benchmark with m4ndock AX650

[erquren]

test in my new borad m4ndock

root@maixbox:/home/erquren/ncnn/benchmark# ./benchncnn loop_count = 4 num_threads = 8 powersave = 2 gpu_device = -1 cooling_down = 1 squeezenet min = 10.61 max = 10.70 avg = 10.65 squeezenet_int8 min = 13.24 max = 13.69 avg = 13.36 mobilenet min = 11.86 max = 17.97 avg = 13.50 mobilenet_int8 min = 12.18 max = 12.39 avg = 12.26 mobilenet_v2 min = 13.05 max = 13.25 avg = 13.13 mobilenet_v3 min = 11.73 max = 11.83 avg = 11.79 shufflenet min = 10.64 max = 10.78 avg = 10.70 shufflenet_v2 min = 9.52 max = 10.10 avg = 9.69 mnasnet min = 12.35 max = 12.45 avg = 12.42 proxylessnasnet min = 14.67 max = 14.89 avg = 14.74 efficientnet_b0 min = 18.44 max = 18.89 avg = 18.60 efficientnetv2_b0 min = 28.70 max = 38.43 avg = 31.21 regnety_400m min = 26.09 max = 26.31 avg = 26.20 blazeface min = 3.88 max = 3.90 avg = 3.89 googlenet min = 39.32 max = 40.25 avg = 39.70 googlenet_int8 min = 38.71 max = 39.35 avg = 38.93 resnet18 min = 26.48 max = 26.82 avg = 26.64 resnet18_int8 min = 28.24 max = 28.64 avg = 28.37 alexnet min = 26.66 max = 26.89 avg = 26.75 vgg16 min = 134.20 max = 137.33 avg = 136.20 vgg16_int8 min = 163.34 max = 166.19 avg = 164.29 resnet50 min = 60.32 max = 61.99 avg = 60.83 resnet50_int8 min = 65.30 max = 65.89 avg = 65.68 squeezenet_ssd min = 37.05 max = 37.44 avg = 37.16 squeezenet_ssd_int8 min = 40.16 max = 40.72 avg = 40.40 mobilenet_ssd min = 29.51 max = 30.12 avg = 29.71 mobilenet_ssd_int8 min = 26.21 max = 26.73 avg = 26.37 mobilenet_yolo min = 68.87 max = 70.13 avg = 69.62 mobilenetv2_yolov3 min = 47.03 max = 47.96 avg = 47.36 yolov4-tiny min = 71.71 max = 73.12 avg = 72.16 nanodet_m min = 27.14 max = 27.47 avg = 27.31 yolo-fastest-1.1 min = 14.04 max = 14.15 avg = 14.09 yolo-fastestv2 min = 11.90 max = 12.17 avg = 12.07 vision_transformer min = 1132.29 max = 1163.10 avg = 1143.91 FastestDet min = 11.51 max = 11.94 avg = 11.68

贴贴nihui

[nihui]

板子图片的背景p成白色 尺寸为长边300像素，按比例缩小，比如 280300 或者 300260 这样，保存为质量90的jpg格式 榜单排序按照 mobilenet 耗时，应该在 jetson nano 上面，rpi5 下面

[nihui]

图片有点太模糊了啊

[nihui]

谢谢～