[TensorFlow] TensorFlow and GEMM

Common Terms

BLAS is a specification and de-facto standard for linear algebra applications.
ACML, ATLAS, MKL, OpenBLAS are examples of BLAS libraries.
There are three levels to BLAS functionality. (Level 3 = GEMM)
GEMM stands for General Matrix Multiply.

Convolution and GEMM

Do im2col and then do matrix multiplication!
* im2col duplicates each section of the image multiple times; however, this waste of memory is outweighed by the performance gain we get by transforming convolution into regular matrix multiplication.
https://petewarden.com/2015/04/20/why-gemm-is-at-the-heart-of-deep-learning/

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.