shuangchenli / processing-engine

A Scala w/ Chisel based implementation of a processing engine generator for neural network accelerators.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

processing-engine

A Scala w/ Chisel based implementation of a processing engine generator for neural network accelerators.

Introduction

The nPE is a highly configurable processing engine that can be topologically configured to support various DNN accelerator architectures. Its functionality is able to support inner products with varying amounts of spatial and temporal parallelism, nonlinear activation functions, scalar addition, scalar multiplication, and max functions. It supports many existing architectures and dataflows, such as those in Eyeriss, Flexflow, and ShiDianNao.

About

A Scala w/ Chisel based implementation of a processing engine generator for neural network accelerators.

License:MIT License


Languages

Language:Verilog 61.1%Language:Jupyter Notebook 26.7%Language:Scala 8.8%Language:Python 2.4%Language:Tcl 0.8%Language:JavaScript 0.2%