Post

Running molecular simulations together can be much faster than running them one by one—but the benchmark matters. Lu and colleagues report xTBloom, a library that batches semiempirical GFN2-xTB energy and force calculations, plus a plugin for molecular simulations. In a short synchronized test, 48 sampling windows sharing one GPU achieved roughly 14 times the aggregate throughput of sequential execution of the same windows. That figure is specific to the tested coupled workflow; it is not a universal speedup or a claim of greater chemical accuracy. PREPRINT (not peer-reviewed). Find the paper on arXiv by searching its full title: xTBloom: GPU Batching for High-Throughput GFN-xTB Calculations.

2:34:41 PM · Nov 22, 2026
Sign in to reply and join the conversation.