NVIDIA proposes K-point batching and mixed-precision optimization for plane-wave DFT
Understand this story
In a technical blog post, NVIDIA proposes combining K-point batching with mixed precision to alleviate the computational bottleneck of plane-wave DFT on GPUs. The method targets efficiency when using plane-wave DFT (whose mainstream implementation is VASP) to generate training data for machine learning interatomic potentials (MLIPs). NVIDIA says training next-generation MLIPs requires tens of millions to hundreds of millions of DFT calculations to obtain ground-truth energies, forces, and stresses across different structures and chemical compositions, and that plane-wave DFT is the gold standard for generating such data. The blog post does not provide the speedup relative to existing implementations or accuracy validation results for the method.
AI overview based on source reports · Updated 2 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- NVIDIA Technical BlogTackling GPU Bottlenecks in Plane-Wave DFT with K-Point Batching and Mixed Precision
Training a next-generation machine learning interatomic potential (MLIP) requires ground truth in the form of 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 described as the gold standard for generating that data.
Activity for this story
- Current comparable activity
- 3
- Peak comparable activity
- 10Nov 19 00:00
- Change in the past 24 hours
- -50%
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.