# BigDL-LLM Installation: GPU ## Windows ### Prerequisites BigDL-LLM on Windows supports Intel iGPU and dGPU. ```eval_rst .. important:: BigDL-LLM on Windows only supports PyTorch 2.1. ``` To apply Intel GPU acceleration, there're several prerequisite steps for tools installation and environment preparation: * Step 1: Install [Visual Studio 2022](https://visualstudio.microsoft.com/downloads/) Community Edition and select "Desktop development with C++" workload, like [this](https://learn.microsoft.com/en-us/cpp/build/vscpp-step-0-installation?view=msvc-170#step-4---choose-workloads) * Step 2: Install or update to latest [GPU driver](https://www.intel.com/content/www/us/en/download/785597/intel-arc-iris-xe-graphics-windows.html) * Step 3: Install [Intel® oneAPI Base Toolkit](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html) 2024.0 ### Install BigDL-LLM From PyPI We recommend using [miniconda](https://docs.conda.io/en/latest/miniconda.html) to create a python 3.9 enviroment: ```eval_rst .. important:: ``bigdl-llm`` is tested with Python 3.9, 3.10 and 3.11. Python 3.9 is recommended for best practices. ``` The easiest ways to install `bigdl-llm` is the following commands: ``` conda create -n llm python=3.9 libuv conda activate llm pip install --pre --upgrade bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu ``` ### Install BigDL-LLM From Wheel If you encounter network issues when installing IPEX, you can also install BigDL-LLM dependencies for Intel XPU from source achieves. First you need to download and install torch/torchvision/ipex from wheels listed below before installing `bigdl-llm`. Download the wheels on Windows system: ``` wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/torch-2.1.0a0%2Bcxx11.abi-cp39-cp39-win_amd64.whl wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/torchvision-0.16.0a0%2Bcxx11.abi-cp39-cp39-win_amd64.whl wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/intel_extension_for_pytorch-2.1.10%2Bxpu-cp39-cp39-win_amd64.whl ``` You may install dependencies directly from the wheel archives and then install `bigdl-llm` using following commands: ``` pip install torch-2.1.0a0+cxx11.abi-cp39-cp39-win_amd64.whl pip install torchvision-0.16.0a0+cxx11.abi-cp39-cp39-win_amd64.whl pip install intel_extension_for_pytorch-2.1.10+xpu-cp39-cp39-win_amd64.whl pip install --pre --upgrade bigdl-llm[xpu] ``` ```eval_rst .. note:: All the wheel packages mentioned here are for Python 3.9. If you would like to use Python 3.10 or 3.11, you should modify the wheel names for ``torch``, ``torchvision``, and ``intel_extension_for_pytorch`` by replacing ``cp39`` with ``cp310`` or ``cp311``, respectively. ``` ### Runtime Configuration To use GPU acceleration on Windows, several environment variables are required before running a GPU example. Make sure you are using CMD (Anaconda Prompt if using conda) as PowerShell is not supported: ```cmd call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat" ``` Please also set the following environment variable if you would like to run LLMs on: ```eval_rst .. tabs:: .. tab:: Intel iGPU .. code-block:: cmd set SYCL_CACHE_PERSISTENT=1 set BIGDL_LLM_XMX_DISABLED=1 .. tab:: Intel Arc™ A300-Series or Pro A60 .. code-block:: cmd set SYCL_CACHE_PERSISTENT=1 .. tab:: Other Intel dGPU Series There is no need to set further environment variables. ``` ```eval_rst .. note:: For **the first time** that **each model** runs on Intel iGPU/Intel Arc™ A300-Series or Pro A60, it may take several minutes to compile. ``` ### Troubleshooting #### 1. Error loading `intel_extension_for_pytorch` If you met error when importing `intel_extension_for_pytorch`, please ensure that you have completed the following steps: * Ensure that you have installed Visual Studio with "Desktop development with C++" workload. * Make sure that the correct version of oneAPI, specifically 2024.0, is installed. * Ensure that `libuv` is installed in your conda environment. This can be done during the creation of the environment with the command: ```cmd conda create -n llm python=3.9 libuv ``` If you missed `libuv`, you can add it to your existing environment through ```cmd conda install libuv ``` * Make sure you have configured oneAPI environment variables in your command prompt through ```cmd call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat" ``` Please note that you need to set these environment variables again once you have a new command prompt window. ## Linux ### Prerequisites BigDL-LLM for GPU supports on Linux has been verified on: * Intel Arc™ A-Series Graphics * Intel Data Center GPU Flex Series * Intel Data Center GPU Max Series ```eval_rst .. important:: BigDL-LLM on Linux supports PyTorch 2.0 and PyTorch 2.1. ``` ```eval_rst .. important:: We currently support the Ubuntu 20.04 operating system and later. ``` ```eval_rst .. tabs:: .. tab:: PyTorch 2.1 To enable BigDL-LLM for Intel GPUs with PyTorch 2.1, here are several prerequisite steps for tools installation and environment preparation: * Step 1: Install Intel GPU Driver version >= stable_775_20_20231219. We highly recommend installing the latest version of intel-i915-dkms using apt. .. seealso:: Please refer to our `driver installation `_ for general purpose GPU capabilities. See `release page `_ for latest version. * Step 2: Download and install `Intel® oneAPI Base Toolkit `_ with version 2024.0. OneDNN, OneMKL and DPC++ compiler are needed, others are optional. .. seealso:: We recommend you to use `this offline package `_ to install oneapi. .. tab:: PyTorch 2.0 To enable BigDL-LLM for Intel GPUs with PyTorch 2.0, here're several prerequisite steps for tools installation and environment preparation: * Step 1: Install Intel GPU Driver version >= stable_775_20_20231219. Highly recommend installing the latest version of intel-i915-dkms using apt. .. seealso:: Please refer to our `driver installation `_ for general purpose GPU capabilities. See `release page `_ for latest version. * Step 2: Download and install `Intel® oneAPI Base Toolkit `_ with version 2023.2. OneDNN, OneMKL and DPC++ compiler are needed, others are optional. .. seealso:: We recommend you to use `this offline package `_ to install oneapi. ``` ### Install BigDL-LLM From PyPI We recommend using [miniconda](https://docs.conda.io/en/latest/miniconda.html) to create a python 3.9 enviroment: ```eval_rst .. important:: ``bigdl-llm`` is tested with Python 3.9, 3.10 and 3.11. Python 3.9 is recommended for best practices. ``` ```eval_rst .. tabs:: .. tab:: PyTorch 2.1 .. code-block:: bash conda create -n llm python=3.9 conda activate llm pip install --pre --upgrade bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu .. note:: The ``xpu`` option will install BigDL-LLM with PyTorch 2.1 by default, which is equivalent to .. code-block:: bash pip install --pre --upgrade bigdl-llm[xpu_2.1] -f https://developer.intel.com/ipex-whl-stable-xpu .. tab:: PyTorch 2.0 .. code-block:: bash conda create -n llm python=3.9 conda activate llm pip install --pre --upgrade bigdl-llm[xpu_2.0] -f https://developer.intel.com/ipex-whl-stable-xpu ``` ### Install BigDL-LLM From Wheel If you encounter network issues when installing IPEX, you can also install BigDL-LLM dependencies for Intel XPU from source archives. First you need to download and install torch/torchvision/ipex from wheels listed below before installing `bigdl-llm`. ```eval_rst .. tabs:: .. tab:: PyTorch 2.1 .. code-block:: bash # get the wheels on Linux system for IPEX 2.1.10+xpu wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/torch-2.1.0a0%2Bcxx11.abi-cp39-cp39-linux_x86_64.whl wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/torchvision-0.16.0a0%2Bcxx11.abi-cp39-cp39-linux_x86_64.whl wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/intel_extension_for_pytorch-2.1.10%2Bxpu-cp39-cp39-linux_x86_64.whl Then you may install directly from the wheel archives using following commands: .. code-block:: bash # install the packages from the wheels pip install torch-2.1.0a0+cxx11.abi-cp39-cp39-linux_x86_64.whl pip install torchvision-0.16.0a0+cxx11.abi-cp39-cp39-linux_x86_64.whl pip install intel_extension_for_pytorch-2.1.10+xpu-cp39-cp39-linux_x86_64.whl # install bigdl-llm for Intel GPU pip install --pre --upgrade bigdl-llm[xpu] .. tab:: PyTorch 2.0 .. code-block:: bash # get the wheels on Linux system for IPEX 2.0.110+xpu wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/torch-2.0.1a0%2Bcxx11.abi-cp39-cp39-linux_x86_64.whl wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/torchvision-0.15.2a0%2Bcxx11.abi-cp39-cp39-linux_x86_64.whl wget https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_stable/xpu/intel_extension_for_pytorch-2.0.110%2Bxpu-cp39-cp39-linux_x86_64.whl Then you may install directly from the wheel archives using following commands: .. code-block:: bash # install the packages from the wheels pip install torch-2.0.1a0+cxx11.abi-cp39-cp39-linux_x86_64.whl pip install torchvision-0.15.2a0+cxx11.abi-cp39-cp39-linux_x86_64.whl pip install intel_extension_for_pytorch-2.0.110+xpu-cp39-cp39-linux_x86_64.whl # install bigdl-llm for Intel GPU pip install --pre --upgrade bigdl-llm[xpu_2.0] ``` ```eval_rst .. note:: All the wheel packages mentioned here are for Python 3.9. If you would like to use Python 3.10 or 3.11, you should modify the wheel names for ``torch``, ``torchvision``, and ``intel_extension_for_pytorch`` by replacing ``cp39`` with ``cp310`` or ``cp311``, respectively. ``` ### Runtime Configuration To use GPU acceleration on Linux, several environment variables are required or recommended before running a GPU example. ```eval_rst .. tabs:: .. tab:: Intel Arc™ A-Series and Intel Data Center GPU Flex For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series, we recommend: .. code-block:: bash # Required step. Configure oneAPI environment variables source /opt/intel/oneapi/setvars.sh # Recommended Environment Variables export USE_XETLA=OFF export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1 .. tab:: Intel Data Center GPU Max For Intel Data Center GPU Max Series, we recommend: .. code-block:: bash # Required step. Configure oneAPI environment variables source /opt/intel/oneapi/setvars.sh # Recommended Environment Variables export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1 export ENABLE_SDP_FUSION=1 Please note that ``libtcmalloc.so`` can be installed by ``conda install -c conda-forge -y gperftools=2.10`` ``` ### Known issues #### 1. Potential suboptimal performance with Linux kernel 6.2.0 For Ubuntu 22.04 and driver version < stable_775_20_20231219, the performance on Linux kernel 6.2.0 is worse than Linux kernel 5.19.0. You can use `sudo apt update && sudo apt install -y intel-i915-dkms intel-fw-gpu` to install the latest driver to solve this issue (need to reboot OS). Tips: You can use `sudo apt list --installed | grep intel-i915-dkms` to check your intel-i915-dkms's version, the version should be latest and >= `1.23.9.11.231003.15+i19-1`. #### 2. Driver installation unmet dependencies error: intel-i915-dkms The last apt install command of the driver installation may produce the following error: ``` The following packages have unmet dependencies: intel-i915-dkms : Conflicts: intel-platform-cse-dkms Conflicts: intel-platform-vsec-dkms ``` You can use `sudo apt install -y intel-i915-dkms intel-fw-gpu` to install instead. As the intel-platform-cse-dkms and intel-platform-vsec-dkms are already provided by intel-i915-dkms. ### Troubleshooting #### 1. Cannot open shared object file: No such file or directory Error where libmkl file is not found, for example, ``` OSError: libmkl_intel_lp64.so.2: cannot open shared object file: No such file or directory ``` ``` Error: libmkl_sycl_blas.so.4: cannot open shared object file: No such file or directory ``` The reason for such errors is that oneAPI has not been initialized properly before running BigDL-LLM code or before importing IPEX package. * Step 1: Make sure you execute setvars.sh of oneAPI Base Toolkit before running BigDL-LLM code. * Step 2: Make sure you install matching versions of BigDL-LLM/pytorch/IPEX and oneAPI Base Toolkit. BigDL-LLM with PyTorch 2.1 should be used with oneAPI Base Toolkit version 2024.0. BigDL-LLM with PyTorch 2.0 should be used with oneAPI Base Toolkit version 2023.2.