Command Palette
Search for a command to run...
QEF_Benchmark Electronic Structure Calculation Benchmark Dataset
QEF_benchmark is a set of high-performance computing (HPC) performance verification and test inputs officially maintained by the Quantum ESPRESSO Foundation (QEF). It is used to evaluate the computational performance, scalability and energy efficiency of the first-principles electronic architecture software Quantum ESPRESSO under different hardware platforms (CPU / GPU / vector engine) and parallel strategies (MPI / OpenMP / Hybrid).
This dataset is based on the core modules of Quantum ESPRESSO (such as pw.x and ph.x) and covers a variety of typical computational scenarios, from small molecules to large-scale metal surface systems. It provides standardized test tasks for HPC environments and is the de facto standard for verifying computing node performance, compiler optimizations, and parallel efficiency.
Dataset structure
The dataset is located in the benchmarks/ directory of the official GitHub repository, and the overall directory structure is as follows:
bash
benchmarks/
├── AUSURF112/ # Au (111) 表面 112 原子基准(中型 DFT 测试)
├── ALCHEMI/ # 电子态投影 ALCHEMI 分析基准
├── C60/ # 富勒烯体系(C60 分子)基准
├── GRAPHENE/ # 石墨烯平面结构基准
├── H2O-256/ # 256 个水分子体系(MD/SCF 性能测试)
├── Si256/ # 硅超胞体系(并行扩展性基准)
├── Si512/ # 更大规模硅体系,用于强扩展测试
├── NITROGEN/ # 氮分子能带与电子密度基准
├── WATER_MOLECULE/ # 小体系 DFT 验证(入门与功能测试)
├── README.md # 官方基准使用说明
Each subdirectory contains:
- Input file (.in): The main control file for the Quantum ESPRESSO module;
- Pseudopotential file (.UPF): The pseudopotential required for the system;
- Reference output (.out): Standard result comparison;
- Script file (.sh): Example of running scripts in parallel (such as in a Slurm or MPI environment).
Dataset Contents
- AUSURF112: DFT (SCF), gold surface Au(111) containing 112 atoms, used to test the FFT performance and parallel communication efficiency of medium-sized systems.
- CNT10POR8: DFT (Relax), (10,8) carbon nanotube system, used for SCF convergence and memory usage assessment in large systems.
- GRIR443 / GRIR686: DFT (SCF), a graphene/iridium (111) interface system (443 and 686 atoms) used to evaluate multi-node parallelism and FFT communication performance.
- PHONON: DFPT (ph.x), a polycrystalline silicon, quartz, LaO, CrI₃ and other systems, used to test the performance of phonon computing and linear response modules.
- PSIWAT: Hybrid DFT (EXX), a water molecule system used to evaluate communication and memory performance in hybrid functional computing.
- Other inputs: Additional systems (such as CuO, H₂O) are used for functional verification and test input references.
- Small-benchmarks: DFT / Relax, for small multi-element systems (such as Si, C, Fe, H₂O, etc.), including automated running scripts and pseudopotential files, used for rapid performance regression, single-node efficiency evaluation and compilation optimization testing.
Build AI with AI
From idea to launch — accelerate your AI development with free AI co-coding, out-of-the-box environment and best price of GPUs.