Update doc to include building sherpa-onnx with GPU for Linux arm64. (#…

…685)

docs/source/onnx/install/linux.rst

@@ -30,7 +30,7 @@ All you need is to run:
         #
         make -j6
 
-   .. tab:: Nvidia GPU (CUDA)
+   .. tab:: Nvidia GPU (CUDA, x64)
 
       .. code-block:: bash
 
@@ -43,12 +43,48 @@ All you need is to run:
 
       .. hint::
 
-          You need to install CUDA toolkit. Otherwise, you would get
+          You need to install CUDA toolkit 11.8. Otherwise, you would get
           errors at runtime.
 
           You can refer to `<https://k2-fsa.github.io/k2/installation/cuda-cudnn.html>`_
           to install CUDA toolkit.
 
+   .. tab:: Nvidia GPU (CUDA 10.2, arm64, e.g., Jetson Nano B01)
+
+      .. code-block:: bash
+
+        git clone https://github.com/k2-fsa/sherpa-onnx
+        cd sherpa-onnx
+        mkdir build
+        cd build
+
+        cmake \
+          -DSHERPA_ONNX_LINUX_ARM64_GPU_ONNXRUNTIME_VERSION=1.11.0 \
+          -DCMAKE_BUILD_TYPE=Release \
+          -DBUILD_SHARED_LIBS=ON \
+          -DSHERPA_ONNX_ENABLE_GPU=ON \
+          ..
+
+        make
+
+   .. tab:: Nvidia GPU (CUDA 11.4, arm64, e.g., Jetson Orin NX)
+
+      .. code-block:: bash
+
+        git clone https://github.com/k2-fsa/sherpa-onnx
+        cd sherpa-onnx
+        mkdir build
+        cd build
+
+        cmake \
+          -DSHERPA_ONNX_LINUX_ARM64_GPU_ONNXRUNTIME_VERSION=1.16.0 \
+          -DCMAKE_BUILD_TYPE=Release \
+          -DBUILD_SHARED_LIBS=ON \
+          -DSHERPA_ONNX_ENABLE_GPU=ON \
+          ..
+
+        make
+
 After building, you will find an executable ``sherpa-onnx`` inside the ``bin`` directory.
 
 That's it!