_perf_us": 399.89,
"t_hw_us": 169.09
},
"20": {
"baseline_perf_us": 4.935,
"t_hw_us": 1.0
}
},
"quant_matmul": {
"1": {
"baseline_perf_us": 40.7,
"t_hw_us": 30.41
},
"2": {
"baseline_perf_us": 58.08,
"t_hw_us": 39.1
},
"3": {
"baseline_perf_us": 56.23,
"t_hw_us": 38.75
},
"4": {
"baseline_perf_us": 165.94,
"t_hw_us": 139.03
},
"7": {
"baseline_perf_us": 43.31,
"t_hw_us": 30.41
},
"9": {
"baseline_perf_us": 56.49,
"t_hw_us": 39.72
},
"10": {
"baseline_perf_us": 19.19,
"t_hw_us": 9.93
},
"11": {
"baseline_perf_us": 19.82,
"t_hw_us": 9.93
},
"12": {
"baseline_perf_us": 54.7,
"t_hw_us": 39.72
},
"13": {
"baseline_perf_us": 59.975,
"t_hw_us": 43.26
},
"14": {
"baseline_perf_us": 15.57,
"t_hw_us": 10.49
},
"15": {
"baseline_perf_us": 40.64,
"t_hw_us": 32.12
},
"16": {
"baseline_perf_us": 5.985,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 29.155,
"t_hw_us": 9.94
},
"18": {
"baseline_perf_us": 2.67,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 191.33,
"t_hw_us": 158.89
},
"20": {
"baseline_perf_us": 196.18,
"t_hw_us": 158.89
},
"5": {
"baseline_perf_us": 1067.5,
"t_hw_us": 106.75
},
"6": {
"baseline_perf_us": 1390.3,
"t_hw_us": 139.03
},
"8": {
"baseline_perf_us": 1390.3,
"t_hw_us": 139.03
}
},
"top_k": {
"1": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"3": {
"baseline_perf_us": 209.7,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 25.8,
"t_hw_us": 2.58
},
"9": {
"baseline_perf_us": 14.8,
"t_hw_us": 1.48
},
"10": {
"baseline_perf_us": 1242.3,
"t_hw_us": 124.23
},
"11": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 27.3,
"t_hw_us": 2.73
},
"13": {
"baseline_perf_us": 17.6,
"t_hw_us": 1.76
},
"14": {
"baseline_perf_us": 25.0,
"t_hw_us": 2.5
},
"15": {
"baseline_perf_us": 209.9,
"t_hw_us": 20.99
},
"16": {
"baseline_perf_us": 152.1,
"t_hw_us": 15.21
},
"17": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 26.1,
"t_hw_us": 2.61
},
"19": {
"baseline_perf_us": 18.2,
"t_hw_us": 1.82
},
"20": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
}
},
"transpose": {
"1": {
"baseline_perf_us": 389.055,
"t_hw_us": 167.77
},
"2": {
"baseline_perf_us": 27.39,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 75.8,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 562.85,
"t_hw_us": 167.77
},
"5": {
"baseline_perf_us": 542.95,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 10.45,
"t_hw_us": 1.47
},
"7": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 12.935,
"t_hw_us": 2.58
},
"9": {
"baseline_perf_us": 14.8,
"t_hw_us": 1.48
},
"10": {
"baseline_perf_us": 460.62,
"t_hw_us": 124.23
},
"11": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 8.75,
"t_hw_us": 2.73
},
"13": {
"baseline_perf_us": 11.5,
"t_hw_us": 1.15
},
"14": {
"baseline_perf_us": 13.31,
"t_hw_us": 2.27
},
"15": {
"baseline_perf_us": 136.4,
"t_hw_us": 13.64
},
"16": {
"baseline_perf_us": 152.1,
"t_hw_us": 15.21
},
"17": {
"baseline_perf_us": 11.93,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 11.88,
"t_hw_us": 2.61
},
"19": {
"baseline_perf_us": 8.67,
"t_hw_us": 1.31
},
"20": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
}
},
"unique": {
"1": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"2": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 209.7,
"t_hw_us": 20.97
},
"4": {
"baseline_perf_us": 4479.3,
"t_hw_us": 447.93
},
"5": {
"baseline_perf_us": 1677.7,
"t_hw_us": 167.77
},
"6": {
"baseline_perf_us": 14545.6,
"t_hw_us": 1454.56
},
"7": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 25.8,
"t_hw_us": 2.58
},
"9": {
"baseline_perf_us": 14.8,
"t_hw_us": 1.48
},
"10": {
"baseline_perf_us": 2241.3,
"t_hw_us": 224.13
},
"11": {
"baseline_perf_us": 13.1,
"t_hw_us": 1.31
},
"12": {
"baseline_perf_us": 27.3,
"t_hw_us": 2.73
},
"13": {
"baseline_perf_us": 11.5,
"t_hw_us": 1.15
},
"14": {
"baseline_perf_us": 25.0,
"t_hw_us": 2.5
},
"15": {
"baseline_perf_us": 209.9,
"t_hw_us": 20.99
},
"16": {
"baseline_perf_us": 76.1,
"t_hw_us": 7.61
},
"17": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 26.1,
"t_hw_us": 2.61
},
"19": {
"baseline_perf_us": 104.7,
"t_hw_us": 10.47
},
"20": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
}
},
"weight_quant_batch_matmul": {
"1": {
"baseline_perf_us": 7.61,
"t_hw_us": 2.63
},
"2": {
"baseline_perf_us": 10.52,
"t_hw_us": 5.27
},
"3": {
"baseline_perf_us": 11.49,
"t_hw_us": 5.29
},
"4": {
"baseline_perf_us": 13.26,
"t_hw_us": 1.81
},
"5": {
"baseline_perf_us": 14.08,
"t_hw_us": 4.94
},
"6": {
"baseline_perf_us": 17.36,
"t_hw_us": 9.22
},
"7": {
"baseline_perf_us": 23.695,
"t_hw_us": 16.45
},
"8": {
"baseline_perf_us": 16.48,
"t_hw_us": 10.51
},
"9": {
"baseline_perf_us": 41.735,
"t_hw_us": 31.53
},
"10": {
"baseline_perf_us": 22.555,
"t_hw_us": 15.76
},
"11": {
"baseline_perf_us": 16.63,
"t_hw_us": 10.38
},
"12": {
"baseline_perf_us": 38.46,
"t_hw_us": 28.29
},
"13": {
"baseline_perf_us": 38.915,
"t_hw_us": 28.64
},
"14": {
"baseline_perf_us": 49.51,
"t_hw_us": 37.05
},
"15": {
"baseline_perf_us": 62.15,
"t_hw_us": 44.9
},
"16": {
"baseline_perf_us": 34.18,
"t_hw_us": 17.22
},
"17": {
"baseline_perf_us": 57.32,
"t_hw_us": 37.86
},
"18": {
"baseline_perf_us": 64.15,
"t_hw_us": 41.99
},
"19": {
"baseline_perf_us": 67.14,
"t_hw_us": 52.54
},
"20": {
"baseline_perf_us": 216.01,
"t_hw_us": 147.99
}
},
"dilation_2d": {
"1": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"3": {
"baseline_perf_us": 419.5,
"t_hw_us": 41.95
},
"4": {
"baseline_perf_us": 423.4,
"t_hw_us": 42.34
},
"5": {
"baseline_perf_us": 378.2,
"t_hw_us": 37.82
},
"6": {
"baseline_perf_us": 838.9,
"t_hw_us": 83.89
},
"7": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 11.2,
"t_hw_us": 1.12
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 18.9,
"t_hw_us": 1.89
},
"18": {
"baseline_perf_us": 102.8,
"t_hw_us": 10.28
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 51.2,
"t_hw_us": 5.12
}
},
"strided_slice": {
"1": {
"baseline_perf_us": 4.35,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 16.17,
"t_hw_us": 10.48
},
"3": {
"baseline_perf_us": 6.34,
"t_hw_us": 1.31
},
"4": {
"baseline_perf_us": 336.88,
"t_hw_us": 83.88
},
"5": {
"baseline_perf_us": 132.43,
"t_hw_us": 41.94
},
"6": {
"baseline_perf_us": 120.04,
"t_hw_us": 20.97
},
"7": {
"baseline_perf_us": 4.15,
"t_hw_us": 1.31
},
"8": {
"baseline_perf_us": 3.49,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 5.34,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 7.32,
"t_hw_us": 2.5
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 9.35,
"t_hw_us": 2.62
},
"13": {
"baseline_perf_us": 2.78,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 3.51,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 3.16,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 5.74,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 379.61,
"t_hw_us": 163.87
}
},
"nms": {
"1": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"3": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"4": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"5": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"6": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
}
},
"roi_align": {
"1": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"2": {
"baseline_perf_us": 209.7,
"t_hw_us": 20.97
},
"3": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"4": {
"baseline_perf_us": 52.4,
"t_hw_us": 5.24
},
"5": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"6": {
"baseline_perf_us": 1194.1,
"t_hw_us": 119.41
},
"7": {
"baseline_perf_us": 37.4,
"t_hw_us": 3.74
},
"8": {
"baseline_perf_us": 54.3,
"t_hw_us": 5.43
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 36.5,
"t_hw_us": 3.65
},
"11": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"12": {
"baseline_perf_us": 120.5,
"t_hw_us": 12.05
},
"13": {
"baseline_perf_us": 51.4,
"t_hw_us": 5.14
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 102.8,
"t_hw_us": 10.28
},
"16": {
"baseline_perf_us": 25.3,
"t_hw_us": 2.53
},
"17": {
"baseline_perf_us": 68.1,
"t_hw_us": 6.81
},
"18": {
"baseline_perf_us": 49.2,
"t_hw_us": 4.92
},
"19": {
"baseline_perf_us": 12.6,
"t_hw_us": 1.26
},
"20": {
"baseline_perf_us": 102.4,
"t_hw_us": 10.24
}
},
"conv_3d_backprop_filter": {
"1": {
"baseline_perf_us": 64.5,
"t_hw_us": 6.45
},
"2": {
"baseline_perf_us": 101.65,
"t_hw_us": 12.58
},
"3": {
"baseline_perf_us": 1318.33,
"t_hw_us": 310.69
},
"4": {
"baseline_perf_us": 83.9,
"t_hw_us": 8.39
},
"5": {
"baseline_perf_us": 110.6,
"t_hw_us": 11.06
},
"6": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 105.6,
"t_hw_us": 10.56
},
"8": {
"baseline_perf_us": 224.41,
"t_hw_us": 52.1
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 505.17,
"t_hw_us": 99.36
},
"11": {
"baseline_perf_us": 99.3,
"t_hw_us": 9.93
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 70.4,
"t_hw_us": 7.04
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 165.33,
"t_hw_us": 26.91
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 26.2,
"t_hw_us": 2.62
},
"18": {
"baseline_perf_us": 4994.47,
"t_hw_us": 1242.76
},
"19": {
"baseline_perf_us": 2224.35,
"t_hw_us": 536.87
},
"20": {
"baseline_perf_us": 1554.62,
"t_hw_us": 440.86
}
},
"add_rms_norm_dynamic_quant": {
"1": {
"baseline_perf_us": 677.46,
"t_hw_us": 167.78
},
"2": {
"baseline_perf_us": 675.31,
"t_hw_us": 167.77
},
"3": {
"baseline_perf_us": 676.82,
"t_hw_us": 167.78
},
"4": {
"baseline_perf_us": 590.2,
"t_hw_us": 146.82
},
"5": {
"baseline_perf_us": 676.11,
"t_hw_us": 167.78
},
"6": {
"baseline_perf_us": 678.8,
"t_hw_us": 167.77
},
"7": {
"baseline_perf_us": 103.35,
"t_hw_us": 41.95
},
"8": {
"baseline_perf_us": 103.07,
"t_hw_us": 41.95
},
"9": {
"baseline_perf_us": 104.0,
"t_hw_us": 41.95
},
"10": {
"baseline_perf_us": 104.11,
"t_hw_us": 41.94
},
"11": {
"baseline_perf_us": 110.96,
"t_hw_us": 41.96
},
"12": {
"baseline_perf_us": 104.95,
"t_hw_us": 41.89
},
"13": {
"baseline_perf_us": 56.59,
"t_hw_us": 21.01
},
"14": {
"baseline_perf_us": 103.95,
"t_hw_us": 41.95
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 14.74,
"t_hw_us": 2.61
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 53.65,
"t_hw_us": 20.98
}
},
"engram_gate_fusion": {
"1": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"3": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"4": {
"baseline_perf_us": 15.5,
"t_hw_us": 1.55
},
"5": {
"baseline_perf_us": 59.7,
"t_hw_us": 5.97
},
"6": {
"baseline_perf_us": 236.6,
"t_hw_us": 23.66
},
"7": {
"baseline_perf_us": 236.6,
"t_hw_us": 23.66
},
"8": {
"baseline_perf_us": 236.6,
"t_hw_us": 23.66
},
"9": {
"baseline_perf_us": 18.5,
"t_hw_us": 1.85
},
"10": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 15.5,
"t_hw_us": 1.55
},
"13": {
"baseline_perf_us": 22.5,
"t_hw_us": 2.25
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 29.3,
"t_hw_us": 2.93
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 15.9,
"t_hw_us": 1.59
},
"18": {
"baseline_perf_us": 15.5,
"t_hw_us": 1.55
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
}
},
"mhc_sinkhorn": {
"1": {
"baseline_perf_us": 5.36,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 5.82,
"t_hw_us": 1.0
},
"3": {
"baseline_perf_us": 8.47,
"t_hw_us": 1.0
},
"4": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"5": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"6": {
"baseline_perf_us": 15.9,
"t_hw_us": 1.59
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 4.17,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 5.65,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 15.5,
"t_hw_us": 1.55
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
}
}
},
"level4": {
"gqa": {
"1": {
"baseline_perf_us": 14.21,
"t_hw_us": 1.97
},
"2": {
"baseline_perf_us": 11.75,
"t_hw_us": 2.29
},
"3": {
"baseline_perf_us": 22.0,
"t_hw_us": 5.9
},
"4": {
"baseline_perf_us": 36.95,
"t_hw_us": 19.88
},
"5": {
"baseline_perf_us": 22.01,
"t_hw_us": 3.93
},
"6": {
"baseline_perf_us": 94.705,
"t_hw_us": 23.59
},
"7": {
"baseline_perf_us": 13.2,
"t_hw_us": 2.62
},
"8": {
"baseline_perf_us": 135.07,
"t_hw_us": 79.54
},
"9": {
"baseline_perf_us": 112.4,
"t_hw_us": 79.54
},
"10": {
"baseline_perf_us": 150.66,
"t_hw_us": 79.54
},
"11": {
"baseline_perf_us": 105.575,
"t_hw_us": 83.97
},
"12": {
"baseline_perf_us": 106.845,
"t_hw_us": 84.13
},
"13": {
"baseline_perf_us": 379.425,
"t_hw_us": 314.88
},
"14": {
"baseline_perf_us": 216.885,
"t_hw_us": 170.39
},
"15": {
"baseline_perf_us": 406.145,
"t_hw_us": 315.19
},
"16": {
"baseline_perf_us": 382.49,
"t_hw_us": 315.19
},
"17": {
"baseline_perf_us": 991.325,
"t_hw_us": 178.26
},
"18": {
"baseline_perf_us": 10.6,
"t_hw_us": 1.39
},
"19": {
"baseline_perf_us": 17.58,
"t_hw_us": 3.93
},
"20": {
"baseline_perf_us": 16.285,
"t_hw_us": 1.97
}
},
"mha": {
"1": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 21.365,
"t_hw_us": 3.93
},
"3": {
"baseline_perf_us": 38.855,
"t_hw_us": 15.73
},
"4": {
"baseline_perf_us": 19.46,
"t_hw_us": 3.93
},
"5": {
"baseline_perf_us": 9.88,
"t_hw_us": 1.23
},
"6": {
"baseline_perf_us": 13.55,
"t_hw_us": 3.93
},
"7": {
"baseline_perf_us": 140.85,
"t_hw_us": 79.54
},
"8": {
"baseline_perf_us": 138.0,
"t_hw_us": 79.54
},
"9": {
"baseline_perf_us": 186.56,
"t_hw_us": 79.54
},
"10": {
"baseline_perf_us": 35.94,
"t_hw_us": 20.98
},
"11": {
"baseline_perf_us": 205.34,
"t_hw_us": 167.85
},
"12": {
"baseline_perf_us": 203.715,
"t_hw_us": 167.94
},
"13": {
"baseline_perf_us": 375.97,
"t_hw_us": 314.88
},
"14": {
"baseline_perf_us": 207.015,
"t_hw_us": 168.43
},
"15": {
"baseline_perf_us": 383.15,
"t_hw_us": 314.88
},
"16": {
"baseline_perf_us": 379.385,
"t_hw_us": 315.19
},
"17": {
"baseline_perf_us": 308.49,
"t_hw_us": 252.64
},
"18": {
"baseline_perf_us": 382.425,
"t_hw_us": 315.19
},
"19": {
"baseline_perf_us": 14.77,
"t_hw_us": 3.93
},
"20": {
"baseline_perf_us": 12.54,
"t_hw_us": 3.93
}
},
"mla": {
"1": {
"baseline_perf_us": 82.9,
"t_hw_us": 23.78
},
"2": {
"baseline_perf_us": 404.45,
"t_hw_us": 95.11
},
"4": {
"baseline_perf_us": 105.11,
"t_hw_us": 47.37
},
"5": {
"baseline_perf_us": 199.59,
"t_hw_us": 47.55
},
"7": {
"baseline_perf_us": 15.7,
"t_hw_us": 1.57
},
"8": {
"baseline_perf_us": 97.23,
"t_hw_us": 25.07
},
"9": {
"baseline_perf_us": 145.73,
"t_hw_us": 50.13
},
"10": {
"baseline_perf_us": 223.995,
"t_hw_us": 94.0
},
"11": {
"baseline_perf_us": 218.515,
"t_hw_us": 106.17
},
"12": {
"baseline_perf_us": 144.58,
"t_hw_us": 91.24
},
"13": {
"baseline_perf_us": 75.72,
"t_hw_us": 23.78
},
"15": {
"baseline_perf_us": 646.11,
"t_hw_us": 169.01
},
"16": {
"baseline_perf_us": 224.405,
"t_hw_us": 94.0
},
"17": {
"baseline_perf_us": 178.665,
"t_hw_us": 42.25
},
"18": {
"baseline_perf_us": 577.925,
"t_hw_us": 253.52
},
"19": {
"baseline_perf_us": 173.125,
"t_hw_us": 42.25
},
"20": {
"baseline_perf_us": 82.99,
"t_hw_us": 23.78
},
"3": {
"baseline_perf_us": 106.5,
"t_hw_us": 10.65
},
"6": {
"baseline_perf_us": 251.3355,
"t_hw_us": 74.57
},
"14": {
"baseline_perf_us": 126.5765,
"t_hw_us": 21.14
}
},
"mla_prolog": {
"1": {
"baseline_perf_us": 97.03,
"t_hw_us": 76.61
},
"2": {
"baseline_perf_us": 97.8,
"t_hw_us": 76.63
},
"3": {
"baseline_perf_us": 104.135,
"t_hw_us": 76.89
},
"4": {
"baseline_perf_us": 123.41,
"t_hw_us": 77.18
},
"5": {
"baseline_perf_us": 192.83,
"t_hw_us": 77.77
},
"6": {
"baseline_perf_us": 97.565,
"t_hw_us": 76.63
},
"7": {
"baseline_perf_us": 99.77,
"t_hw_us": 76.74
},
"8": {
"baseline_perf_us": 87.86,
"t_hw_us": 76.74
},
"9": {
"baseline_perf_us": 124.56,
"t_hw_us": 77.18
},
"10": {
"baseline_perf_us": 103.84,
"t_hw_us": 76.89
},
"11": {
"baseline_perf_us": 83.73,
"t_hw_us": 73.9
},
"12": {
"baseline_perf_us": 89.19,
"t_hw_us": 74.15
},
"13": {
"baseline_perf_us": 180.76,
"t_hw_us": 74.9
},
"14": {
"baseline_perf_us": 85.93,
"t_hw_us": 74.02
},
"15": {
"baseline_perf_us": 110.99,
"t_hw_us": 74.4
},
"16": {
"baseline_perf_us": 191.19,
"t_hw_us": 77.69
},
"17": {
"baseline_perf_us": 703.13,
"t_hw_us": 145.25
},
"18": {
"baseline_perf_us": 270.57,
"t_hw_us": 75.4
},
"19": {
"baseline_perf_us": 97.01,
"t_hw_us": 76.61
},
"20": {
"baseline_perf_us": 96.56,
"t_hw_us": 76.61
}
},
"sparse_flash_attention": {
"10": {
"baseline_perf_us": 422.5,
"t_hw_us": 42.25
},
"11": {
"baseline_perf_us": 1590.7,
"t_hw_us": 159.07
},
"12": {
"baseline_perf_us": 845.1,
"t_hw_us": 84.51
},
"15": {
"baseline_perf_us": 461.2,
"t_hw_us": 46.12
},
"17": {
"baseline_perf_us": 141.7,
"t_hw_us": 14.17
},
"1": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
},
"2": {
"baseline_perf_us": 124.5,
"t_hw_us": 12.45
},
"3": {
"baseline_perf_us": 795.4,
"t_hw_us": 79.54
},
"4": {
"baseline_perf_us": 248.6,
"t_hw_us": 24.86
},
"5": {
"baseline_perf_us": 99.4,
"t_hw_us": 9.94
},
"6": {
"baseline_perf_us": 212.2,
"t_hw_us": 21.22
},
"7": {
"baseline_perf_us": 1057.6,
"t_hw_us": 105.76
},
"8": {
"baseline_perf_us": 111.4,
"t_hw_us": 11.14
},
"9": {
"baseline_perf_us": 397.7,
"t_hw_us": 39.77
},
"13": {
"baseline_perf_us": 397.7,
"t_hw_us": 39.77
},
"14": {
"baseline_perf_us": 118.2,
"t_hw_us": 11.82
},
"16": {
"baseline_perf_us": 99.4,
"t_hw_us": 9.94
},
"18": {
"baseline_perf_us": 236.3,
"t_hw_us": 23.63
},
"19": {
"baseline_perf_us": 107.3,
"t_hw_us": 10.73
},
"20": {
"baseline_perf_us": 104.9,
"t_hw_us": 10.49
}
},
"gru": {
"1": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"3": {
"baseline_perf_us": 10.9,
"t_hw_us": 1.09
},
"4": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"5": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"6": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"8": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"14": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"15": {
"baseline_perf_us": 16.4,
"t_hw_us": 1.64
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
}
},
"lstm": {
"1": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"2": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"3": {
"baseline_perf_us": 14.6,
"t_hw_us": 1.46
},
"4": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"5": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"6": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"7": {
"baseline_perf_us": 10.2,
"t_hw_us": 1.02
},
"8": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"9": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"10": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"11": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 10.1,
"t_hw_us": 1.01
},
"14": {
"baseline_perf_us": 21.8,
"t_hw_us": 2.18
},
"15": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"16": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"17": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"18": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"19": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"20": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
}
},
"grouped_matmul_swiglu_quant": {
"1": {
"baseline_perf_us": 192.45,
"t_hw_us": 147.46
},
"2": {
"baseline_perf_us": 195.63,
"t_hw_us": 148.03
},
"3": {
"baseline_perf_us": 220.96,
"t_hw_us": 151.47
},
"4": {
"baseline_perf_us": 260.12,
"t_hw_us": 156.06
},
"5": {
"baseline_perf_us": 440.35,
"t_hw_us": 278.06
},
"6": {
"baseline_perf_us": 197.92,
"t_hw_us": 149.18
},
"7": {
"baseline_perf_us": 142.04,
"t_hw_us": 110.74
},
"8": {
"baseline_perf_us": 373.33,
"t_hw_us": 296.06
},
"9": {
"baseline_perf_us": 256.43,
"t_hw_us": 210.0
},
"10": {
"baseline_perf_us": 87.74,
"t_hw_us": 52.82
},
"11": {
"baseline_perf_us": 37.47,
"t_hw_us": 11.15
},
"12": {
"baseline_perf_us": 10.0,
"t_hw_us": 1.0
},
"13": {
"baseline_perf_us": 55.34,
"t_hw_us": 22.33
},
"14": {
"baseline_perf_us": 244.63,
"t_hw_us": 173.18
},
"15": {
"baseline_perf_us": 19.41,
"t_hw_us": 3.28
},
"16": {
"baseline_perf_us": 20.51,
"t_hw_us": 3.01
},
"17": {
"baseline_perf_us": 24.5,
"t_hw_us": 3.04
},
"18": {
"baseline_perf_us": 19.18,
"t_hw_us": 2.96
},
"19": {
"baseline_perf_us": 19.09,
"t_hw_us": 2.96
},
"20": {
"baseline_perf_us": 130.73,
"t_hw_us": 86.59
}
}
}
}