Related papers: Dual-Pool Token-Budget Routing for Cost-Efficient …

Serving LLMs with a cluster of GPUs is common nowadays, where the serving system must meet strict latency SLOs required by applications. However, the stateful nature of LLM serving requires maintaining huge states (i.e., KVCache) in limited…

Vision-language models (VLMs) excel at image understanding tasks, but the large number of visual tokens imposes significant computational costs, hindering deployment on mobile devices. Many pruning methods rely solely on token importance…

A well-known dilemma in large vision-language models (e.g., GPT-4, LLaVA) is that while increasing the number of vision tokens generally enhances visual understanding, it also significantly raises memory and computational costs, especially…

Long-context LLM serving is bottlenecked by the cost of attending over ever-growing KV caches. Dynamic sparse attention promises relief by accessing only a small, query-dependent subset of the KV state per decoding step and extending the KV…

Production LLM systems often rely on separate models for safety and other classification-heavy steps, increasing latency, VRAM footprint, and operational complexity. We instead reuse computation already paid for by the serving LLM: we train…

Large Language Models (LLMs) demonstrate substantial potential across a diverse array of domains via request serving. However, as trends continue to push for expanding context sizes, the autoregressive nature of LLMs results in highly…

The ever-increasing computation and energy demand for LLM and AI agents call for holistic and efficient optimization of LLM serving systems. In practice, heterogeneous GPU clusters can be deployed in a geographically distributed manner,…

When IP-packet processing is unconditionally carried out on behalf of an operating system kernel thread, processing systems can experience overload in high incoming traffic scenarios. This is especially worrying for embedded real-time…

Large language models (LLMs) are typically served from clusters of GPUs/NPUs that consist of large number of devices. Unfortunately, communication between these devices incurs significant overhead, increasing the inference latency and cost…

Long-context Multimodal Large Language Models (MLLMs) demand substantial computational resources for inference as the growth of their multimodal Key-Value (KV) cache, in response to increasing input lengths, challenges memory and time…

While prior researches focus on CPU-based microservices, they are not applicable for GPU-based microservices due to the different contention patterns. It is challenging to optimize the resource utilization while guaranteeing the QoS for GPU…

Long-context inference in LLMs faces the dual challenges of quadratic attention complexity and prohibitive KV cache memory. While token-level sparse attention offers superior accuracy, its indexing overhead is costly; block-level methods…

Large Language Models (LLMs) have been widely adopted to process long-context tasks. However, the large memory overhead of the key-value (KV) cache poses significant challenges in long-context scenarios. Existing training-free KV cache…

Scaling the input context length of a large language model (LLM) incurs a significant increase in computation cost and memory footprint to maintain the attention key-value (KV) cache. Existing KV cache compression methods suffer from…

Attention computation takes both the time complexity of $O(n^2)$ and the space complexity of $O(n^2)$ simultaneously, which makes deploying Large Language Models (LLMs) in streaming applications that involve long contexts requiring…

Generative Recommender (GR) inference places embedding hot caches (EMB) and KV caches in direct competition for limited GPU HBM: allocating more memory to one improves its efficiency but degrades the other. Existing systems optimize them in…

Large Language Models (LLMs) have excelled in various tasks but perform better in high-resource scenarios, which presents challenges in low-resource scenarios. Data scarcity and the inherent difficulty of adapting LLMs to specific tasks…

A key challenge in scaling shared-L1 multi-core clusters towards many-core (more than 16 cores) configurations is to ensure low-latency and efficient access to the L1 memory. In this work we demonstrate that it is possible to scale up the…

Large language model (LLM) serving is now limited by the key-value (KV) cache. During decode, each new token rereads prior KV state, so attention becomes a bandwidth- and capacity-heavy memory task. HBM-PIM helps by moving attention closer…

Scaling long-context ability is essential for Large Language Models (LLMs). To amortize the memory consumption across multiple devices in long-context training, inter-data partitioning (a.k.a. Data Parallelism) and intra-data partitioning…