AMD Ryzen 9 5950X 16-Core testing with a ASUS ROG CROSSHAIR VIII HERO (WI-FI) (4006 BIOS) and AMD Radeon RX 6700/6700 XT / 6800M on Ubuntu 22.04 via the Phoronix Test Suite.
Compare your own system(s) to this result file with the
Phoronix Test Suite by running the command:
phoronix-test-suite benchmark 2208131-PTS-NCNNMNN216
ncnn mnn 2022,
"NCNN 20220729 - Target: Vulkan GPU - Model: FastestDet",
Lower Results Are Better
"A",2.26,2.42,2.83
"B",2.25,2.25,2.35
"C",2.4,2.34,2.28,2.31,2.36,2.26,2.34,2.41,2.39,2.34,2.26,2.38,2.29,2.3
"D",2.3,2.36,2.37,2.31,2.38,2.35,2.27,2.28,2.72,2.32,2.34,2.28,2.78,2.3,2.34
"E",2.33,2.37,2.42,2.36,2.28,2.77,2.26,2.3,2.38,2.33,2.39,2.35,2.42,2.23,2.39
"NCNN 20220729 - Target: Vulkan GPU - Model: vision_transformer",
Lower Results Are Better
"A",219.88,221.06,221.3
"B",221.79,217.07,220.83
"C",220.71,220.25,219.63,219.53,219.59,219.07,217.88,218.06,219.65,219.16,219.26,223.29,220.11,218.27,219.12
"D",220.48,218.87,221.66,218.96,221.35,218.86,221.66,218.15,221.03,219.72,221.94,220.56,219.86,222.26,219.24
"E",219.66,222.17,221.21,221.2,221.96,219.62,220.6,219.63,219.08,220.87,221.65,220.45,220.48,219.54,224.6
"NCNN 20220729 - Target: Vulkan GPU - Model: regnety_400m",
Lower Results Are Better
"A",2.86,3.18,3.29
"B",2.89,3.26,2.93
"C",2.91,3.21,3.3,3.01,3.01,3.11,2.98,3,3.08,3.06,3.11,3.05,3.21,3.16,3.11
"D",2.92,3.29,3.12,3.21,3.03,3.04,3,3.16,3.02,3.29,3,2.94,3.11,3.1,3.02
"E",2.9,3.03,3.11,3.13,3.13,2.99,3.28,2.94,3,3.01,3.08,3.06,3.23,3.04,2.96
"NCNN 20220729 - Target: Vulkan GPU - Model: squeezenet_ssd",
Lower Results Are Better
"A",5.1,5.31,5.29
"B",4.94,5.27,5.22
"C",5.17,6.33,5.07,5.18,6.08,5.07,5.07,5.26,5.35,6.23,5.08,6.37,5.44,5.22,5.15
"D",5.2,5.16,5.34,5.06,6.29,5.05,6.3,5.1,5.12,5.4,6.35,5.37,5.65,5.91,5.16
"E",4.96,5.08,6.78,5.37,5.14,6.14,5.23,5.62,5.16,5.33,5.32,5.42,5.21,6.26,6.18
"NCNN 20220729 - Target: Vulkan GPU - Model: yolov4-tiny",
Lower Results Are Better
"A",14.62,14.62,14.49
"B",14.39,14.65,14.39
"C",14.92,14.48,14.46,14.49,14.72,14.55,14.77,14.36,14.61,14.58,14.43,14.45,14.48,14.64,14.42
"D",14.88,14.53,14.55,14.61,14.48,14.39,14.73,14.55,14.83,14.55,14.44,14.61,14.53,14.62,14.62
"E",14.36,14.57,14.57,14.49,14.55,14.48,14.47,14.45,14.36,14.75,14.49,14.5,18.24,14.53,14.68
"NCNN 20220729 - Target: Vulkan GPU - Model: resnet50",
Lower Results Are Better
"A",5.01,5,4.95
"B",5,4.9,4.93
"C",5.02,5.13,5.18,4.99,5.17,4.86,4.89,5.03,4.94,5.05,4.91,4.96,4.92,4.95,4.98
"D",5.03,5.02,5.35,4.83,4.97,4.95,5,4.88,5.26,4.92,5.04,5,4.91,4.79,5.09
"E",4.8,5.13,5.04,5.04,4.95,4.93,5,4.81,4.92,5.14,5.03,5.06,4.96,4.92,4.8
"NCNN 20220729 - Target: Vulkan GPU - Model: alexnet",
Lower Results Are Better
"A",1.97,2.17,2.09
"B",2.31,2.33,2.08
"C",2.11,2.04,2.2,2.03,1.9,2.07,2.37,2.1,2.11,2.19,2.06,2.03,2.06,2.07,2.03
"D",2.04,2.02,1.95,2.04,2.1,2.13,2.01,2.05,2.13,2.06,2.15,2.22,2.16
"E",1.98,2.13,2.11,2.2,2.27,1.99,2.08,2.03,2.09,2.15,2.32,2.13
"NCNN 20220729 - Target: Vulkan GPU - Model: resnet18",
Lower Results Are Better
"A",2.56,2.67,2.65
"B",2.5,2.54,2.58
"C",2.61,2.57,2.67,2.51,2.51,2.69,2.54,2.64,2.69,2.69,2.66,2.57,2.5,2.58,2.62
"D",2.59,2.49,2.73,2.45,2.64,2.64,2.76,2.47,2.45,2.65,2.72,2.7,2.45,2.63,2.78
"E",2.41,2.72,2.63,2.69,2.61,2.41,2.56,2.59,2.76,2.84,2.69,2.61,2.58
"NCNN 20220729 - Target: Vulkan GPU - Model: vgg16",
Lower Results Are Better
"A",6.68,6.81,6.83
"B",6.86,6.88,7.05
"C",6.81,6.66,6.68,6.63,6.55,6.74,6.96,6.89,6.83,6.83,6.8,6.67,6.65,6.78,6.75
"D",6.68,6.77,6.77,6.6,6.74,6.8,6.74,6.72,6.72,6.84,6.86,6.88,6.77,6.65,6.66
"E",6.76,6.75,6.83,6.76,6.85,6.84,6.77,6.69,6.66,6.87,6.81,6.77,6.89,6.81,6.9
"NCNN 20220729 - Target: Vulkan GPU - Model: googlenet",
Lower Results Are Better
"A",3.66,3.65,3.69
"B",3.59,3.86,3.69
"C",3.77,3.68,3.66,3.69,3.62,3.78,3.41,3.65,3.74,3.65,3.86,3.51,3.69,3.64,3.58
"D",3.6,4.06,3.63,3.74,3.66,3.72,3.73,3.66,3.64,3.71,3.9,3.59,3.85,3.66,3.52
"E",3.65,3.61,3.75,3.92,3.61,3.53,3.91,3.73,3.7,3.74,3.83,3.75,3.74,3.65,3.65
"NCNN 20220729 - Target: Vulkan GPU - Model: blazeface",
Lower Results Are Better
"A",1.6,1.61,1.72
"B",1.61,1.63,1.57
"C",1.69,1.63,1.62,1.58,1.63,1.74,1.65,1.61,1.68,1.67,1.75,1.68,1.62,1.65,1.56
"D",1.62,1.61,1.68,1.56,1.65,1.66,1.72,1.64,1.55,1.74,1.59,1.67,1.66,1.61,1.59
"E",1.54,1.69,1.63,1.65,1.67,1.64,1.69,1.56,1.75,1.68,1.67,1.68,1.74,1.73,1.57
"NCNN 20220729 - Target: Vulkan GPU - Model: efficientnet-b0",
Lower Results Are Better
"A",5,4.86,4.92
"B",4.99,4.84,4.83
"C",4.88,4.85,4.89,4.84,4.95,4.98,4.83,4.84,5.06,5.06,4.92,4.88,4.85,4.86,4.84
"D",4.78,4.85,4.85,4.84,4.83,4.85,4.86,4.83,4.83,4.91,4.96,4.92,4.84,4.84,4.86
"E",4.82,4.94,4.86,4.86,4.93,4.83,4.86,4.85,4.89,4.86,4.85,4.88,4.86,4.82,4.85
"NCNN 20220729 - Target: Vulkan GPU - Model: mnasnet",
Lower Results Are Better
"A",1.93,1.97,2
"B",1.99,2.04,2.1
"C",1.93,2.03,2,1.98,2.03,2.04,2.04,2.04,2.05,1.99,2.03,1.94,1.97,2.11,2.04
"D",1.87,2.09,2.04,2.12,1.96,1.99,2.02,2.08,2.05,2,2.07,2.02,2.17
"E",1.99,1.96,2.04,1.96,2.03,2,2.04,2.02,1.97,2.08,2.22,2.03,2.08,2.08
"NCNN 20220729 - Target: Vulkan GPU - Model: shufflenet-v2",
Lower Results Are Better
"A",1.9,2.08,2.14
"B",2.12,2.08,2.15
"C",1.97,2.01,2.05,2.07,2.18,2.12,2.1,2.21,2.14,2.06,2.12,2.01,2.12,2.14,2.01
"D",1.99,2.13,2.19,2.13,2.11,2,2.15,2.06,2.03,2.09,2.17,2.01,2.12,2.01,2.1
"E",2.04,2.08,2.12,2.05,2.02,2.02,2.18,2.12,2.05,2.13,2.12,2.11,2.04,2.09,2.11
"NCNN 20220729 - Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3",
Lower Results Are Better
"A",2.36,2.31,2.38
"B",2.35,2.39,2.33
"C",2.2,2.42,2.34,2.41,2.3,2.34,2.28,2.35,2.35,2.41,2.35,2.34,2.33,2.41,2.33
"D",2.29,2.37,2.38,2.38,2.4,2.38,2.35,2.36,2.38,2.26,2.41,2.44,2.32,2.33,2.31
"E",2.28,2.44,2.56,2.39,2.44,2.45,2.28,2.43,2.4,2.63,2.47,2.39,2.38,2.41,2.25
"NCNN 20220729 - Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2",
Lower Results Are Better
"A",1.83,1.96,1.96
"B",1.86,2.01,1.98
"C",1.79,1.99,1.96,1.95,2.04,2.02,1.97,1.98,1.92,1.97,1.97,1.86,2.45,1.94,2.02
"D",1.84,1.93,2.01,1.98,1.98,2.04,2.01,1.93,2.02,1.98,1.99,2.05,1.93,2,1.97
"E",1.94,2.04,1.98,2.04,2.15,2.01,1.9,1.96,1.99,1.98,1.96,2.02,2.07,1.98,2.04
"NCNN 20220729 - Target: Vulkan GPU - Model: mobilenet",
Lower Results Are Better
"A",9.42,9.24,9.68
"B",9.41,9.14,9.48
"C",9.07,11.18,9.12,8.78,9.57,9.11,10.25,10.82,9.56,11.08,11.06,10.06,9.36,9.27,9.25
"D",11.27,10.07,9.79,9.56,9.26,9.91,9.59,9.81,9.25,9.9,9.94,11.28,9.41,9.19,10.23
"E",11.17,9.32,9.44,9.11,11.19,9.35,10.24,11.23,11.38,11.06,8.83,9.61,9.68,9.89,9.31
"Mobile Neural Network 2.0 - Model: inception-v3",
Lower Results Are Better
"A",25.752,26.387,26.082
"B",25.676,25.859,25.882
"C",25.956,25.796,25.963
"D",25.287,26.041,25.93,26.212,25.976,26.343,25.946,25.758,26.4,26.096,25.566,25.867,26.229,25.633,25.575
"E",26.164,25.876,25.671,25.86,25.511,25.871,26.03,25.943,26.42,25.997,26.222,25.884,25.961,26.246,26.192
"Mobile Neural Network 2.0 - Model: mobilenet-v1-1.0",
Lower Results Are Better
"A",2.604,2.605,2.667
"B",2.523,2.578,2.524
"C",2.616,2.542,2.511
"D",2.467,2.627,2.626,2.567,2.548,2.645,2.571,2.844,2.69,2.624,2.452,2.581,2.61,2.461,2.567
"E",2.493,2.597,2.628,2.496,2.473,2.659,2.541,2.569,2.612,2.635,2.491,2.602,2.589,2.698,2.637
"Mobile Neural Network 2.0 - Model: MobileNetV2_224",
Lower Results Are Better
"A",3.427,3.463,3.458
"B",3.413,3.459,3.394
"C",3.453,3.479,3.472
"D",3.298,3.475,3.495,3.434,3.552,3.579,3.46,3.587,3.525,3.424,3.33,3.482,3.496,3.246,3.451
"E",3.366,3.549,3.442,3.508,3.224,3.535,3.534,3.403,3.506,3.46,3.426,3.568,3.515,3.586,3.457
"Mobile Neural Network 2.0 - Model: SqueezeNetV1.0",
Lower Results Are Better
"A",5.203,5.293,5.207
"B",5.148,5.243,5.04
"C",5.3,5.262,5.489
"D",5.137,5.41,5.461,5.281,5.453,5.465,5.477,5.26,5.429,5.313,5.085,5.305,5.436,4.841,5.175
"E",5.46,5.284,5.222,5.687,5.255,5.485,5.475,5.167,5.7,5.214,5.283,5.486,5.31,5.402,5.296
"Mobile Neural Network 2.0 - Model: resnet-v2-50",
Lower Results Are Better
"A",21.971,22.046,21.852
"B",20.854,21.337,22.502
"C",20.905,21.01,21.379
"D",20.997,21.518,21.835,22.182,22.092,22.061,21.51,21.906,22.196,21.372,21.234,21.693,22.018,21.208,21.735
"E",21.248,21.304,21.87,21.615,21.11,21.499,22.101,21.463,21.737,21.504,20.965,21.666,21.856,21.867,21.236
"Mobile Neural Network 2.0 - Model: squeezenetv1.1",
Lower Results Are Better
"A",3.112,3.366,3.15
"B",3.263,3.305,3.114
"C",3.197,3.282,3.355
"D",3.125,3.375,3.407,3.365,3.363,3.387,3.364,3.377,3.399,3.127,3.11,3.376,3.392,2.982,3.322
"E",3.359,3.333,3.328,3.377,3.165,3.376,3.368,3.059,3.353,3.321,3.134,3.381,3.337,3.391,3.158
"Mobile Neural Network 2.0 - Model: mobilenetV3",
Lower Results Are Better
"A",1.87,1.939,1.899
"B",1.847,1.88,1.854
"C",1.836,1.877,1.881
"D",1.799,1.931,1.898,1.944,1.961,1.951,1.961,1.977,1.966,1.861,1.81,1.916,1.962,1.838,1.914
"E",1.839,1.921,1.982,1.843,1.805,1.947,1.887,1.833,1.941,1.93,1.82,1.943,1.971,1.956,1.905
"NCNN 20220729 - Target: CPU - Model: FastestDet",
Lower Results Are Better
"A",5.15,5,4.78
"B",4.91,5.03,4.78,4.45,4.79,5.13,4.82,4.99,5.01,4.93,5.14,4.78,4.9,4.94,4.9
"C",5.04,4.77,4.97
"D",4.9,4.84,4.85
"E",4.87,4.49,5.15
"NCNN 20220729 - Target: CPU - Model: vision_transformer",
Lower Results Are Better
"A",123.66,123.53,123.45
"B",123.06,122.93,122.71,122.87,122.78,122.64,122.84,123,122.14,123.43,122.41,123.52,122.53,122.69,123.87
"C",122.71,122.89,122.3
"D",123.19,124.41,124.02
"E",122.69,124.65,122.99
"NCNN 20220729 - Target: CPU - Model: regnety_400m",
Lower Results Are Better
"A",12.8,12.91,12.79
"B",12.92,12.69,12.79,12.74,12.91,12.92,13,12.66,12.75,12.95,13.02,12.71,12.99,12.36,12.73
"C",12.82,12.93,12.96
"D",12.83,12.91,12.8
"E",12.9,12.59,12.87
"NCNN 20220729 - Target: CPU - Model: squeezenet_ssd",
Lower Results Are Better
"A",18.29,18.65,18.01
"B",19.08,17.82,18.53,18.49,18.77,19.03,18.14,18.28,19.3,19.28,18.39,18.19,18.84,17.84,18.04
"C",18.32,18.64,18.81
"D",18.21,18.72,18.24
"E",17.9,18.53,18.07
"NCNN 20220729 - Target: CPU - Model: yolov4-tiny",
Lower Results Are Better
"A",21.35,21.78,21.4
"B",21.66,21.18,21.34,21.22,21.2,21.47,21.01,21.08,21.59,21.7,21.03,21.7,22.07,21.07,20.65
"C",21.8,21.32,21.59
"D",21.34,21.22,21.33
"E",21.21,21.7,21.39
"NCNN 20220729 - Target: CPU - Model: resnet50",
Lower Results Are Better
"A",21.03,21.98,21.41
"B",20.41,20.64,21.44,20.52,21.38,21.58,21.25,21.41,20.98,20.59,21.05,21.54,20.98,20.38,20.6
"C",21.42,21.92,21.75
"D",21.32,21.46,21.65
"E",21.16,21.26,21.76
"NCNN 20220729 - Target: CPU - Model: alexnet",
Lower Results Are Better
"A",7.7,7.8,7.87
"B",7.84,7.66,7.7,7.63,7.63,7.89,7.87,7.92,7.53,7.57,7.73,7.82,7.7,7.79,7.73
"C",7.84,7.54,7.62
"D",7.67,7.74,7.76
"E",12.41,7.77,7.72
"NCNN 20220729 - Target: CPU - Model: resnet18",
Lower Results Are Better
"A",12.19,12.14,12.18
"B",12.27,12.31,12.1,12.13,12.15,12.29,12.08,12.13,12.11,12.01,12.62,12.22,12.66,12.05,12.07
"C",12.15,12.08,12.02
"D",12.41,11.83,12.23
"E",11.99,12.15,12.27
"NCNN 20220729 - Target: CPU - Model: vgg16",
Lower Results Are Better
"A",47.64,47.32,47.77
"B",47.89,46.8,46.93,47.24,47.53,48.34,47.54,48.15,47.36,47.6,47.99,47.14,47.56,46.97,47.73
"C",47.64,47.58,47.51
"D",47.35,47.19,47.48
"E",47.41,47.68,48.72
"NCNN 20220729 - Target: CPU - Model: googlenet",
Lower Results Are Better
"A",12.59,11.57,12.4
"B",11.69,11.6,12.24,11.51,11.38,11.44,11.62,11.36,11.35,11.41,12.29,11.42,12.22,11.31,12.45
"C",11.36,11.38,11.4
"D",11.83,11.43,11.42
"E",11.35,11.51,11.74
"NCNN 20220729 - Target: CPU - Model: blazeface",
Lower Results Are Better
"A",1.83,1.83,1.81
"B",1.8,1.76,1.81,1.79,1.84,1.85,1.83,1.8,1.78,1.81,1.84,1.8,1.82,1.73,1.76
"C",1.81,1.82,1.82
"D",1.97,1.86,1.82
"E",1.84,1.81,1.84
"NCNN 20220729 - Target: CPU - Model: efficientnet-b0",
Lower Results Are Better
"A",5.9,5.95,5.88
"B",5.9,5.87,5.86,5.96,5.9,5.92,6.01,5.86,5.89,5.9,6.19,5.86,5.92,6.2,6.25
"C",5.88,5.89,5.92
"D",6,6.16,5.89
"E",5.91,5.89,5.91
"NCNN 20220729 - Target: CPU - Model: mnasnet",
Lower Results Are Better
"A",3.9,3.9,3.88
"B",3.89,3.89,3.87,3.9,3.9,3.91,3.96,3.89,3.9,3.9,4.6,3.88,3.93,4.17,3.91
"C",3.87,3.89,3.91
"D",3.93,3.89,3.88
"E",3.87,3.91,3.91
"NCNN 20220729 - Target: CPU - Model: shufflenet-v2",
Lower Results Are Better
"A",4.27,4.32
"B",4.3,4.32,4.28,4.22,4.32,4.31,4.3,4.22,4.3,4.28,4.4,4.35,4.34,4.26
"C",4.28,4.32,4.31
"D",4.27,4.28,4.24
"E",4.28,4.27,4.29
"NCNN 20220729 - Target: CPU-v3-v3 - Model: mobilenet-v3",
Lower Results Are Better
"A",3.78,3.78,3.77
"B",3.74,3.77,3.75,3.89,3.79,3.77,3.82,3.76,3.76,3.78,4.39,3.77,3.81,3.75,3.76
"C",3.75,3.78,3.77
"D",3.79,3.76,3.76
"E",3.74,3.76,3.77
"NCNN 20220729 - Target: CPU-v2-v2 - Model: mobilenet-v2",
Lower Results Are Better
"A",4.29,4.31,4.27
"B",4.28,4.27,4.27,4.31,4.3,4.34,4.31,4.28,4.28,4.3,4.94,4.26,4.31,4.28,4.28
"C",4.26,4.29,4.29
"D",4.31,4.35,4.28
"E",4.31,4.29,4.3
"NCNN 20220729 - Target: CPU - Model: mobilenet",
Lower Results Are Better
"A",11.18,11.75,11.44
"B",11.18,16.49,11.85,11.19,11.21,12.01,11.56,11.46,12.37,11.24,12.51,11.67,11.55,11.71,10.87
"C",11.7,11.95,11.47
"D",11.6,11.86,12.07
"E",11.55,11.7,11.67