QUARK: A Framework for Quantum Computing Application Benchmarking
summary
QUARK is an open‑source, modular framework that enforces an application-centric metrics top class, lets users plug in mapping, solver and device classes, and outputs reproducible artifacts (CSV + config YAML) via a CLI. It runs a real BMW press‑shop allocation benchmark (10 press shops × 10 plants, 120 pages of constraints) with classical (linear/CUBO) mappings and solvers including C plexus / rubies, and a generative‑model benchmark using QCBM and GAN training on Qiskit. Benchmarks show simulator GPU runs (8 A100s vs 265 CPUs) giving up to a 300x GPU speedup.
AI-written summary · about
chapters (10)
- 0:00QUARK framework intro
- 1:24Why QUARK for benchmarking
- 3:13Quantum benchmarking challenges
- 4:56Reproducibility with QUARK
- 6:01QUARK modular architecture
- 7:46Optimization benchmark walkthrough
- 10:17ML benchmark setup
- 13:47Generative modeling results
- 18:02GPU vs CPU performance
- 20:29QUARK community summary
related talks
12mApplication-oriented Quantum Computing Benchmarking with QUARK
19mSupporting End-Users in Realizing Quantum Computing Applications
47mUnboxing Quantum Computing
15mQOS: Quantum Operating System
21mSoftware for Quantum Computing- Easing Solutions for Complex
20mJKQ: JKU Tools for Quantum Computing
45mQuantum application development for multi-processor systems using CUDA-Q
47mQuantum Computing