Published February 10, 2026 | Version v1
Presentation Open

When to use GPU for Control? Benchmarking Fusion ML Inference Latency

Contributors

Related person:

Supervisor:

  • 1. ROR icon Princeton University

Description

Currently, most ML-based controllers for nuclear fusion are CPU based via Keras2C. This study investigates the effect of GPU vs CPU inference for ML-based tasks on fusion-related ML.

Files

ML Inference for Fusion Benchmark.pdf

Files (1.1 MB)

Name Size Download all
md5:11ae1ee95f1b4df7e6c5146368d4269e
1.1 MB Preview Download

Additional details

Dates

Available
2025-12-09