Quantum computing is rapidly moving toward cloud-native, High-Performance Computing (HPC) models. However, current job submission systems rely on sequential, exclusive-use execution, causing severe resource under-utilization and excessive user wait times. This paper introduces QUDA (Quantum Unified Device Architecture), a distributed orchestration platform designed to solve this bottleneck by decoupling the control and execution planes. QUDA employs a master-worker hierarchy in the form of the Joint Unified Gateway (JUG) for global control and distributed Quantum Processor Schedulers (QuPS) for hardware-aware execution. By abstracting hardware complexity and leveraging OpenQASM 3.0, QUDA enables backend-agnostic portability across heterogeneous systems. We evaluate QUDA using W-state probes and combinatorial optimization benchmarks, demonstrating significant gains in system-wide throughput and qubit utilization over direct submission models. QUDA provides a scalable foundation for modern, distributed quantum workloads.
Alejandro Olvera, Harry Fu, Song Fu· 2026 International Conferenc...· 0 citations
KV-cache offload is widely used to stretch GPU memory for LLM serving, but its storage behavior has not been characterized at the block-device level. In this paper, we study LMCache through realworld multi-session workloads that span same/different context $\times$ same/different prompt, using over 100 stateless requests per workload. Our comparison is observational rather than factorial: it contrasts two realizable deployment snapshots—LMCache 0.3.0 with buffered I/O and the kernel page cache, and LMCache 0.3.16 with prefix-aware deduplication and O_DIRECT. Key findings define the paper. First, the legacy stack appears almost read-free at the SSD layer in all four workloads, but a flush_ram experiment shows that this is conditional on page-cache warmth: once the cache is evicted, the same path issues about 1 GB of sequential reads in 683 ms. Second, deduplication delivers a dramatic write reduction only for the exact-repeat workload: one set collapses from 1.62 GB to 11.75 MB of writes, while the other three sets remain in the 2–5 GB range. Third, O_DIRECT converts a hidden and bimodal read cost into an explicit and stable one: on one set, warm TTFT rises from 27 ms to 93 ms because each of the 99 warm requests re-reads the same ~12 MB partial chunk from SSD. The main conclusion is therefore not “newer is better,” but a system trade-off: predictability versus average-case latency.
Ying He, Dingsen Shi, Yanbo Dai et al.· 2026 International Conferenc...· 0 citations