DPS: Dual-Mode Precision LLM Serving with Semi-Unified Memory
cs.DC, cs.AI, cs.ET, cs.PF, cs.SE
Submitted: 2026-09-28
Updated: 2026-09-28
Sources
- DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
- DeepSeek-V3 Technical Report
- Efficient Memory Management for Large Language Model Serving with PagedAttention
- SGLang: Efficient Execution of Structured Language Model Programs
- Qwen3 Technical Report
- Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
- An Inquiry into Datacenter TCO for LLM Inference with FP8
- Efficient Post-training Quantization with FP8 Formats
- NestedFP: High-Performance, Memory-Efficient Dual-Precision Floating Point Support for LLMs
- LLaMA: Open and Efficient Foundation Language Models
- QQQ: Quality Quattuor-Bit Quantization for Large Language Models
- LongCodeBench: Evaluating Coding LLMs at 1M Context Windows
- Qwen3-Coder-Next Technical Report
- Any-Precision Deep Neural Networks
- MatGPTQ: Efficient and Accurate Inference over Nested Quantized Models
- GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models
- SLOs-Serve: Optimized Serving of Multi-SLO LLMs
- Revisiting Service Level Objectives and System Level Metrics in Large Language Model Serving
- MorphServe: Efficient and Workload-Aware LLM Serving via Runtime Quantized Layer Swapping and KV Cache Resizing
- Prism: Cost-Efficient Multi-LLM Serving via GPU Memory Ballooning
Related papers
- iScheduler: Reinforcement Learning-Driven Continual Optimization for Large-Scale Resource Investment Problems
- SAMM: Sharded Automated Market Maker
- InferScale: GPU-Native KV Injection for Personalized LLM Serving
- Vigil: Accountable Liveness against Selective Silence
- Steelhead: Interleaving Partially Synchronous and Asynchronous Commit Rules on a Shared DAG
- Democratizing MoE inference on commodity GPUs with CoMoE