Source-linked AI summary
IBM Quantum Computers: Evolution, Performance, and Future Directions
M. AbuGhanem
TL;DR
IBM Quantum’s hardware evolution and performance metrics require comprehensive historical documentation across processor generations and systems. This paper reviews that progression and reports milestones including Condor’s 1,121-qubit processor, Egret’s QV of 512, and IBM systems’ reported reliability and stability.
Problem
The paper addresses the need for comprehensive performance metrics documenting IBM Quantum’s hardware evolution during the NISQ era.
Method
The study reviews performance metrics and summarizes the progression of IBM Quantum processors across current and retired systems, including their technological attributes and revisions.
Results
IBM Quantum’s progression includes Condor with 1,121 superconducting qubits, Egret with a QV of 512, and systems reporting over 95% collective uptime.
Takeaways & Limitations
The documented evolution provides a comparative framework for IBM Quantum systems and highlights technological strides in hardware and software development.
Abstract
from arXiv · showhide
Quantum computers represent a transformative frontier in computational technology, promising exponential speedups beyond classical computing limits. IBM Quantum has led significant advancements in both hardware and software, providing access to quantum hardware via IBM Cloud since 2016, achieving a milestone with the world's first accessible quantum computer. This article explores IBM's quantum computing journey, focusing on the development of practical quantum computers. We summarize the evolution and advancements of IBM Quantum's processors across generations, including their recent breakthrough surpassing the 1,000-qubit barrier. The paper reviews detailed performance metrics across various hardware, tracing their evolution over time and highlighting IBM Quantum's transition from the noisy intermediate-scale quantum (NISQ) computing era towards fault-tolerant quantum computing capabilities.
I. INTRODUCTION
Quantum computing offers a potential route beyond classical computational limits, while IBM Quantum has advanced accessible superconducting hardware, software, and performance measurement across processor generations.
- Quantum computers use quantum-mechanical principles to potentially solve problems beyond classical computers’ reach.
- Since 2016, IBM and other companies have provided cloud access to quantum computers for academic and commercial users.
- IBM Quantum has advanced scalable superconducting processors by increasing qubit counts, improving coherence times, and implementing error-correction techniques.
- The paper reviews IBM hardware through performance metrics including relaxation times, frequencies, anharmonicity, readout errors, gate errors, connection errors, and gate times.
- 1,121 superconducting qubits mark Condor’s 2023 milestone, while Heron’s 133 qubits achieved a three-times-lower error rate than IBM’s previous processor.
- Heron improves device performance by 3-5x over 127-qubit Eagle processors and virtually eliminates cross-talk.
- IBM Quantum System Two uses three Heron processors and modular architecture to support parallel circuit execution for quantum-centric supercomputing.
A. Canary
IBM’s processor progression spans compact and medium-scale families, increasingly sophisticated packaging and connectivity, and higher-performance designs such as Egret and Condor.
- Canary: The Canary family comprises compact processors with 5-16 qubits on an optimized two-dimensional lattice.
- Condor: 1,121 superconducting qubits characterize Condor, which uses cross-resonance gates and a 50% increase in qubit density.
- Falcon: The Falcon family targets medium-scale circuits with QV 128 and heavy-hexagonal connectivity optimized for cross-resonance two-qubit gates.
- Egret: 33 qubits and QV 512 characterize Egret, whose tunable couplers improve two-qubit gate speed and fidelity.
- Egret: 99.9% fidelity was achieved by many Egret gates while spectator errors were minimized.
E. Eagle
IBM’s Eagle family advanced superconducting processors through a 127-qubit heavy-hexagonal design, improved coherence, and continued deployment across IBM Quantum systems. The section also situates Eagle alongside later Osprey, Heron, and Condor processors and reports Eagle r3 performance metrics.
- Eagle family configuration: 127 qubits and QV 128 define Eagle’s processor configuration, using advanced packaging and a heavy-hexagonal layout.Qubits connect to two or three neighbors in a tessellated-hexagon pattern.
- Eagle family configuration: Heavy-hexagonal connectivity reduces errors from interactions between adjacent qubits while maintaining processor reliability and functionality.The layout limits each qubit’s neighboring connections to two or three.
- Eagle generations: Eagle r1 was introduced in December 2021, while Eagle r3 arrived in December 2022 with enhanced coherence properties and design continuity.Eagle r1 used design elements and parameters similar to Falcon r5.11 and supported fast qubit readout.
- Eagle r3 performance: 7.571 × 10^-3 median ECR error, 2.411 × 10^-4 median SX error, and 1.350 × 10^-2 median readout error were reported for Eagle r3.The same report gives median T1 of 262.69 µs and median T2 of 176.67 µs as of August 1, 2024.
IV. PERFORMANCE AND CHARACTERISTICS OF IBM’S QUANTUM
IBM Quantum’s systems evolved from early retired machines and simulators to current hardware characterized by detailed performance measurements. The section also describes Qiskit’s role in constructing, transforming, executing, and post-processing quantum circuits and in supporting iterative workflows.
- Retired systems and simulators: IBM Quantum’s offerings span retired early systems, advanced processors, and simulators that supported quantum algorithm experimentation.Several early systems, including ibmq 5 yorktown and ibmq 16 melbourne, were retired in 2021.
- Current-system performance: 15 up-to-date IBM Quantum machines are analyzed using coherence times, frequencies, anharmonicity, readout errors, readout length, flip probabilities, and gate-error metrics.The performance summaries cover both single-qubit and two-qubit gates.
- Qiskit: Qiskit is IBM’s open-source software development kit for quantum information science, supporting circuit construction and analysis across quantum and classical operations.Its circuit abstractions include qubit operations, measurements, unitaries, Cliffords, isometries, Fourier transforms, and classical computations.
- Qiskit applications: Qiskit supported error mitigation, fault-tolerant magic-state preparation beyond break-even fidelity, and studies using up to 133 qubits and thousands of two-qubit entangling gates.These examples connect the software stack with both error-management research and larger quantum experiments.
- Qiskit workflow: Qiskit workflows translate classical problems into circuits, transform circuits, execute them on target backends, and post-process results into solutions.The workflow can iterate by generating new circuits from earlier results and combine quantum and classical computing.
D. Scale to large numbers of qubits
Utility-scale quantum computing requires circuits exceeding 100 qubits and 1,000 gates, while IBM’s hardware, software, and execution workflow support large-scale experiments and detailed performance evaluation.
- D. Scale to large numbers of qubits: More than 100 qubits and over 1,000 gates define the utility-scale task regime discussed in the paper.The paper illustrates this regime with a 100-qubit GHZ-state experiment.
- D. Scale to large numbers of qubits: The Qiskit workflow maps a problem, optimizes circuits and operators, executes a quantum primitive, and analyzes the resulting data.Hardware execution can use dynamical decoupling and adjustable resilience levels; higher resilience requires longer processing times.
- D. Scale to large numbers of qubits: IBM’s hardware inventory includes current processors and retired systems, with performance characterized through qubit, gate, coherence, and readout metrics.The reported metrics include relaxation and coherence times, gate errors and times, and readout properties.
- D. Scale to large numbers of qubits: The 156-qubit Heron r2 system ibm fez reports a median CZ error of 2.848 × 10^-3.The cited table identifies its basis gates as CZ, ID, RZ, SX, and X.
- D. Scale to large numbers of qubits: The 133-qubit Heron system ibm torino reports a median CZ error of 4.769 × 10^-3.Heron systems use CZ, ID, RZ, SX, and X as basis gates in the cited hardware summaries.
A. The IBM’s era of quantum utility
IBM’s 2023 utility-era work demonstrated quantum circuits at a scale beyond brute-force classical simulation and motivated a shift toward heterogeneous, parallel quantum-classical computing.
- A. The IBM’s era of quantum utility: In 2023, an IBM–UC Berkeley experiment demonstrated a path toward practical quantum computing by executing circuits beyond brute-force classical simulation.The result is presented as a groundbreaking experiment showing a practical route forward.
- A. The IBM’s era of quantum utility: 100 qubits and 3,000 gates were run without prior knowledge of the outcomes, using IBM Quantum hardware and software.The paper identifies this as the first time IBM Quantum reached this combined scale without knowing the results in advance.
- A. The IBM’s era of quantum utility: IBM Quantum advocates parallelism, concurrent classical computing, and dynamic circuits beyond traditional circuit models.The proposed direction is a heterogeneous architecture combining scalable parallel circuit execution with advanced classical computation.
- A. The IBM’s era of quantum utility: IBM’s quantum-centric supercomputing vision targets advanced utility-scale work and a seamless development environment, potentially before fault tolerance.The vision is tied to updates and an extended roadmap announced at the 2023 IBM Quantum Summit.
B. IBM Quantum’s roadmap
IBM’s roadmap extends to 2033 by increasing executable gate counts and developing hardware for error-corrected, increasingly large-scale quantum computation.
- B. IBM Quantum’s roadmap: Heron targets 5,000 gates by 2024, with later processor generations using quality improvements to reach higher gate counts.The roadmap spans a decade of quantum innovation through 2033.
- B. IBM Quantum’s roadmap: Starling is projected for 2029 to execute 100 million gates across 200 qubits using error correction based on the Gross code.The paper describes this as a pivotal roadmap milestone.
- B. IBM Quantum’s roadmap: Blue Jay is envisioned for 2033 with capability for 1 billion gates across 2,000 qubits.The paper describes this as a nine-order-of-magnitude increase in gate execution capability since IBM’s first cloud-based devices in 2016.
- B. IBM Quantum’s roadmap: Flamingo, Crossbill, and Kookaburra are intended to demonstrate Gross-code technology using l-, m-, and c-couplers, respectively.These processors are part of the innovation roadmap supporting the planned error-correction approach.
C. IBM Quantum safe
IBM Quantum Safe addresses cryptographic preparation for practical quantum computing through security evaluation, post-quantum adoption, and new quantum-safe technologies.
- C. IBM Quantum safe: Advancing quantum technology creates a need for cryptographic methods based on problems difficult for both quantum and classical computers.The paper frames this need as part of preparing for practical quantum computing.
- C. IBM Quantum safe: The roadmap extends toward quantum-centric supercomputers with thousands of logical qubits in 2030 and beyond.Figure 8 presents IBM’s technology roadmap for this development trajectory.
- C. IBM Quantum safe: IBM Quantum Safe helps enterprises evaluate cryptographic security and update cybersecurity strategies for the practical-quantum-computing era.The stated focus is enterprise preparation for changing cryptographic risks.
- C. IBM Quantum safe: IBM’s Quantum Safe roadmap advances research, industry collaboration, post-quantum cryptographic adoption, and quantum-safe technologies.The program includes the IBM Quantum Safe Explorer, launched in October 2023.
VII. CONCLUSION
The study traces IBM Quantum’s rapid hardware and software evolution through performance evaluations across current and retired systems. It presents comparative evidence of technological progress and future prospects.
- IBM combines hardware advancements with software frameworks such as Qiskit in its continuing development of quantum technology.
- IBM Quantum’s journey is evaluated through comprehensive performance data from current and retired quantum computers.The evaluations provide a comparative framework for examining system evolution.
- Comparative metrics across systems document significant technological strides over time.The paper uses these metrics to illustrate IBM Quantum’s evolution and future prospects.
Appendix A: Characteristics of IBM’s Quantum Computers
The appendix compiles historical performance and hardware characteristics for IBM Quantum systems from the NISQ era. Its tables support comparisons across processor generations, including qubit counts, connectivity, errors, coherence, gates, and calibration records.
- Performance summaries: The appendix documents coherence times, qubit frequencies, gate error rates, qubit counts, basis gates, connections, and calibration dates.The records cover retired systems and older performance data from current systems.
- Performance summaries: 433-qubit IBM Seattle uses the Osprey r1 processor with median ECR error 2.155 × 10−2 and median T1 88.35 µs.The same table reports median SX error 6.256 × 10−4, median readout error 4.910 × 10−2, and median T2 58.73 µs.
- Processor specifications: IBM’s listed systems span processor families including Osprey, Eagle, Hummingbird, and Falcon across machines ranging from 5 to 433 qubits.The appendix includes systems such as Seattle, Sherbrooke, Ithaca, Montreal, Mumbai, and several smaller Falcon processors.
- Processor specifications: 127-qubit IBM Washington uses the Eagle r3 processor and has 288 CX qubit connections.Its listed basis gates are CX, ID, IF ELSE, RZ, SX, and X.