This is the original code of paper "Computation Error Analysis of Block Floating Point Arithmetic Oriented Convolution Neural Network Accelerator Design". This work is based on caffe. The function "caffe_cpu_gemm" is replaced to simulate the behavior of block floating point(BFP) on CPUs.
-
Notifications
You must be signed in to change notification settings - Fork 0
SongZRui/BlockFloatingPointCNN
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
About
This is the original code of paper "Computation Error Analysis of Block Floating Point Arithmetic Oriented Convolution Neural Network Accelerator Design"
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published