Attend ACME seminar with guest Jingtong Hu, Nov. 16

Join the ASU Center for Semiconductor Microelectronics, or ACME, for a seminar with guest Jingtong Hu, a professor in the Department of Electrical and Computer Engineering at the University of Pittsburgh. His research focuses on embedded systems, on-device artificial intelligence and digital health, with an emphasis on human-centric ambient intelligence.
Hu’s work has received three best paper awards, including the Donald O. Pederson Best Paper Award from IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, as well as five best paper nominations from DAC, ASP-DAC and ESWEEK. He is also a recipient of the IEEE TCCPS Mid-Career Award, DAC Under-40 Innovators Award, JSPS Invitational Fellowship, Carl Friedrich von Siemens Research Fellowship, Humboldt Research Fellowship, Outstanding New Faculty Awards and ACM SIGDA Meritorious Service Award.
On-device large language model, or LLM, inference is driven by critical needs for privacy, low-latency safety and off-grid availability. Yet sustained inference can push mobile GPU temperatures past safe thresholds within minutes, forcing devices into aggressive throttling that undermines the benefits of local execution. Current thermal management treats this as a hardware-only problem, relying on reactive frequency scaling that is blind to the workload it serves.
In this talk, Hu will present three systems that rethink thermal management as a cross-layer, workload-aware design objective spanning model internals and human perception.
First, he will introduce GreenDynaMix, a runtime that converts MoE expert precision into a dynamic thermal control knob. A dual-PID control loop couples DVFS-driven frequency reduction with sensitivity-aware precision demotion, compensating for throttling-induced throughput loss while maintaining output quality.
Second, Hu will present TRAIL, which extends this approach to vision-language models, or VLMs, by dynamically replacing thermally redundant transformer layers with precomputed linear projections. The approach recovers more than 50% of throttling-induced throughput loss while maintaining 95.9% accuracy on VLM benchmarks.
Third, Hu will describe GreenRead, a perceptual throughput pacing framework that uses gaze tracking to match token generation speed to human reading rates, eliminating wasted heat from tokens generated faster than users can consume them. Combining GreenRead with GreenDynaMix on a Samsung Galaxy S25 yields thermal benefits exceeding either technique alone, pointing toward a vertically integrated approach in which a single algorithmic decision propagates thermal relief across the entire inference stack.
ACME seminar: How to Make LLMs Cool: Breaking the Thermal Wall through Thermally Adaptive Algorithms and Human-in-the-Loop Pacing for On-Device Inference
Monday, Nov. 16, 2026
10:30–11:30 a.m.
Goldwater Center (GWC) 487, Tempe campus [map]
