Skip to content

M5 Optimizations using Metal Performance Primitives / Metal - Tensor engines? #2

Description

@vade

Hi

Firstly thank you for this work - this is super interesting. I'm curious if youve looked at the newer Metal Performance Primitives for dispatching metal work for running ML workload within a command buffer / command stream?

https://machinelearning.apple.com/research/exploring-llms-mlx-m5

https://developer.apple.com/download/files/Metal-Performance-Primitives-Programming-Guide.pdf

Im wondering if MLX would provide any optimizations to make this run on M5 much faster?

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions