In-Memory Computing Accelerator for Bioinformatics

General-Purpose Accelerator for Bioinformatics

Overview

This project focuses on designing and developing an efficient in-memory computing accelerator specifically tailored for bioinformatics workloads. By leveraging 3D DRAM technology and processing-in-memory principles, this accelerator aims to dramatically improve performance and energy efficiency for genomic and proteomic analysis tasks.

Key Objectives

• Design a hardware accelerator that exploits the unique computational patterns of bioinformatics applications
• Leverage 3D DRAM’s high bandwidth and memory density for near-data processing
• Minimize data movement between memory and processing units
• Support diverse multi-omics workloads (genomics, proteomics, metabolomics)

Collaborations

• Prof. Tajana S. Rosing (UCSD) - Principal Advisor
• Prof. Rob Knight’s group (UCSD) - Bioinformatics expertise and real-world workload characterization
• Niema Moshiri - Computational biology collaboration for phylogenetics and sequence analysis

Technologies & Tools

• 3D DRAM architecture and memory technology exploration
• FPGA prototyping and validation
• Hardware-software co-design methodology
• Python-based simulation and analysis

Impact

This work aims to bridge the gap between computational biology demands and hardware capabilities, enabling faster and more energy-efficient analysis of large-scale genomic and proteomic datasets.