RK356X 产品用户使用 NPU 前需要在终端使用 rsetup 开启 NPU:
sudo rsetup -> Overlays -> Manage overlays -> Enable NPU，最后重启系统。如 overlays 选项中无
Enable NPU选项，请通过:sudo rsetup -> System -> System Update升级系统, 重启后执行上述步骤开启 NPU。
因为GFW原因，升级一直失败啊，怎么解决？？骂人的话我就不讲了。cao
国内radxa升级系统太慢了不成功怎么办
系统启动后，在/boot/dtbo目录里面将rk3568-npu-enable.dtbo.disabled重命名为rk3568-npu-enable.dtbo，然后运行sudo u-boot-update
运行了sudo u-boot-update么？运行了就不需要用rsetup了
运行了。看来是不需要rsetup了。
我测试了一下yolo11,结果如下，这是用了NPU了没？
WARNING 
Unable to automatically guess model task, assuming ‘task=detect’. Explicitly define task for your model, i.e. ‘task=detect’, ‘segment’, ‘classify’,‘pose’ or ‘obb’.
Loading yolo11n_3566_rknn_model for RKNN inference…
W rknn-toolkit-lite2 version: 2.3.0
I RKNN: [09:31:06.536] RKNN Runtime Information, librknnrt version: 1.6.0 (9a7b5d24c@2023-12-13T17:31:11)
I RKNN: [09:31:06.536] RKNN Driver Information, version: 0.8.8
I RKNN: [09:31:06.538] RKNN Model Information, version: 6, toolkit version: 2.3.0(compiler version: 2.3.0 (c949ad889d@2024-11-07T11:39:30)), target: RKNPU lite, target platform: rk3566, framework name: ONNX, framework layout: NCHW, model inference type: static_shape
W RKNN: [09:31:06.538] RKNN Model version: 2.3.0 not match with rknn runtime version: 1.6.0
W RKNN: [09:31:06.634] query RKNN_QUERY_INPUT_DYNAMIC_RANGE error, rknn model is static shape type, please export rknn with dynamic_shapes
W Query dynamic range failed. Ret code: RKNN_ERR_MODEL_INVALID. (If it is a static shape RKNN model, please ignore the above warning message.)
image 1/1 /home/radxa/Documents/yolo/bus.jpg: 640x640 4 persons, 1 bus, 517.6ms
Speed: 48.6ms preprocess, 517.6ms inference, 22.2ms postprocess per image at shape (1, 3, 640, 640)
sudo dmesg | grep rknpu
[ 8.873992] OF: reserved mem: initialized node rknpu, compatible id shared-dma-pool
[ 11.654368] RKNPU fde40000.npu: RKNPU: rknpu iommu is enabled, using iommu mode
[ 11.654480] RKNPU fde40000.npu: Looking up rknpu-supply from device tree
[ 11.656537] [drm] Initialized rknpu 0.8.8 20230428 for fde40000.npu on minor 1
[ 11.656900] RKNPU fde40000.npu: Looking up rknpu-supply from device tree
看上去是正常的：I RKNN: [09:31:06.536] RKNN Driver Information, version: 0.8.8
NPU 驱动没问题，看起来是静态模型当动态用了，建议使用纯静态的方式，请参考：
和
https://docs.radxa.com/zero/zero3/app-development/rknn_toolkit_lite2_yolov5
有些无关紧要的warn,看末尾几行实际已经出结果了
就是感觉500ms+太慢了。回头用v8再试试。