mLR: Scalable Laminography Reconstruction based on Memoization Conference: The International Conference for High Performance Computing, Networking, Storage and Analysis; November 16–21, 2025; St Louis, MO, USA The International Conference for High Performance Computing, Networking, Storage and Analy
Bin Ma, Viktor Nikitin, Xi Wang, Tekin Bicer, Dong Li · 2025
Read mLR: Scalable Laminography Reconstruction based on Memoization Conference: The International Conference for High Performance Computing, Networking, Storage and Analysis; November 16–21, 2025; St Louis, MO, USA The International Conference for High Performance Computing, Networking, Storage and Analy by Bin Ma, Viktor Nikitin, Xi Wang, Tekin Bicer, Dong Li (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-2511-01893.txt.
Description
Bin Ma and four colleagues speed up laminography, a tomographic method for inspecting flat objects layer by layer. The accurate ADMM-FFT reconstruction algorithm is notoriously slow and memory-hungry. Their system, mLR, is built on a simple observation: across the algorithm's iterations, many nearly identical Fourier transforms get computed again and again.
Memoization replaces those repeated transforms with cached results. A series of techniques makes the caching both beneficial and scalable. One is variable offloading. It saves CPU memory and spreads the work across graphics processors, within and across compute nodes. With mLR the team reconstructed a 2K-by-2K-by-2K input. That is the largest problem laminography reconstruction has tackled with this algorithm, under limited memory. Running time fell by about 53 percent on average. It fell by up to about 65 percent, compared with the original method.
Catalog record
| CREATOR | Bin Ma, Viktor Nikitin, Xi Wang, Tekin Bicer, Dong Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-01893