Research

From NANOxCOMP H2020 Project
(Difference between revisions)
Jump to: navigation, search
Line 228: Line 228:
 
We propose a method to overcome the main drawback in stochastic computing, '''low accuracy''' or related '''long computing times'''. Our method manipulates stochastic bit streams with the aid of feedback mechanisms. We implement error-free arithmetic multiplier and adder circuits by considering performance parameters area, delay, and accuracy.
 
We propose a method to overcome the main drawback in stochastic computing, '''low accuracy''' or related '''long computing times'''. Our method manipulates stochastic bit streams with the aid of feedback mechanisms. We implement error-free arithmetic multiplier and adder circuits by considering performance parameters area, delay, and accuracy.
  
[[Image:Research-stochastic-1.png|center|none|800px|link=]]
+
[[Image:nanoxcomp_RC_modeling.png|center|none|600px|link=]]
  
 
<!--        YAYIN      -->
 
<!--        YAYIN      -->

Revision as of 22:56, 12 April 2017

We aim to develop a complete synthesis and performance optimization methodology for switching nano-crossbar arrays that leads to the design and construction of an emerging nanocomputer. Our objectives are 1) synthesizing Boolean functions with area optimization; 2) achieving fault tolerance; 3) performing performance optimization by considering area, delay, power, and accuracy; 4) implementing arithmetic and memory elements; and 5) realizing a synchronous state machine.

Contents

Logic Synthesis

We study implementation of Boolean functions with nano-crossbar arrays where each crosspoint behaves as a diode, a FET, and a four-terminal switch. For these three types, we give array size formulations for a given Boolean function. Additionally, we focus on four-terminal switch based implementations and propose an algorithm that implements Boolean functions with optimal array sizes.

Nanoarray logic synthesis.png


Selected Publications
title: Synthesis on Switching Lattices of Dimension-Reducible Boolean Functions
authors: Anna Bernasconi, Valentina Ciriani, Luca Frontini, and Gabriella Trucco
presented at: International Conference on Very Large Scale Integration (VLSI-SoC),
Tallinn, Estonia, 2016

PDF.png
Paper

PPT.jpg
Slides

title: Synthesis and Performance Optimization of a Switching Nano-crossbar Computer
authors: Dan Alexandrescu, Mustafa Altun, Lorena Anghel, Anna Bernasconi,
Valentina Ciriani, and Mehdi Tahoori
presented at: Euromicro Conference on Digital System Design (DSD),
Limassol, Cyprus, 2016.

PDF.png
Paper

PDF.png
Slides

Developed Tools
title: Optimal Synthesis Tool
authors: Ceylan Morgul and Mustafa Altun
description: Two optimal synthesis tools Tool-1 and Tool-2 are developed in Matlab and Python, respectively. Both tools aim to synthesize a given target Boolean functions with an optimal size of four-terminal switch based arrays .

ZIP.png
Tool


Fault Tolerance

We examine reconfigurable crossbar arrays by considering randomly occurred stuck-open and stuck-closed crosspoint faults. In the presence of permanent faults, a fast and accurate heuristic algorithm is proposed that uses the techniques of index sorting, backtracking, and row matching. In the presence of transient faults, tolerance analysis is performed by formally and recursively determining tolerable fault positions

Nanoarray fault tolerance.png


Selected Publications
title: Permanent and Transient Fault Tolerance for Reconfigurable Nano-Crossbar Arrays
authors: Onur Tunali and Mustafa Altun
accepted in: IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2016.

PDF.png
Paper

Developed Tools
title: Fault Tolerant Logic Mapping Tool
authors: Onur Tunali and Mustafa Altun
description: The tool is developed in Matlab. It aims to map logic funtions into fault crossbars such that each crosspoint has an independent fault probability up to 20%.

ZIP.png
Tool


Performance Modeling and Analysis

Accurate Arithmetic Implementations

We propose a method to overcome the main drawback in stochastic computing, low accuracy or related long computing times. Our method manipulates stochastic bit streams with the aid of feedback mechanisms. We implement error-free arithmetic multiplier and adder circuits by considering performance parameters area, delay, and accuracy.

Selected Publications
title: Power-Delay-Area Performance Modeling and Analysis for Nano-Crossbar Arrays
authors: Ceylan Morgul, Furkan Peker, and Mustafa Altun
presented at: IEEE Computer Society Annual Symposium on VLSI (ISVLSI),
Pittsburgh, USA, 2016.

PDF.png
Paper

PPT.jpg
Poster

Personal tools
Namespaces

Variants
Actions
NANOxCOMP
Toolbox