Key Findings
Researchers have demonstrated quantum computational advantage via random-circuit sampling (RCS) using the IBM Nighthawk r2 architecture. Utilizing the 120-qubit ibm_phoenix processor, the team operated over 61 active qubits. The experiment produced 106 samples in 19 seconds of active quantum processing unit (QPU) time. Under baseline rejection-sampling assumptions, reproducing this ensemble on the Oak Ridge National Laboratory Frontier supercomputer would require an estimated 1.2 × 1027 machine floating-point operations (FLOPs), totaling roughly 110 years of exascale compute.
Technical Details
The experiment employed a four-color brickwork schedule across 102 active couplers, averaging 25.5 native controlled-Z (CZ) gates per cycle. Fidelity was verified across depths of 4 to 40 cycles using mirror benchmarking with randomized compiling and multi-partition cross-entropy benchmarking (XEB). At a depth of 36 cycles, which incorporates 918 CZ gates, the experimental fidelity reached FXEB ≈ 2.3 × 10−3. This depth was selected because tensor-network contraction costs saturate at full system size with a width of w = 64.
Background & Context
The study compared the Nighthawk r2 architecture against its predecessor, the Nighthawk r1 (ibm_miami). The per-cycle fidelity improved from 0.836 on the r1 to 0.872 on the r2, resulting in a fidelity gain of more than an order of magnitude at deep cycle depths. To ensure the results were not due to classical simulations of decoupled sub-circuits, the team performed “severing attacks.” At 36 cycles, all severing-attack modes scored below 2.2 × 10−4, confirming the benchmark operates on the classically hard side of the noise-induced phase transition.
Significance & Outlook
The circuits were submitted via standard Qiskit Runtime cloud queues without bespoke hardware calibration, demonstrating the accessibility of the hardware. While the current runtime estimate provides a baseline, the authors note that future classical developments—such as GPU-accelerated multi-amplitude reuse, matrix-product state approximations, or batched tensor slicing—could potentially lower classical runtimes.
Get our weekly technology intelligence — free
Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.
Subscribe Free — Weekly Tech Intelligence
By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.
- Your email and selected fields are used only to deliver the newsletter.
- We never share your information with third parties.
- You can unsubscribe anytime via the link in each email.
See our Privacy Policy for details.
Takes about a minute · Unsubscribe anytime
