分布式、并行与集群计算

Metastable failure is a recent abstraction of a pattern of failures that occurs frequently in real-world distributed storage systems. In this paper, we propose a formal analysis and modeling of metastable failures in replicated storage…

The consensus problem, briefly stated, consists of having processes in an asynchronous distributed system agree on a value. It is widely known that the consensus problem does not have a deterministic solution that ensures both termination…

Large language models (LLMs) require vast amounts of GPU compute to train, but limited availability and high costs of GPUs make homogeneous clusters impractical for many organizations. Instead, assembling heterogeneous clusters by pooling…

The exploration and application of Large Language Models (LLMs) is thriving. To reduce deployment costs, continuous batching has become an essential feature in current service frameworks. The effectiveness of continuous batching relies on…

Learning and constructing large-scale graphs has attracted attention in recent decades, resulting in a rich literature that introduced various systems, tools, and algorithms. Grale is one of such tools that is designed for offline…

The growth of Artificial Intelligence (AI) and large language models has enabled the use of Generative AI (GenAI) in cloud data centers for diverse AI-Generated Content (AIGC) tasks. Models like Stable Diffusion introduce unavoidable delays…

With the exponential growth of smart devices connected to wireless networks, data production is increasing rapidly, requiring machine learning (ML) techniques to unlock its value. However, the centralized ML paradigm raises concerns over…

Hardware accelerators like GPUs are now ubiquitous in data centers, but are not fully supported by common cloud abstractions such as Functions as a Service (FaaS). Many popular and emerging FaaS applications such as machine learning and…

Modern AI workloads such as large language models (LLMs) and retrieval-augmented generation (RAG) impose severe demands on memory, communication bandwidth, and resource flexibility. Traditional GPU-centric architectures struggle to scale…

FastSet is a distributed protocol for decentralized finance and settlement, which is inspired from both actors and blockchains. Account holders cooperate by making claims, which can include payments, holding and transferring assets,…

This work presents HotSwap, a novel provider-side cold-start optimization for serverless computing. This optimization reduces cold-start time when booting and loading dependencies at runtime inside a function container. Previous research…

Flowshop problems play a prominent role in operations research, and have considerable practical significance. The single-machine flowshop problem is of particular theoretical interest. Until now the problem of minimizing late jobs or job…

Byzantine fault-tolerant (BFT) web services provide critical integrity guarantees for distributed applications but face significant latency challenges that hinder interactive user experiences. We propose a novel two-layer architecture that…

Intel(r) Software Guard Extensions (SGX) was originally released on client platforms and later extended to single socket server platforms. As developers have become familiar with the capabilities of the technology, the applicability of this…

Large language models have been widely adopted across different tasks, but their auto-regressive generation nature often leads to inefficient resource utilization during inference. While batching is commonly used to increase throughput,…

Long context fine-tuning of large language models(LLMs) involves training on datasets that are predominantly composed of short sequences and a small proportion of longer sequences. However, existing approaches overlook this long-tail…

We present "Reciprocating Locks", a novel mutual exclusion locking algorithm, targeting cache-coherent shared memory (CC), that enjoys a number of desirable properties. The doorway arrival phase and the release operation both run in…

Autoscaling GPU inference workloads in Kubernetes remains challenging due to the reactive and threshold-based nature of default mechanisms such as the Horizontal Pod Autoscaler (HPA), which struggle under dynamic and bursty traffic patterns…

Large language model (LLM) based agentic workflows have become a popular paradigm for coordinating multiple specialized agents to solve complex tasks. To improve serving efficiency, existing LLM systems employ prefix caching to reuse…

Computational models have become one of the prevalent methods to model complex phenomena. To accurately model complex interactions, such as detailed biomolecular interactions, scientists often rely on multiscale models comprised of several…