Implementing CNN Layers on the Manticore Cluster-Based Many-Core Architecture

TL;DR

This paper explores implementing CNN layers on the Manticore cluster-based many-core architecture, analyzing their characteristics and trade-offs for efficient neural network processing.

Contribution

It introduces specific implementations of CNN layers tailored for the Manticore architecture, highlighting their unique features and performance considerations.

Findings

Demonstrates effective CNN layer implementations on Manticore

Analyzes trade-offs in performance and resource utilization

Provides insights into architecture-specific optimizations

Abstract

This document presents implementations of fundamental convolutional neural network (CNN) layers on the Manticore cluster-based many-core architecture and discusses their characteristics and trade-offs.