Skip to content
HARDWARE · QUANTUM ERROR CORRECTION

Quantum Machines connects its controller to NVIDIA’s error-correction software

A test sends simulated error checks out for interpretation and returns a correction decision during the same run.

Published Source report Sources checked
Reading tools for text size, copying, printing and your reading list
My reading list
A technician stands at an open rack in a row of blue-and-white cabinets labeled ALPS.
The Alps supercomputer in Switzerland uses NVIDIA GH200 hardware. Quantum Machines ran its decoder on the CPU of a separate GH200 system.Illustrative photoMarco Abram / CSCS

has a timing problem: identifying a fault is useful only if the answer arrives when the computation needs it. Quantum Machines has tested a connection that sends error-check data from its controller to software that interprets the pattern and returns a correction decision.[1]

On November 16, the company reported connecting its OPX1000 controller to NVIDIA’s CUDA-Q QEC decoding server. A decoder interprets error-check patterns to infer the needed correction. This test replayed simulated data rather than freshly measured errors. PyMatching ran on the CPU of an NVIDIA GH200 system; GPU dispatch remains future work.[1]

The company reports 2.4 microseconds for two-way communication. Decoding time was measured separately and increased with code distance, a parameter governing the error-correcting code’s size and protection. The communication figure is not a measured complete correction time.[1]

Each tested configuration used 10,000 repeated runs and simulated error-check records generated with an assumed 0.1% physical error probability. The correction decision returned within the same run. Early installation for OPX1000 customers is planned for December 2026.[1]

The software arrangement matters beyond a single stopwatch result. NVIDIA’s server guide separates the program controlling a processor from the service predicting corrections. A compatible controller sends measurements and receives logical correction decisions. Its configuration identifies the decoder and how to interpret the measurements, letting researchers change the decoding pipeline without treating every experiment as an entirely separate system.[2]

The guide also makes a crucial distinction: acknowledging receipt of measurements does not mean decoding has finished. Its examples test functionality, not a guaranteed quantum-processor feedback time. That is why the useful next result is a complete, clearly bounded test using measured qubit errors, with response-time variation and correction quality reported together. Faster communication alone cannot establish better protection of quantum information.[2]

Keep your reading trail

Evidence & context

2 linked sources

Company-reported controller/server benchmark with simulated errors; no demonstrated live-qubit correction improvement.

How to read the evidence and dates

An announcement records what an organization stated. An experiment reports an observation. A simulation or resource estimate depends on its assumptions. QubitWire’s analysis connects these sources; the labels are not quality scores.

Published is the article’s first QubitWire publication. Updated marks a substantive content change. Sources checked records a review of the underlying evidence. An older paper can support a new explanation, and another press release about the same work is not independent replication.

Sources

  1. Quantum Machines: OPX1000 CUDA-Q QEC decoding-server results

    November 16, 2026 company announcement. Pre-sampled simulated syndromes; CPU PyMatching; communication and decoding metrics separate.

  2. NVIDIA: CUDA-Q QEC Decoding Server User Guide

    Living technical documentation, checked November 17. Describes configuration, correction flow and timing limits; not independent reproduction of the QM benchmark.

Keep up with these topics

Explore the subjects covered in this story.

READ NEXT

Continue exploring

All Hardware stories
  1. More on error correction

    Alice & Bob’s GPU Decoding Test Uses Synthetic Errors

    The NVQLink proof of concept connects control electronics to GPU decoding. It does not yet perform real-time error correction.

    Published · Source report · Sources checked · About 2 min read
  2. Hardware / Quantum control

    The fast feedback loop inside a hybrid quantum experiment

    Quantum Machines describes a September CUDA-Q demonstration linking live qubits, control electronics and GPU computing. Its timing claim measures a specific exchange.

    Published · Sources checked · About 2 min read