Palesi and Colleagues Propose Dependency-Aware Scheduling for Multi-Core Quantum Systems

Scheduling quantum circuits on multicore processors now occurs by assigning each gate as soon as its dependencies and resources are available, enabling greater parallelism across cores. Rajeswari Suance P S of the Indian Institute of Technology Guwahati, Chandigarh University, and the University of Catania and colleagues have devised a method for organising quantum calculations, key as quantum computers increase in size and complexity. Current systems limit the number of qubits, and this approach utilises multiple processing cores to overcome these restrictions.

By scheduling each step of a calculation as soon as its requirements are met, the team achieved a 40 per cent reduction in processing time compared to traditional methods, improving how efficiently cores use resources. Rajeswari Suance P S and colleagues tackle the challenge of increasing the processing power of quantum computers by distributing calculations across multiple cores, a strategy mirroring the move to multicore processors in classical computing. As quantum computers grow, simply adding more qubits to a single chip becomes increasingly difficult, and this new approach instead connects smaller processing units, each containing a limited number of qubits, to work in parallel. This is akin to adding more lanes to a motorway to handle increased traffic, improving overall system capacity. The researchers of Technology Guwahati, Chandigarh University, and the University of Catania have developed a scheduling method that assigns each step of a quantum calculation as soon as its requirements are met, unlike traditional ‘layered scheduling’ which organises tasks like an assembly line.

Greedy scheduling delivers substantial gains in multicore quantum circuit completion times

A 40 per cent reduction in makespan, the total time to complete a quantum circuit, occurred using a new greedy scheduling strategy developed by researchers from University of Catania, Chandigarh University, and Indian Institute of Technology Guwahati. The 40 per cent threshold represents a major leap towards practical quantum computation with multicore systems, exceeding the limitations of previous ‘layered scheduling’ methods. These older methods processed gates sequentially, hindering parallel processing and creating bottlenecks that underutilised resources, preventing full exploitation of multicore potential.

The investigations conducted by the University of Catania, Chandigarh University, and Indian Institute of Technology Guwahati team revealed a 40 per cent reduction in makespan, alongside improved core utilisation and enhanced resource management within the multicore system. Meticulous tracking of gate-level dependencies within quantum circuits enabled this outcome, moving beyond the constraints of traditional ‘layered scheduling’ which often left computational qubits idle while awaiting intercore communication. The new greedy scheduling strategy also accounts for ‘entanglement link-availability’, a key factor in Network-on-Chip systems where qubits distribute themselves across multiple cores and rely on quantum teleportation for interaction. Evaluation showed that overlapping communication latency with computation particularly benefited circuits with differing intercore communication distances, with those featuring both short and long-range qubit interactions experiencing the most considerable gains.

Demonstrating performance gains with a layered scheduling baseline for multicore quantum

Multicore designs offer a potential route to scaling quantum computers beyond the limitations of building ever-larger single chips. However, this work currently benchmarks only against layered scheduling, leaving open the question of how this greedy approach compares to other, potentially more sophisticated, algorithms. While a 40 per cent reduction in processing time is encouraging, it remains unclear whether these gains will hold as circuits become sharply more complex and the interaction between cores intensifies.

Optimising how calculations divide and execute across multiple processors is a worthwhile pursuit, and a baseline for evaluating more complex scheduling algorithms designed for multi-core quantum computers has now been established. The team’s new scheduling strategy unlocks greater parallelism in multi-core quantum computers by assigning gates for execution immediately when resources permit, a contrast to previous ‘layered scheduling’ which processed gates in sequential blocks. A 40 per cent reduction in the time to complete quantum circuits demonstrates the effectiveness of this approach, highlighting the benefits of exploiting fine-grained parallelism across multiple cores, and paving the way for more complex algorithms and larger problem sizes to address key challenges in scaling quantum computation beyond single-chip processors.

The researchers demonstrated a 40% reduction in the time required to execute quantum circuits using a new greedy scheduling strategy on multi-core quantum architectures. This matters because efficient scheduling is crucial for overcoming limitations in scaling quantum computers beyond single processors. The strategy improves core utilisation by assigning calculations for execution as soon as resources are available, allowing for greater parallelism. The authors established a baseline for evaluating more complex scheduling algorithms designed for these systems, potentially enabling larger and more complex quantum computations.