Toward an ecosystem-agnostic standard for quantum runtime architecture

Research on achieving Quantum Utility (QU) with software approaches covers all layers of Quantum Computing Optimization Middleware (QCOM) and requires execution on real quantum hardware (QH). Due to the nascent nature of the technology domain and the proprietary strategies of both large and small players, popular runtimes for executing quantum workloads lack flexibility in programming models, scheduling, and hardware access patterns, including queuing, which creates roadblocks for researchers and slows innovation. These problems are further exacerbated by emerging hybrid operating models that place Graphical Processing Unit (GPU) supercomputing and Quantum Intermediate Representation (QIR) at the heart of real-time computations across quantum and distributed resources. There is a need for a widely adopted runtime platform (RP) driven by the open-source community that can be deployed to work in a distributed manner between Quantum Processing Units (QPUs), GPUs, control hardware, and external computational resources and provide required flexibility in terms of programming and configuration models. The product discovery approach of the Value Proposition Canvas (VPC) and software architecture techniques including Architecture Drivers (ADs), Quality Attributes (QAs), Attribute-Driven Design (ADD) were employed to design a blueprint architecture. The intent of the blueprint is to move beyond the concept of just software that runs quantum workloads and focus on practical challenges of existing runtimes that block efforts of QC optimization research. Once embraced by the community, it will open doors to the evaluation of new algorithmic strategies not widely used as of now.