AWS and NVIDIA are expanding their hybrid quantum computing capabilities by integrating NVIDIA’s open-source CUDA-Q platform into Amazon Braket. This collaboration allows researchers to develop and test hybrid quantum-classical workflows using NVIDIA’s GPU-accelerated simulators within Braket’s managed environment.
A notable enhancement in the latest CUDA-Q release is its pulse-level programming interface. As explained by AWS’s Stefan Natu, AWS collaborated with NVIDIA to map this interface onto Braket’s intermediate representation, initially targeting QuEra’s hardware. QuEra is the first quantum device where CUDA-Q workloads operate entirely through Braket, rather than via standalone integration. This approach exemplifies Braket’s role as a unified platform for quantum innovation.
The integration addresses the increasing demand for classical compute resources, essential for tasks like quantum error correction, circuit simulation, and algorithm design. By leveraging CUDA-Q, users can achieve significantly faster simulation runtimes compared to other open-source tools. For example, GPU-based simulations on Braket demonstrate up to 350x speed improvements over CPU-based simulations, enabling more efficient testing of quantum circuits.
Braket also supports seamless deployment of these workflows on diverse quantum hardware, including systems from IonQ, Rigetti, and IQM. This capability allows researchers to transition between simulators and real quantum processors with minimal code changes, facilitating scalable and practical quantum algorithm development.
Looking ahead, AWS and NVIDIA are collaborating to address future challenges in hybrid quantum-classical computing, such as ultra-low latency co-processing and AI-enabled quantum simulations. This partnership aligns with the vision of quantum-accelerated supercomputing, combining classical and quantum technologies for advanced computational applications.
For more details and example implementations, refer to the original blog post.
December 5, 2024
