PX2的一些资料

官方论坛：https://devtalk.nvidia.com/default/board/182/drive-px2/
官方网站：https://www.nvidia.com/en-us/self-driving-cars/ap2x/
开发网站：https://developer.nvidia.com/drive/downloads

docker

docker run --device=/dev/nvhost-ctrl --device=/dev/nvhost-ctrl-gpu --device=/dev/nvhost-prof-gpu --device=/dev/nvmap --device=/dev/nvhost-gpu --device=/dev/nvhost-as-gpu -v /usr/lib/aarch64-linux-gnu/tegra:/usr/lib/aarch64-linux-gnu/tegra -v /opt/zhangmm:/opt flacjacket/pytorch-tx2

测试情况：中止

我们当前的分割模型为mxnet及tensorrt，且使用python进行推断
mxnet可以在Nvidia的TX2上安装，但没有提供PX2版本
PX2虽然内置了Tensorrt，但却不支持python接口，只能用C++来调用：
-(2019年5月13号的回答)https://devtalk.nvidia.com/default/topic/1051664/general/tensorrt-python-example-code-for-reference-no-directory-for-python-as-per-documentation
-https://devtalk.nvidia.com/default/topic/1029055/tensorrt-3-0-python-api

note

gpu info: /usr/local/cuda-9.2/samples/1_Utilities/deviceQuery
resource usage info: sudo tegrastats. (website: https://devtalk.nvidia.com/default/topic/1022290/general/how-to-monitor-discrete-gpu-workload-usage-on-px2-/)
Please refer to “GR3D_FREQ xx%@1275” not GR3D_PCI for dGPU usage. Thanks. (https://devtalk.nvidia.com/default/topic/1035959/driveworks/tegrastatus-gr3d_pci-is-always-0-/)

gpu

https://devtalk.nvidia.com/default/topic/1036238/general/can-t-detect-the-dgpu-utilization-through-tegrastat/3
https://devtalk.nvidia.com/default/topic/1013464/jetson-tx2/gpu-out-of-memory-when-the-total-ram-usage-is-2-8g/post/5168834/#5168834
https://devtalk.nvidia.com/default/topic/1036238/general/can-t-detect-the-dgpu-utilization-through-tegrastat/post/5269344/#5269344
https://devtalk.nvidia.com/default/topic/1022290/general/how-to-monitor-discrete-gpu-workload-usage-on-px2-/
https://docs.nvidia.com/cuda/cuda-for-tegra-appnote/index.html#memory-management
https://devtalk.nvidia.com/default/topic/1049071/pytorch-for-jetson-nano-version-1-1-0-now-available/