The paper investigates how different parallelism strategies—tensor parallelism (TP), pipeline parallelism (PP), and hybrids—affect the trade‑off between time‑to‑first‑token (TTFT) and inter‑token latency (ITL) in large language model inference. It introduces a framework that maps feasible TTFT‑ITL regions and Pareto frontiers for various workloads, revealing a shift from TP‑dominant to PP‑dominant configurations as prefill work increases. The authors also present an SLO‑aware selection method based on goodput, showing that stricter TTFT goals favor PP, while relaxed constraints move toward TP, and they demonstrate that longer contexts further favor PP for scalability and SLO attainment.
arXiv:2604. 07472v2 Announce Type: replace Abstract: Serving large language model (LLM) inference in cloud environments requires jointly optimizing model selection, GPU provisioning, parallelism configuration, and workload routing under latency, accuracy, memory, and budget constraints.
arXiv:2602. 24044v2 Announce Type: replace-cross Abstract: Large Language Model (LLM) adapters enable low-cost model specialization, but introduce complex caching and scheduling challenges in distributed serving systems where hundreds of adapters must be hosted concurrently.
arXiv:2412. 04504v2 Announce Type: replace-cross Abstract: As large language models (LLMs) grow in popularity for their diverse capabilities, improving the efficiency of their inference systems has become increasingly critical.
arXiv:2602. 21788v2 Announce Type: replace-cross Abstract: Scaling long-context capabilities is crucial for Large Language Models (LLMs).
arXiv:2606. 10493v1 Announce Type: cross Abstract: Local deployment of large Mixture-of-Experts (MoE) models falls short of the service quality achieved in cloud-scale environments, even under low-concurrency workloads.
arXiv:2504. 11320v4 Announce Type: replace-cross Abstract: Large language models now serve millions of users daily, with providers incurring costs exceeding $700,000 per day.
arXiv:2606. 17104v1 Announce Type: cross Abstract: As large language models (LLMs) are increasingly deployed in latency- and cost-sensitive settings, inference efficiency has become a central systems challenge.
arXiv:2609.08307v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used as backends for intelligent web services, but serving them across the edge continuum requires bala...
arXiv:2610.17689v1 Announce Type: new Abstract: Serverless inference is emerging as an efficient deployment model for Large Language Models (LLMs) due to its ability to elastically scale resources, i...
arXiv:2511. 10480v3 Announce Type: replace-cross Abstract: Optimizing the performance of large language models (LLMs) on large-scale AI training and inference systems requires a scalable and expressive mechanism to model distributed workload execution.
Scepsy is a serving system designed to efficiently schedule arbitrary multi‑LLM agentic workflows on GPU clusters. It leverages the observation that each LLM’s share of execution time remains relatively stable across requests, profiling LLMs under various parallelism levels to build an Aggregate LLM Pipeline that predicts throughput and latency. Using this predictor, Scepsy searches for optimal GPU allocations—balancing fractional GPU shares, tensor parallelism, and replica counts—and then heuristically places them on the cluster to reduce fragmentation and honor network topology, achieving up to 2.5× higher throughput and 1.0–3.3× lower latency compared to baseline approaches.
arXiv:2609.13592v1 Announce Type: cross Abstract: GPU memory bandwidth and capacity limit throughput in large language model (LLM) inference. The GPU memory system consists of a primary tier of high-...