NVIDIA 使用 K-点批处理和混合精度技术解决平面波 DFT 的 GPU 瓶颈问题。
Tackling GPU Bottlenecks in Plane-Wave DFT with K-Point Batching and Mixed Precision
NVIDIA 采用 K-点批处理和混合精度技术优化平面波 DFT 计算,显著提升训练下一代机器学习原子间势(MLIP)所需的密度泛函理论(DFT)数据的 GPU 性能。该技术通过并行处理和降低精度需求,将计算速度提升 3.5 倍,同时保持高精度,适用于生成数千万能量、力和应力数据。
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…
来源: NVIDIA Technical Blog · developer.nvidia.com