The open-source xDQC compiler helps quantum applications run across multiple QPUs, using CUDA-Q to optimise distributed execution instead of relying on a single quantum processor.
memQ has open-sourced its Extensible Distributed Quantum Compiler xDQCxDQC, a compiler designed to help quantum applications run across multiple Quantum Processing Units QPUsQPUs instead of depending on a single quantum processor. Built on NVIDIA’s open-source CUDA-Q platform, xDQC is intended to simplify the development of distributed quantum applications while allowing researchers to scale beyond the limits of individual QPUs.
Unlike conventional quantum compilers that target a single processor, xDQC partitions a quantum circuit across multiple QPUs and manages communication between them. It automatically handles qubit allocation, schedules operations across different processors and coordinates entanglement resources required for distributed execution, reducing the amount of manual optimisation needed when developing large quantum workloads.
The compiler has been demonstrated on hybrid quantum-classical workloads running across eight interconnected QPUs, where it simulated circuits involving hundreds of qubits. By combining CUDA-Q’s hybrid programming model with distributed compilation, xDQC enables developers to explore larger quantum workloads while continuing to use classical GPUs as part of the overall computing pipeline.
The project is designed as an extensible framework rather than a fixed compiler implementation. Developers can adapt it to different quantum hardware configurations as distributed quantum computing evolves, making it suitable for research into multi-QPU architectures and future quantum-networked systems.
By releasing xDQC as open-source software, memQ is making distributed quantum compilation more accessible to researchers and developers working on next-generation quantum infrastructure. The release builds on the open CUDA-Q ecosystem while providing a practical framework for experimenting with quantum applications that span multiple processors instead of being constrained by the capacity of a single QPU.
