From 07c00d153ec2bd12e0fe4372e857ac8b04816af3 Mon Sep 17 00:00:00 2001
From: Fangjun Kuang <csukuangfj@gmail.com>
Date: Tue, 23 Jul 2024 18:32:57 +0800
Subject: [PATCH] Add doc for installing CUDA 12.4 (#1296)

Note that torch 2.4 supports CUDA 12.4
---
 docs/source/installation/cuda-cudnn.rst | 77 +++++++++++++++++++++++++
 1 file changed, 77 insertions(+)

diff --git a/docs/source/installation/cuda-cudnn.rst b/docs/source/installation/cuda-cudnn.rst
index 5dc78b4d5..01f4139b2 100644
--- a/docs/source/installation/cuda-cudnn.rst
+++ b/docs/source/installation/cuda-cudnn.rst
@@ -692,3 +692,80 @@ The output should look like the following:
   Built on Tue_Feb__7_19:32:13_PST_2023
   Cuda compilation tools, release 12.1, V12.1.66
   Build cuda_12.1.r12.1/compiler.32415258_0
+
+CUDA 12.4
+---------
+
+You can use the following commands to install CUDA 12.4. We install it
+into ``/star-fj/fangjun/software/cuda-12.4.0``. You can replace it
+if needed.
+
+.. code-block:: bash
+
+  wget https://developer.download.nvidia.com/compute/cuda/12.4.0/local_installers/cuda_12.4.0_550.54.14_linux.run
+
+  chmod +x cuda_12.4.0_550.54.14_linux.run
+
+  ./cuda_12.4.0_550.54.14_linux.run \
+    --silent \
+    --toolkit \
+    --installpath=/star-fj/fangjun/software/cuda-12.4.0 \
+    --no-opengl-libs \
+    --no-drm \
+    --no-man-page
+
+Install cuDNN for CUDA 12.4
+^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Now, install ``cuDNN`` for CUDA 12.4.
+
+.. code-block:: bash
+
+  wget https://huggingface.co/csukuangfj/cudnn/resolve/main/cudnn-linux-x86_64-8.9.7.29_cuda12-archive.tar.xz
+
+  tar xvf cudnn-linux-x86_64-8.9.7.29_cuda12-archive.tar.xz --strip-components=1 -C /star-fj/fangjun/software/cuda-12.4.0
+
+Set environment variables for CUDA 12.4
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Note that we have to set the following environment variables after installing
+CUDA 11.8. You can save the following code to ``activate-cuda-12.4.sh``
+and use ``source activate-cuda-12.4.sh`` if you want to activate CUDA 12.4.
+
+.. code-block:: bash
+
+  export CUDA_HOME=/star-fj/fangjun/software/cuda-12.4.0
+  export PATH=$CUDA_HOME/bin:$PATH
+  export LD_LIBRARY_PATH=$CUDA_HOME/lib64:$LD_LIBRARY_PATH
+  export LD_LIBRARY_PATH=$CUDA_HOME/lib:$LD_LIBRARY_PATH
+  export LD_LIBRARY_PATH=$CUDA_HOME/extras/CUPTI/lib64:$LD_LIBRARY_PATH
+  export CUDAToolkit_ROOT_DIR=$CUDA_HOME
+  export CUDAToolkit_ROOT=$CUDA_HOME
+
+  export CUDA_TOOLKIT_ROOT_DIR=$CUDA_HOME
+  export CUDA_TOOLKIT_ROOT=$CUDA_HOME
+  export CUDA_BIN_PATH=$CUDA_HOME
+  export CUDA_PATH=$CUDA_HOME
+  export CUDA_INC_PATH=$CUDA_HOME/targets/x86_64-linux
+  export CFLAGS=-I$CUDA_HOME/targets/x86_64-linux/include:$CFLAGS
+  export CUDAToolkit_TARGET_DIR=$CUDA_HOME/targets/x86_64-linux
+
+To check that you have installed CUDA 12.4 successfully, please run:
+
+.. code-block:: bash
+
+  which nvcc
+
+  nvcc --version
+
+The output should look like the following:
+
+.. code-block:: bash
+
+  /star-fj/fangjun/software/cuda-12.4.0/bin/nvcc
+
+  nvcc: NVIDIA (R) Cuda compiler driver
+  Copyright (c) 2005-2024 NVIDIA Corporation
+  Built on Tue_Feb_27_16:19:38_PST_2024
+  Cuda compilation tools, release 12.4, V12.4.99
+  Build cuda_12.4.r12.4/compiler.33961263_0