event

Ph.D. Dissertation Defense - James Read

Title:  Compute-in-Memory for Efficient AI: Cross-Layer Modeling and Hardware-Software Co-Design

Committee:

Dr. Shimeng Yu, ECE, Chair, Advisor

Dr. Asif Khan, ECE, Co-Advisor

Dr. Callie Hao, ECE

Dr. Sung-Kyu Lim, ECE

Dr. Muhannad Bakir, ECE

Dr. Ada Gavrilovska, Coc

Media

No media selected

Summary

Modern AI models spend much of their energy moving data between memory and processors. Compute-in-memory removes most of this movement by computing directly inside the memory arrays that store the model. The most efficient analog designs promise up to one hundred times better energy efficiency but are inherently noisy, and the noise erodes model accuracy. This dissertation develops open-source simulation tools that evaluate efficiency and accuracy together and uses them to compare memory technologies, compute circuits, packaging options, and AI model architectures.

Details

Monday
Jul 20 2026
01:00pm - 03:00pm

Location: Room 118, Kendeda

In campus calendar: No

Sidebar Content

No sidebar content

Groups

Status

  • Workflow status: Published
  • Created by: Daniela Staiculescu
  • Created: 07/09/2026
  • Modified By: Daniela Staiculescu
  • Modified: 07/09/2026

Related Links

Categories

Keywords

User Data

Target Audience