Fall 2025

Priya Narasimhan, AMD Research

Co-Designing Inference Accelerators for Rapidly Evolving Models

4:00 PM · 32-G449 · In person

Lessons from deploying accelerator designs that must keep pace with changing model operators, memory access patterns, and serving constraints.

Mateo Alvarez, University of Michigan

Smarter Memory Systems for Irregular Parallel Workloads

4:00 PM · 32-G449 · Hybrid

Architectural support for reducing tail latency and improving bandwidth efficiency in graph analytics and sparse tensor codes.

Aisha Green, Google

Warehouse-Scale Scheduling Meets Architecture

4:00 PM · 32-G449 · Zoom

Cross-layer opportunities between cluster schedulers and node-level architecture for training and serving large models.