Georgia Institute of Technology
Optimizing LightningSim: Implementing Trace Compression and Analysis optimizations for acceleration of Simulation for High-Level Synthesis Designs
Abstract
dc:description.abstractHigh-level synthesis (HLS) synthesizes high level languages such as C/C++ into hardware description languages (HDL), saving development time on complex hardware designs and opening the door to hardware design to more developers. However, HLS is not conductive to agile development workflow due to slow synthesis and simulation times. LightningSim is a tool used for dynamic behavior modeling and cycle-level estimation of designs built for AMD Vitis HLS as a faster alternative to C/RTL cosimulation. This work aims to improve the developer experience for specialized hardware and facilitate high-level hardware development workflows by utilizing the highly repetitive execution nature of many HLS designs to optimize LightningSim for more efficient memory usage, trace generation, and analysis. This work introduces three optimizations into LightningSim: static fixed-bound loop analysis, trace compression and analysis, and a custom compressed trace file format. In the experiments, we show between 85.80\% and 99.87\% trace file compression, and up to 1.31x trace generation speedup and 4.35x trace analysis speedup.
Degree
thesis:*- Level thesis:degree_level
- Masters
- Department dc:contributor.department
- Electrical and Computer Engineering
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Paul, Rachel Lynne
- Advisor dc:contributor.advisor
-
- Hao, Cong
- Committee members dc:contributor.committeemember
-
- Krishna, Tushar
- Kim, Hyesoon
Subjects
dc:subject × 2Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/75317