Related papers: Locality-aware Fair Scheduling in LLM Serving

High-demand LLM inference services (e.g., ChatGPT and BARD) support a wide range of requests from short chat conversations to long document reading. To ensure that all client requests are processed fairly, most major LLM inference services…

In LLM serving, reusing the KV cache of prompts across requests is critical for reducing TTFT and serving costs. Cache-affinity scheduling, which co-locates requests with the same prompt prefix to maximize KV cache reuse, often conflicts…

The efficient deployment of large language models (LLMs) in online settings requires optimizing inference performance under stringent latency constraints, particularly the time-to-first-token (TTFT) and time-per-output-token (TPOT). This…

Large Language Model (LLM) inference, where a trained model generates text one word at a time in response to user prompts, is a computationally intensive process requiring efficient scheduling to optimize latency and resource utilization. A…

Large language model (LLM) inference systems face a fundamental tension between minimizing Time-to-First-Token (TTFT) latency for new requests and maintaining a high, steady token generation rate (low Time-Per-Output-Token, or TPOT) for…

Serving numerous users and requests concurrently requires good fairness in Large Language Models (LLMs) serving system. This ensures that, at the same cost, the system can meet the Service Level Objectives (SLOs) of more users , such as…

Diffusion large language models (dLLMs) represent a significant advancement in text generation, offering parallel token decoding capabilities. However, existing open-source implementations suffer from quality-speed trade-offs that impede…

Fraud detection and anti-money-laundering (AML) compliance are high-value domains for large language models (LLMs), but their serving requirements differ sharply from generic chat workloads. Compliance prompts are often prefix-heavy,…

Large language models (LLMs) with different architectures and sizes have been developed. Serving each LLM with dedicated GPUs leads to resource waste and service inefficiency due to the varying demand of LLM requests. A common practice is…

Conventional operating system scheduling algorithms are largely content-ignorant, making decisions based on factors such as latency or fairness without considering the actual intents or semantics of processes. Consequently, these algorithms…

Diffusion Language Models (DLMs) promise highly parallel text generation, yet their practical inference speed is often bottlenecked by suboptimal decoding schedulers. Standard approaches rely on 'scattered acceptance'-committing high…

Hosting diverse large language model workloads in a unified resource pool through co-location is cost-effective. For example, long-running chat services generally follow diurnal traffic patterns, which inspire co-location of batch jobs to…

We study offline scheduling for large language model (LLM) serving under a fixed KV-cache memory budget, where requests have heterogeneous prompt (prefill) and response (decode) lengths. Prompt tokens determine initial KV usage, and each…

Fairness has become a crucial aspect in the development of trustworthy machine learning algorithms. Current fairness metrics to measure the violation of demographic parity have the following drawbacks: (i) the average difference of model…

High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks…

The bulk of the research on Long Term Evolution/Long Term Evolution-Advanced packet scheduling is concentrated in the downlink and the uplink is comparatively less explored. In up-link, channel aware scheduling with throughput maximization…

In Large Language Model (LLM) inference, the output length of an LLM request is typically regarded as not known a priori. Consequently, most LLM serving systems employ a simple First-come-first-serve (FCFS) scheduling strategy, leading to…

Large language models (LLMs) are increasingly employed for complex tasks that process multiple generation calls in a tree structure with shared prefixes of tokens, including few-shot prompting, multi-step reasoning, speculative decoding,…

Modern deployment of large language models (LLMs) frequently involves both inference serving and continuous retraining to stay aligned with evolving data and user feedback. Common practices separate these workloads onto distinct servers in…

LLM serving is increasingly dominated by decode attention, which is a memory-bound operation due to massive KV cache loading from global memory. Meanwhile, real-world workloads exhibit substantial, hierarchical shared prefixes across…