From 107ae8499d0b65029e36e50d8ca5d44de206b0e5 Mon Sep 17 00:00:00 2001 From: tpoisonooo Date: Fri, 23 Sep 2022 10:14:00 +0800 Subject: [PATCH] docs(jetsons.md): fix desc (#1090) * docs(jetsons.md): fix desc * docs(jetsons.md): update --- docs/en/01-how-to-build/jetsons.md | 8 +++++--- docs/zh_cn/01-how-to-build/jetsons.md | 10 ++++++---- 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/docs/en/01-how-to-build/jetsons.md b/docs/en/01-how-to-build/jetsons.md index f995dea3a..add87bd9b 100644 --- a/docs/en/01-how-to-build/jetsons.md +++ b/docs/en/01-how-to-build/jetsons.md @@ -175,16 +175,18 @@ It takes about 1 hour 40 minutes to install MMCV on a Jetson Nano. So, please be #### Install ONNX ```shell +# Execute one of the following commands pip install onnx +conda install -c conda-forge onnx ``` -#### Install h5py +#### Install h5py and pycuda -Model Converter employs HDF5 to save the calibration data for TensorRT INT8 quantization. +Model Converter employs HDF5 to save the calibration data for TensorRT INT8 quantization and needs `pycuda` to copy device memory. ```shell sudo apt-get install -y pkg-config libhdf5-100 libhdf5-dev -pip install versioned-hdf5 +pip install versioned-hdf5 pycuda ``` ```{note} diff --git a/docs/zh_cn/01-how-to-build/jetsons.md b/docs/zh_cn/01-how-to-build/jetsons.md index accd986fb..973ffdf33 100644 --- a/docs/zh_cn/01-how-to-build/jetsons.md +++ b/docs/zh_cn/01-how-to-build/jetsons.md @@ -158,16 +158,18 @@ conda activate mmdeploy - 安装 ONNX ```shell - pip install onnx + # 以下方式二选一 + python3 -m pip install onnx + conda install -c conda-forge onnx ``` -- 安装 h5py +- 安装 h5py 和 pycuda - Model Converter 使用 HDF5 存储 TensorRT INT8 量化的校准数据。 + Model Converter 使用 HDF5 存储 TensorRT INT8 量化的校准数据；需要 pycuda 拷贝显存 ```shell sudo apt-get install -y pkg-config libhdf5-100 libhdf5-dev - pip install versioned-hdf5 + pip install versioned-hdf5 pycuda ``` ### 安装 SDK 的依赖项