Heterogenous in-memory accelerators combining memristor-based crossbar for matrix multiplication and content-addressable memories
Grant Data
Project Title
Heterogenous in-memory accelerators combining memristor-based crossbar for matrix multiplication and content-addressable memories
Principal Investigator
Professor Li, Can
(Principal Investigator (PI))
Co-Investigator(s)
Professor Wong Ngai
(Co-Investigator)
Strachan John Paul
(Principal investigator)
Duration
24
Start Date
2024-01-01
Amount
89600
Conference Title
Heterogenous in-memory accelerators combining memristor-based crossbar for matrix multiplication and content-addressable memories
Keywords
NA
Discipline
Electronics
HKU Project Code
G-HKU707/23
Grant Type
Germany/Hong Kong Joint Research Scheme 2023/24
Funding Year
2023
Status
On-going
Objectives
1. Develop a heterogeneous memristor array-based hardware accelerator chip that is performance and energy efficiency optimized for locality-sensitive hashing and similarity searches in addition to the matrix multiplication. 2. Architect the micro-structure of the heterogeneous system and establish performance benchmarks for comprehensive evaluation.3. Conduct workshops and presentations in our institute and our partner’s institute on ""Multitype Memristor Array-based Next-generation Heterogeneous AI Accelerator Architectures""