PPoPP 2026
Sat 31 January - Wed 4 February 2026 Sydney, Australia
co-located with HPCA/CGO/PPoPP/CC 2026
Tue 3 Feb 2026 12:30 - 12:50 at Balmoral - Mixed Precision and Quantization Chair(s): Dingwen Tao

Mixed-precision methods offer the potential to achieve better performance while maintaining accuracy comparable to that of high-precision formats. However, the adoption of mixed precision—particularly with 16-bit formats—in scientific computing remains limited due to precision truncation.

To address this challenge, we propose HierCut, a mixedprecision strategy that leverages cutoff schemes in molecular dynamics (KOKKOS package of LAMMPS) by assigning different precision levels to particles across distinct cutoff layers. We further introduce techniques to improve the performance and accuracy of simulations using 16-bit numerical formats, and develop error analysis methods to guide the configuration of mixed-precision ratios. Our scheme achieves accuracy comparable to high-precision results, delivering a speedup of up to 3.75x over the original FP64 implementation of LAMMPS and up to 1.40x over the optimized FP32 kernels, based on NVIDIA A100 GPUs. Our implementation can be effectively scaled to boost large-scale multi-GPU simulations.

Tue 3 Feb

Displayed time zone: Hobart change

11:30 - 12:50
Mixed Precision and QuantizationMain Conference at Balmoral
Chair(s): Dingwen Tao Institute of Computing Technology, Chinese Academy of Sciences
11:30
20m
Talk
RoMeo: Mitigating Dual-dimensional Outliers with Rotated Mixed Precision Quantization
Main Conference
Qihao Zhang Tsinghua University, MingLiang Tang Tsinghua University, Mingshu Zhai Tsinghua University, Kinman Lei Tsinghua University, Jidong Zhai Tsinghua University
DOI
11:50
20m
Talk
High-Throughput Non-Uniformly Quantized 3-bit LLM Inference
Main Conference
YuAng Chen Chinese University of Hong Kong, Wenqi Zeng Hong Kong University of Science and Technology, Jeffrey Xu Yu Chinese University of Hong Kong
DOI
12:10
20m
Talk
JanusQuant: Accurate and Efficient 2-bit KV Cache Quantization for Long-Context Inference
Main Conference
Chengyu Sun Wuhan University, Yaqi Xia Wuhan University, Hulin Wang , Donglin Yang Nvidia Corporation, Xiaobo Zhou University of Macau, Dazhao Cheng WuHan University
DOI
12:30
20m
Talk
HierCut: Enabling 16-bit Format Mixed Precision for Molecular Dynamics through Hierarchical CutoffBest Artifact Award
Main Conference
zeyu song Tsinghua University, Lin Gan Tsinghua University, Xiaohui Duan Shandong University, Jiayu Fu Tsinghua University, Zhengrui Li Tsinghua University, Yinuo Wang Tsinghua University, Guangzhao Li Chinese Academy of Sciences, Guangwen Yang Tsinghua University
DOI