Reproduce “SweepANN: A GPU-Accelerated Large-K Approximate Vector Search for Drug Discovery” 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
Zhenyu Zhang, Qi Lin, Nan Zhang, Zhichao Cao
SweepANN ranks molecule libraries for protein pockets by scanning every compressed library row on GPUs and using a retained seed to filter candidates before top-K selection. It returns the compressed-score top-K, not guaranteed FP32-reference top-K. On four RTX PRO 6000 GPUs, it processed 500M rows for 26,562 pockets in 61.04 seconds of resident time with 99.63% mean recall; minimum recall was 98.52%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.