Related papers: PipeLive: Efficient Live In-place Pipeline Paralle…

Modern serving systems for Mixture-of-Experts (MoE) models adopt hybrid data-expert parallelism: expert parallelism (EP) shards experts across GPUs to scale capacity, while data parallelism (DP) replicates attention layers across instances…

KV cache in autoregressive LLMs eliminates redundant recomputation but has emerged as the dominant memory and bandwidth bottleneck during inference, notably with long contexts and test-time scaling. KV quantization is a key lever for…

Deploying large language models (LLMs) for online inference is often constrained by limited GPU memory, particularly due to the growing KV cache during auto-regressive decoding. Hybrid GPU-CPU execution has emerged as a promising solution…

In this work, we investigate a typical scenario in code generation where a developer edits existing code in real time and requests a code assistant, e.g., a large language model, to re-predict the next token or next line on the fly.…

Multi-agent LLM systems on edge devices need to hand off latent context efficiently, but the practical choices today are expensive re-prefill or full-precision KV transfer. We study QKVShare, a framework for quantized KV-cache handoff…

Pipeline parallelism is a key technique for scaling large-model training, but modern workloads exhibit runtime variability in computation and communication. Existing pipeline systems typically consume static, profiled, or adaptively…

This paper introduces an efficient Vision-Language Model (VLM) pipeline specifically optimized for deployment on embedded devices, such as those used in robotics and autonomous driving. The pipeline significantly reduces the computational…

Prefill-Decode (PD) disaggregation has become the standard architecture for modern LLM inference engines, which alleviates the interference of two distinctive workloads. With the growing demand for multi-turn interactions in chatbots and…

The key-value (KV) cache is a foundational optimization in Transformer-based large language models (LLMs), eliminating redundant recomputation of past token representations during autoregressive generation. However, its memory footprint…

Multi-agent large language model (LLM) systems are increasingly adopted for complex language processing tasks that require communication and coordination among agents. However, these systems often suffer substantial overhead from repeated…

The expanding context windows in large language models (LLMs) have greatly enhanced their capabilities in various applications, but they also introduce significant challenges in maintaining low latency, particularly in Time to First Token…

Serving long-context LLMs is costly because attention computation grows linearly with context length. Dynamic sparse attention algorithms (DSAs) mitigate this by attending only to the key-value (KV) cache of critical tokens. However, with…

DNN training is time-consuming and requires efficient multi-accelerator parallelization, where a single training iteration is split over available accelerators. Current approaches often parallelize training using intra-batch…

In real-time Visual SLAM systems, local mapping must operate under strict latency constraints, as delays degrade map quality and increase the risk of tracking failure. GPU parallelization offers a promising way to reduce latency. However,…

This paper presents an FPGA runtime framework that demonstrates the feasibility of using dynamic partial reconfiguration (DPR) for time-sharing an FPGA by multiple realtime computer vision pipelines. The presented time-sharing runtime…

Large Language Models (LLMs) increasingly require processing long text sequences, but GPU memory limitations force difficult trade-offs between memory capacity and bandwidth. While HBM-based acceleration offers high bandwidth, its capacity…

Large Language Model (LLM) inference services demand exceptionally high availability and low latency, yet multi-GPU Tensor Parallelism (TP) makes them vulnerable to single-GPU failures. We present AnchorTP, a state-preserving elastic TP…

Efficient large-scale inference of transformer-based large language models (LLMs) remains a fundamental systems challenge, frequently requiring multi-GPU parallelism to meet stringent latency and throughput targets. Conventional tensor…

Generative large language models (LLMs) have garnered significant attention due to their exceptional capabilities in various AI tasks. Traditionally deployed in cloud datacenters, LLMs are now increasingly moving towards more accessible…

Retrieval-Augmented Generation (RAG) systems suffer from severe time-to-first-token (TTFT) bottlenecks due to long input sequences. Existing KV cache reuse methods face a fundamental trade-off: prefix caching requires identical prefixes…