Models of Cache Coherence for Hybrid Memory Systems Assisted by Quantum

Abstract

Hybrid memory systems, which integrate classical volatile memory with emerging non-volatile and quantum-assisted memory technologies, present novel challenges for cache coherence protocols. Ensuring data consistency across heterogeneous memory layers is critical for maintaining system reliability and performance, especially as quantum memory modules introduce probabilistic behaviors and longer access latencies compared to traditional SRAM or DRAM. This paper proposes new formal models of cache coherence specifically designed for hybrid memory architectures augmented by quantum-assisted memory components. The proposed models extend conventional coherence paradigms by incorporating quantum state behavior, error rates, and latency variations inherent to quantum memory. We develop a hybrid coherence protocol that orchestrates synchronization between classical caches and quantum memory cells, accounting for quantum decoherence effects and stochastic access delays. To evaluate the effectiveness of the models, we simulate a multi-core processor environment with integrated quantum memory modules using a custom hybrid memory simulator. The simulator models cache line states, memory access ordering, and coherence messages under realistic workload scenarios including parallel computing and quantum error correction routines. Our experimental results demonstrate that the hybrid coherence models reduce coherence-related stalls by up to 28% compared to baseline classical coherence protocols adapted naïvely to quantum memory. Additionally, the models enable improved cache hit rates and overall memory system throughput, compensating for quantum memory latency overheads. The formal framework also provides a foundation for verifying coherence correctness in hybrid memory systems with probabilistic quantum states, a capability absent in existing classical coherence verification methods. This research advances the state-of-the-art in heterogeneous memory system design by bridging classical and quantum memory coherence domains. The findings highlight the importance of protocol adaptations to accommodate quantum memory characteristics and open avenues for future work in hardware-software co-design, fault-tolerant quantum computing integration, and next-generation memory hierarchy optimizations.