Queuing for Inference: Dynamic Memory Footprints and Costly Preemption
cs.PF, cs.NI
Submitted: 2026-10-24
Updated: 2026-10-24
Related papers
- HybridQC: Hardware-Grounded Simulation of Tightly Integrated Hybrid Quantum-Classical Systems
- Reproducible LLM Inference Benchmarking: A Sequential Isolation Protocol for Regression Testing
- A Controlled Study of Memory Hierarchy Transitions in Quantum Circuit Simulation on Apple M4 Pro Unified Memory Architecture
- Cluster, Route, Escalate: Cascaded Framework for Cost-Aware LLM Serving
- The Caf'e in Amsterdam: When the Incumbent Becomes the Oracle