Clone this repo:
  1. 38a9266 Create a utility library to suppress floating-point denormals, and apply it to every task execution of every thread. by Chao Mei · 12 days ago master
  2. 9394492 Simplify some code and add release assertions to help debug a crash in an application. by Benoit Jacob · 6 weeks ago
  3. b0e97e6 rollback hopefully fixing some application crash by Benoit Jacob · 6 weeks ago
  4. 54774a7 Use std::ptrdiff_t instead of int when calculating memory size to avoid int overflow. by Chao Mei · 7 weeks ago
  5. be760b6 Simplify quantized multiplier by Georgios Pinitas · 10 weeks ago

The ruy matrix multiplication library

This is not an officially supported Google product.

ruy is a matrix multiplication library. Its focus is to cover the matrix multiplication needs of neural network inference engines. Its initial user has been TensorFlow Lite, where it is used by default on the ARM CPU architecture.

ruy supports both floating-point and 8bit-integer-quantized matrices.


ruy is designed to achieve high performance not just on very large sizes, as is the focus of many established libraries, but on whatever are the actual sizes and shapes of matrices most critical in current TensorFlow Lite applications. This often means quite small sizes, e.g. 100x100 or even 50x50, and all sorts of rectangular shapes. It's not as fast as completely specialized code for each shape, but it aims to offer a good compromise of speed across all shapes and a small binary size.


Some documentation will eventually be available in the doc/ directory, see doc/