OpenQSE

I gave a talk at the OpenQSE (Open QHPC Software Ecosystem) initiative’s architecture seminar series focused on the Q-HPC middleware we are developing at Pasqal such as a 2nd level scheduler that will take care of scheduling the quantum tasks from within the HPC allocation. You can find the meeting notes and transcript here. The slides will appear on the GitHub repo. And you can even watch the recording.

In this talk, I presented Pasqal’s approach to quantum-HPC integration, arguing that the primary bottlenecks today are software and operational rather than hardware. After briefly introducing Pasqal’s neutral-atom QPUs and their deployment in HPC and cloud environments, the focus shifted to the middleware stack required to make quantum resources usable inside existing supercomputing workflows. I outlined how hybrid quantum–classical applications fit naturally into HPC usage patterns, and why tight integration with schedulers, job management systems, and monitoring infrastructure is essential for both users and operators.

I briefly described the Quantum Resource Management Interface (QRMI), a thin, vendor-agnostic layer that exposes quantum resources to classical resource managers such as Slurm that we’re developing with partners. Building on this, I introduced the second-level scheduler we are developing at Pasqal. This scheduler mediates access to quantum hardware within an HPC allocation, enforcing policies, priorities, and operational constraints that are difficult or impossible to express in a first-level batch scheduler alone. It will enable new degrees of freedom for hybrid workloads, including job classes, development versus production modes, preemption, and calibration-aware scheduling, while leaving to the HPC scheduler what it does best.

The talk concluded by positioning this middleware and scheduling architecture as a first step toward a more user-centric and operator-friendly Q-HPC environment. By providing consistent runtime environments, integrated observability, and extensible control points for HPC centers, the approach aims to improve developer productivity, reduce environment-induced errors, and allow sites to experiment and innovate on quantum–HPC integration using open, modifiable software