TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Qblox says it has connected NVIDIA DGX Spark to its quantum system controller using NVIDIA NVQLink, measuring round-trip latency below 7.2 microseconds on DGX Spark. The company says the setup is designed to bring GPU computing into real-time quantum workloads such as calibration and error correction; independent validation and the extent of practical performance gains were not provided in the source report.
Qblox announced that it has connected NVIDIA’s DGX Spark compact AI computer to its quantum system controller using NVIDIA NVQLink, reporting round-trip latency below 7.2 microseconds on DGX Spark. The company says the integration is intended to put GPU computing in the real-time control loop for quantum workloads, including calibration and quantum error correction.
Qblox also reported round-trip latency below 4.0 microseconds when using an NVIDIA RTX PRO 6000 Blackwell GPU. The figures were supplied in Qblox’s announcement, as reported by The Quantum Insider on October 22; the source material does not describe an independent test or provide detailed test conditions, such as workload, measurement method or system configuration.
The system combines Qblox’s Q1 sequence processors and LINQ with NVIDIA GPU computing. According to Qblox, the arrangement is intended to let GPU-accelerated calibration routines update qubit parameters while experiments are running, rather than only between experiments. For quantum error correction, the company says syndrome data can be sent to GPU-based decoders at low latency to help identify and correct errors in real time.
Qblox says developers can use NVIDIA CUDA-Q to build and test workloads on desktop hardware, then carry the same code over to GPU servers as quantum systems grow. The announcement describes this as a development path, not evidence that all workloads will transfer unchanged or achieve the same performance on different hardware.
Connecting GPUs to Quantum Control
The announcement addresses a practical challenge in scaling quantum systems: quantum processors need classical computing to monitor and control experiments, and some tasks may need to run quickly enough to affect the ongoing operation. Bringing a GPU into that loop could give researchers access to more computing capacity for tasks such as calibration and decoding error-correction data.
Qblox says the integration is aimed at supporting work toward systems with logical qubits, which encode quantum information in ways designed to protect it from errors. The latency measurements are relevant because control-loop timing can constrain what tasks are feasible. However, the announcement does not establish that the setup has delivered better quantum-computer performance, enabled fault-tolerant operation, or improved error rates. Those outcomes would depend on further testing with specific quantum hardware and workloads.
The desktop-to-server path may also make it easier for researchers to develop workloads on a local system before using larger GPU infrastructure. Whether code portability translates into comparable results across systems remains to be demonstrated publicly.
As an affiliate, we earn on qualifying purchases.
Earlier NVQLink and CUDA-Q Work
The integration extends work Qblox says it began with NVIDIA NVQLink, announced in October 2025. Qblox later announced support for the NVIDIA CUDA-Q real-time API in March 2026. The company characterizes the new announcement as a move from architecture work to measured hardware performance.
NVQLink and CUDA-Q are NVIDIA technologies that Qblox says support the connection between quantum control hardware and GPU computing. The new report focuses on a specific hardware combination—Qblox’s controller with DGX Spark—and gives latency figures for that system and an RTX PRO 6000 Blackwell configuration. It does not provide a full comparison with other controllers, GPU platforms or existing quantum-control arrangements.
“Quantum computers will scale only as fast as the classical control systems around them.”
— Niels Bultink, Qblox CEO and co-founder
Test Conditions and Results Still Open
The announcement gives headline latency thresholds but does not detail how the measurements were made, which specific workloads were used, or whether the figures include the full path from quantum measurement through GPU processing and back to control hardware. The source report also does not include independent verification or a comparison against a baseline system.
It remains unclear which quantum processors have been tested with this setup, whether the described calibration and error-correction functions have been demonstrated during live experiments, and what effect they have on error rates or experiment time. The announcement provides no deployment schedule, customer list, pricing, or performance data at larger data-center scale.
Further Workload Testing Needed
The next useful evidence would be detailed test results showing the measurement setup, the quantum hardware involved, and performance across particular calibration or error-correction workloads. Independent evaluation and comparisons with other system configurations would help establish what the reported latency means in practice.
Qblox and NVIDIA have not announced a specific next milestone in the source material. Researchers and customers will need further demonstrations to determine how reliably workloads run on DGX Spark, whether they transfer to GPU servers as described, and whether the approach produces measurable gains in operating quantum systems.
Key Questions
What did Qblox announce?
Qblox said it integrated NVIDIA DGX Spark with its quantum system controller using NVQLink, reporting round-trip latency below 7.2 microseconds on DGX Spark.
What latency did Qblox report on the RTX PRO 6000 Blackwell?
Qblox reported round-trip latency below 4.0 microseconds on an NVIDIA RTX PRO 6000 Blackwell. The announcement did not provide detailed test conditions or independent validation.
What quantum tasks is the integration intended to support?
Qblox says the integration is intended to support GPU-accelerated qubit calibration and quantum error correction, including sending syndrome data to GPU-based decoders. The announcement does not specify the quantum hardware or provide results showing improved error rates.
Can researchers move code from DGX Spark to larger systems?
Qblox says CUDA-Q lets developers work on desktop hardware and carry the same code to GPU servers as systems scale. The source report does not establish how much code modification may be needed or whether performance remains the same across platforms.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
