Accelerated PyTorch Training on M1 Mac

This page summarizes the projects mentioned and recommended in the original post on news.ycombinator.com

InfluxDB - Power Real-Time Data Analytics at Scale
Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
www.influxdata.com
featured
SaaSHub - Software Alternatives and Reviews
SaaSHub helps you find the best software and product alternatives
www.saashub.com
featured
  • tensorexperiments

    Boilerplate for GPU-Accelerated TensorFlow and PyTorch code on M1 Macbook

  • Since it's tangentially relevant, if you have an M1 Mac I've created [some boilerplate](https://github.com/alexfromapex/tensorexperiments) for working with the latest Tensorflow with GPU acceleration as well.

  • neural-engine

    Everything we actually know about the Apple Neural Engine (ANE)

  • > Is too limited? Too hard to interact with? Not worth the effort?

    IIRC the only way to access ANE is through the Accelerate framework, and it seems to have pretty severe limitations[0].

    Apple has developed a tensorflow plugin[1] but i can't tell you if it uses ANE. Earlier this year they also published a job offer talking about accelerating PyTorch with BNNS and Accelerate[2]. Apparently PyTorch already uses Accelerate and AMX (the matrix coprocessor).

    So might indeed be that ANE is too limited and Accelerate never gets to use it.

    [0] https://github.com/hollance/neural-engine

    [1] https://developer.apple.com/metal/tensorflow-plugin/

    [2] https://github.com/pytorch/pytorch/issues/47702#issuecomment...

  • InfluxDB

    Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.

    InfluxDB logo
  • Pytorch

    Tensors and Dynamic neural networks in Python with strong GPU acceleration

  • > Is too limited? Too hard to interact with? Not worth the effort?

    IIRC the only way to access ANE is through the Accelerate framework, and it seems to have pretty severe limitations[0].

    Apple has developed a tensorflow plugin[1] but i can't tell you if it uses ANE. Earlier this year they also published a job offer talking about accelerating PyTorch with BNNS and Accelerate[2]. Apparently PyTorch already uses Accelerate and AMX (the matrix coprocessor).

    So might indeed be that ANE is too limited and Accelerate never gets to use it.

    [0] https://github.com/hollance/neural-engine

    [1] https://developer.apple.com/metal/tensorflow-plugin/

    [2] https://github.com/pytorch/pytorch/issues/47702#issuecomment...

  • cnn-benchmarks

    Benchmarks for popular CNN models

  • > ResNet > VGG: ResNet-50 is faster than VGG-16 and more accurate than VGG-19 (7.02 vs 9.0); ResNet-101 is about the same speed as VGG-19 but much more accurate than VGG-16 (6.21 vs 9.0).

    https://github.com/jcjohnson/cnn-benchmarks#:~:text=ResNet%2....

  • pytorch-apple-silicon-benchmarks

    Performance of PyTorch on Apple Silicon

  • I started collecting benchmarks of the M1 Max on PyTorch here: https://github.com/lucadiliello/pytorch-apple-silicon-benchm...

  • I started collecting benchmarks of the M1 Max on PyTorch here: https://github.com/lucadiliello/pytorch-apple-silicon-benchm...

NOTE: The number of mentions on this list indicates mentions on common posts plus user suggested alternatives. Hence, a higher number means a more popular project.

Suggest a related project

Related posts

  • PyTorch 2.3: User-Defined Triton Kernels, Tensor Parallelism in Distributed

    1 project | news.ycombinator.com | 10 May 2024
  • Clasificador de imágenes con una red neuronal convolucional (CNN)

    2 projects | dev.to | 1 May 2024
  • penzai: JAX research toolkit for building, editing, and visualizing neural nets

    4 projects | news.ycombinator.com | 21 Apr 2024
  • Tinygrad: Hacked 4090 driver to enable P2P

    5 projects | news.ycombinator.com | 12 Apr 2024
  • Functions and operators for Dot and Matrix multiplication and Element-wise calculation in PyTorch

    1 project | dev.to | 21 Mar 2024