跳到正文
NVIDIA Technical Blog:Agentic AI / Generative AI· Erica Tsai·· 3 天前AI 評分19

以 K 點批次處理及混合精度應對平面波 DFT 的 GPU 瓶頸

Tackling GPU Bottlenecks in Plane-Wave DFT with K-Point Batching and Mixed Precision

AI 導讀

NVIDIA 探討以 K 點批次處理和混合精度,應對平面波 DFT 的 GPU 瓶頸。訓練新一代機器學習原子間勢(MLIP)需要數千萬至數億次 DFT 計算,以取得不同結構和化學組成的能量、力和應力;平面波 DFT 是生成這些資料的黃金標準,VASP 是其主流實作。

正文

Training a next-generation machine learning interatomic potential (MLIP) requires ground truth: tens to hundreds of millions of density functional theory (DFT)...

Training a next-generation machine learning interatomic potential (MLIP) requires ground truth: tens to hundreds of millions of density functional theory (DFT) calculations of energies, forces, and stresses across diverse structures and chemistries. Plane-wave DFT, with VASP as its dominant implementation, is the gold standard for generating that data. It is accurate…

Source

來源:NVIDIA Technical Blog:Agentic AI / Generative AI · developer.nvidia.com