RocketCE for Power

 View Only
Back to discussions
Expand all | Collapse all

Releasing RocketCE OnnxRuntime v1.13.1 with OpenCE v1.8.0

  • 1.  Releasing RocketCE OnnxRuntime v1.13.1 with OpenCE v1.8.0

    ROCKETEER
    Posted 01-12-2023 00:27

    Overview 

    ONNX Runtime is a cross-platform inference and training machine-learning accelerator. 

    ONNX Runtime is an accelerator for machine learning models with multi-platform support and a flexible interface to integrate with hardware-specific libraries. ONNX Runtime can be used with models from PyTorch, Tensorflow/Keras, TFLite, scikit-learn, and other frameworks.  

    This release of RocketCE is based on Release OpenCE v1.8.0 and it includes ONNX Runtime v1.13.1. 

    These packages are supported for both Power 9 and Power 10 Systems of IBM with linux-ppc64le. 

     

    CPU Support 

    RocketCE includes CPU only variant of ONNX Runtime which is built without GPU support. This inclusion allows for training and inferencing to be done on Power 9 and Power 10 systems that do not have GPUs, or on systems where you want to train and inference without using the GPUs. 

    GPU Support 

    This version of Tensorflow will have support for GPU based operations. This can be used in Power 9 systems.

     

    Installation instructions 

    CPU Variant 

    To install CPU only variant of ONNX Runtime, run one of the following commands according to the python versions you have:

    • conda install -c rocketce onnxruntime=1.13.1=hea80eff_cpu_py38_pb3.19_1
    • conda install -c rocketce onnxruntime=1.13.1=hea80eff_cpu_py39_pb3.19_1
    • conda install -c rocketce onnxruntime=1.13.1=hea80eff_cpu_py310_pb3.19_1

     Once the installation is success, validate the installation by running the following command which lists the package onnxruntime which was pulled from rocketce channel  

    • Conda list onnxruntime 

    GPU Variant 

    To install GPU variant of ONNX Runtime, run one of the following commands according to the python versions you have:

    • conda install -c rocketce onnxruntime=1.13.1=h432c9ac_cuda11.2_py38_pb3.19_1   - For CUDA 11.2 version
    • conda install -c rocketce onnxruntime=1.13.1=h432c9ac_cuda11.2_py39_pb3.19_1    - For CUDA 11.2 version
    • conda install -c rocketce onnxruntime=1.13.1=h432c9ac_cuda11.2_py310_pb3.19_1   - For CUDA 11.2 version
    • conda install -c rocketce onnxruntime=1.13.1=h9f2cfa9_cuda11.4_py38_pb3.19_1    - For CUDA 11.4 version
    • conda install -c rocketce onnxruntime=1.13.1=h9f2cfa9_cuda11.4_py39_pb3.19_1    - For CUDA 11.4 version
    • conda install -c rocketce onnxruntime=1.13.1=h9f2cfa9_cuda11.4_py310_pb3.19_1   - For CUDA 11.4 version

       Once the installation is success, validate the installation by running the following command which lists the package onnxruntime which was pulled from rocketce channel  

      • Conda list onnxruntime 


      More information
      This conda package is using the Power 10 MMA capabilities. More information is available in Matrix-multiply assist (MMA) capabilities.


      ------------------------------
      Rajesh Nukala
      Rocket Internal - All Brands
      ------------------------------