TinySPICE plus: Scaling up statistical SPICE simulations on GPU leveraging shared-memory based sparse matrix solution techniques

Lengfei Han, Zhuo Feng

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

6 Scopus citations

Abstract

TinySPICE was a SPICE simulator on GPU developed to achieve dramatic speedups in statistical simulations of small nonlinear circuits, such as standard cell designs and SRAMs. While TinySPICE can perform circuit simulations much faster than traditional SPICE tools for small circuits, it may not be efficient for handling relatively large logic/memory circuit designs due to the embedded dense MNA matrix solver that can result in fast growing memory cost with increasing matrix size. In this work, we present TinySPICE Plus, a full-blown statistical SPICE simulation engine on GPU platform that integrates a highly-optimized shared-memory based sparse matrix solver that is capable of dealing with much larger circuits than TinySPICE while achieving orders of magnitude speedup over traditional CPU-based SPICE simulation engine. Extensive experimental results show that TinySPICE Plus can achieves over 70X speedups for parametric yield analysis of SRAM arrays and variation-aware logic circuit characterizations.

Original languageEnglish
Title of host publication2016 IEEE/ACM International Conference on Computer-Aided Design, ICCAD 2016
ISBN (Electronic)9781450344661
DOIs
StatePublished - 7 Nov 2016
Event35th IEEE/ACM International Conference on Computer-Aided Design, ICCAD 2016 - Austin, United States
Duration: 7 Nov 201610 Nov 2016

Publication series

NameIEEE/ACM International Conference on Computer-Aided Design, Digest of Technical Papers, ICCAD
Volume07-10-November-2016
ISSN (Print)1092-3152

Conference

Conference35th IEEE/ACM International Conference on Computer-Aided Design, ICCAD 2016
Country/TerritoryUnited States
CityAustin
Period7/11/1610/11/16

Fingerprint

Dive into the research topics of 'TinySPICE plus: Scaling up statistical SPICE simulations on GPU leveraging shared-memory based sparse matrix solution techniques'. Together they form a unique fingerprint.

Cite this