Skip to content

README.md

Latest commit

 

History

History
91 lines (61 loc) · 1.76 KB

README.md

File metadata and controls

91 lines (61 loc) · 1.76 KB

SimpleMLBenchmark

For normal users use the "Normal Install" option, for people who want to test DirectML use the "DirectML Install"

Please read the Mode of executions on the end of the readme

Normal Install

  1. Install Python

  2. Clone this repo

  3. Install prerequisites

    pip install --upgrade numpy tqdm Pillow

  1. Install Pytroch

  2. Run the scoring system

    cd SimpleMLBenchmark
    python main.py

DirectML Install

Follow the steps below to get set up with PyTorch on DirectML.

WSL WIN

  1. Download and install Python 3.8.

  2. Install conda

  3. Clone this repo.

  4. Create conda env

    conda create --name pydml -y
    conda activate pydml
  1. Install prerequisites
    pip install torchvision==0.9.0
    pip uninstall torch
    pip install pytorch-directml

Note: The torchvision package automatically installs the torch==1.8.0 dependency, but this is not needed and will cause collisions with the pytorch-directml package. We must uninstall the torch package after installing requirements.

  1. Install other libraries
    pip install --upgrade numpy tqdm Pillow
  1. Run the scoring system
    cd SimpleMLBenchmark
    python main.py

Tests Modes

Autoselect

    python main.py

Force cuda / rocm

    python main.py --cuda

Force DirectML

    python main.py --dml

Force Appple AI Accelerator

    python main.py --apple