Reproduce “Regular mixed-radix DFT matrix factorization for in-place FFT accelerators” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sergey Salishev
The paper formulates mixed-radix FFTs as stages aligned with an in-place, memory-based accelerator. It factors the transform into smaller butterfly operations, twiddle-factor multiplications, and index rearrangements, proving decimation-in-time (DIT) and decimation-in-frequency (DIF) forms. The factorizations support selecting transform lengths and either mode on a fixed datapath through address-generator reconfiguration, within the stated radix and 1r1w-SRAM scope.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.