AGP Picks
View all

Quantum Machines adds real-time NVIDIA CUDA-Q decoding to OPX1000

Nov. 16, 2026
Quantum Machines adds real-time NVIDIA CUDA-Q decoding to OPX1000
By AI, Created 17:26 UTC, Nov 16, 2026, AGP -

Quantum Machines says the OPX1000 can now run NVIDIA CUDA-Q QEC decoders in microseconds through NVIDIA NVQLink, enabling real-time mid-circuit error correction. The early implementation is set for customer installation in December 2026 and is meant to speed testing of fault-tolerant quantum systems.

Why it matters: - Quantum error correction only works if decoding keeps up with the quantum hardware. - The new OPX1000 integration gives quantum controllers a microsecond-latency path to classical decoders. - That can shorten the loop between syndrome measurement and correction, which is central to fault-tolerant quantum computing.

What happened: - Quantum Machines announced first results running NVIDIA CUDA-Q QEC decoders on the OPX1000 quantum control system. - The integration uses the CUDA-Q Realtime API and NVIDIA NVQLink. - The setup allows microsecond-latency callbacks between quantum controllers and classical processors. - OPX1000 customers will be able to run CUDA-Q QEC decoders, or their own decoders, through the CUDA-Q decoding server.

The details: - CUDA-Q Realtime provides a runtime API for microsecond-latency callbacks between quantum controllers and classical compute. - A controller sends a request, a dispatcher routes it to a registered handler, and the response returns to the controller. - Handlers can run on host CPU or GPU threads. - The CUDA-Q QEC decoding server registers decoders as handlers. - The decoding server is transport-agnostic and contains no networking code. - The same QEC software stack, configuration format and decoder setup carry over when the transport layer changes. - Quantum Machines implemented the CUDA-Q Realtime bridge for the OPX1000, giving syndrome measurements a microsecond-latency path to the decoder and a path back for corrections. - A quantum program written in CUDA-Q is compiled once and runs as a single job on the OPX1000. - Syndromes travel to the decoder and the correction returns over the same NVQLink path. - The early implementation uses the CPU dispatch path, with syndrome data passing through the host CPU. - GPU dispatch is next. - The system supports static memory circuits written in CUDA-Q and compiled through Quake to QUA.

Between the lines: - The integration is designed to make decoder choice a software decision rather than a hardware rewrite. - Users can switch between NVIDIA decoders such as PyMatching or the NVIDIA fusion decoder by changing one line of configuration. - Users can also bring their own decoder or use the same path for other classical-to-quantum control tasks. - NVIDIA said the integration gives developers a common path to test different decoders through the same control system and move faster from algorithm research to real-time testing.

What's next: - Quantum Machines said the implementation will be available for customer installation in December 2026. - The current CPU dispatch path will be followed by GPU dispatch. - Further optimization appears most important at code distance 15, where decoding latency rises above 1 microsecond.

The bottom line: - Quantum Machines is pushing quantum error correction from offline analysis into real-time control, with NVIDIA CUDA-Q decoders running directly in the loop on the OPX1000. - In its test runs, the system showed 2.4 microseconds of two-way communication latency across all configurations and sub-microsecond decoding at smaller code distances.

Disclaimer: This article was produced by AGP Wire with the assistance of artificial intelligence based on original source content and has been refined to improve clarity, structure, and readability. This content is provided on an “as is” basis. While care has been taken in its preparation, it may contain inaccuracies or omissions, and readers should consult the original source and independently verify key information where appropriate. This content is for informational purposes only and does not constitute legal, financial, investment, or other professional advice.

Sign up for:

Transportation Industry News

The daily local news briefing you can trust. Every day. Subscribe now.

By signing up, you agree to our Terms & Conditions.

Share this page:

Advanced Search Options

Search for:

Search scope:

Type:

Search in:

Date range:

The last

Sort by:

Sign up for:

Transportation Industry News

The daily local news briefing you can trust. Every day. Subscribe now.

By signing up, you agree to our Terms & Conditions.