A systolic update scheme to overcome memory bandwidth limitations in GPU-accelerated FDTD simulations
Jesse Lu, David Qu, Jim Qu, Ryan Fong, Geun Ho Ahn, Jelena Vučković · 2025
Read A systolic update scheme to overcome memory bandwidth limitations in GPU-accelerated FDTD simulations by Jesse Lu, David Qu, Jim Qu, Ryan Fong, Geun Ho Ahn, Jelena Vučković (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-20610.txt.
Description
Jesse Lu and five co-authors present a 2025 scheme to speed light wave simulations on graphics chips. Such simulations step electric and magnetic fields through a grid over time. They need much memory traffic but little math, which clashes with chips built for heavy math. Demand for fast light design tools grows with AI hardware.
The scheme passes updates from cell to cell in a chain, like an assembly line. Only edge values between blocks need shared memory, so sync points drop sharply. On one top graphics chip, the full three-dimensional method ran at a high rate of cell updates. The authors see a path to cheaper, faster design checks.
Catalog record
| CREATOR | Jesse Lu, David Qu, Jim Qu, Ryan Fong, Geun Ho Ahn, Jelena Vučković |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-20610