Research Papers
Research Papers news — Tuesday 27 October 2026
54 stories from 2 sources, every one in English. Each links to its publisher.
-
ZipFusion: Fusing Lossless KV Decompression with Attention for Efficient LLM Decoding
arXiv cs.DC (Distributed)
arXiv:2610.33719v1 Announce Type: cross Abstract: The key-value (KV) cache places growing demands on GPU memory capacity and bandwidth as large language models serve longer contexts and more concurrent requests. Its footprint limits…
-
A Density--Delay Law for Event-Driven State Progression in Open Distributed Systems
arXiv cs.DC (Distributed)
arXiv:2603.26758v2 Announce Type: replace Abstract: This paper studies delay-induced competition in an honest event-driven longest-chain network and distinguishes selected-tip divergence from lowest-common-ancestor (LCA) competing-branch…
-
Priority-Aware Scheduler for Serverless Functions Provision and Load Balancing
arXiv cs.DC (Distributed)
arXiv:2610.35073v1 Announce Type: new Abstract: Serverless computing has emerged as a pivotal cloud deployment paradigm due to its efficient utilization of on-demand resources. However, the persistent cold start problem continues to hinder…
-
Retrofitting Service Dependency Discovery in Distributed Systems
arXiv cs.DC (Distributed)
arXiv:2510.15490v2 Announce Type: replace Abstract: Modern distributed systems rely on complex networks of interconnected services, creating direct or indirect dependencies that can propagate faults and cause cascading failures. To…
-
SGL: Towards a Bridging Model for Heterogeneous Hierarchical Platforms
arXiv cs.DC (Distributed)
arXiv:2610.33995v1 Announce Type: new Abstract: This paper introduces the Scatter-Gather parallel-programming and parallel-execution model in the form of a simple imperative Scatter-Gather Language (SGL). Its design is based on past…
-
Faster Parallel Batch-Dynamic Algorithms for Low Out-Degree Orientation
arXiv cs.DC (Distributed)
arXiv:2602.17811v3 Announce Type: replace Abstract: A low out-degree orientation directs each edge of an undirected graph with the goal of minimizing the maximum out-degree of a vertex. In the parallel batch-dynamic setting, one can insert…
-
InferScale: Kubernetes-Native Predictive Autoscaling for SLO-Aware AI Inference
arXiv cs.DC (Distributed)
arXiv:2610.34006v1 Announce Type: new Abstract: Kubernetes Horizontal Pod Autoscaler (HPA) is widely used for inference services, but reactive scaling is often too slow when pods take tens of seconds to become ready. The problem gets worse…
-
ActECC: Action Chunk Temporal Harnessing for Edge-Cloud Collaborative VLA Inference System
arXiv cs.DC (Distributed)
arXiv:2603.07949v3 Announce Type: replace Abstract: VLA models have emerged as the mainstream for embodied intelligence, which decouples reasoning and execution via action chunks. To enable real-time deployment of VLA models, existing…
-
Replicate What Varies: Resource Sharing for Domain-Randomized Embodied Simulation
arXiv cs.DC (Distributed)
arXiv:2610.33582v1 Announce Type: new Abstract: Simulation provides a low-cost sandbox where embodied agents can attempt tasks, observe outcomes, and learn from experience. Running many diverse simulation environments concurrently can…
-
Supervision Without Blocking: An Evidence-Driven Redesign of OTP-Style Fault Tolerance for Cooperative Runtimes
arXiv cs.DC (Distributed)
arXiv:2610.34521v1 Announce Type: new Abstract: Actor frameworks on mainstream runtimes advertise Erlang/OTP-style supervision, but OTP's consistency contracts assume a synchronous, single-threaded supervisor, and ports inherit that…
-
COPE: Composable Performance Emulation for Mixture-of-Experts Offloading Policies
arXiv cs.DC (Distributed)
arXiv:2610.33404v1 Announce Type: cross Abstract: As LLMs scale, mixture-of-experts (MoE) architectures increase model capacity while keeping per-token computation manageable through sparse expert activation. This sparsity also enables…
-
ScaleRLT: Exact Log-Probabilities and Fast Deferred Rollouts for Reinforcement Learning on Looped Transformers
arXiv cs.DC (Distributed)
arXiv:2610.34427v1 Announce Type: cross Abstract: Looped language models such as Ouro reuse one block of layers several times per token. A rollout engine can batch the deeper loops of several response positions and run them only when a…
-
Free to Look, Free to Reason: Simple LLM Harnesses Outdiagnose Hand-Crafted Root Cause Analysis Agents
arXiv cs.DC (Distributed)
arXiv:2610.35767v1 Announce Type: cross Abstract: Root cause analysis (RCA) of microservice failures has attracted many LLM agents whose investigation is hand-crafted, through standard operating procedures, fixed stages and role agents…
-
KubePACS: Kubernetes Cluster Using Performant, Highly Available, and Cost Efficient Spot Instances
arXiv cs.DC (Distributed)
arXiv:2604.24027v3 Announce Type: replace Abstract: Cloud users aim to minimize cost while maximizing performance by selecting the most suitable instance types for their workloads. To reduce expenses, spot instances have been widely…
-
DualPace: Decoupled Scheduling and Cooperative GPU--CPU Decoding for LLM Serving
arXiv cs.DC (Distributed)
arXiv:2610.33501v1 Announce Type: new Abstract: Large language model (LLM) serving relies on batching to improve inference throughput, but the key-value (KV) cache grows with sequence length and limits GPU-resident concurrency. CPU…
-
Speeding up Log-Sum-Exp: Kernel Fusion at the Memory Wall, Integer Arithmetic at the Compute Wall
arXiv cs.DC (Distributed)
arXiv:2610.33647v1 Announce Type: new Abstract: While the Log-Sum-Exp (LSE) function is a foundational numerical primitive for stable log-domain computation in modern machine learning, its fast and efficient computation on generic…
-
Who Verifies the Verifiers? Accountable Semantic Consensus for AI-Mediated Web Information Verification
arXiv cs.DC (Distributed)
arXiv:2610.33587v1 Announce Type: cross Abstract: Large language models (LLMs) increasingly decide whether Web evidence supports online claims, often by voting over several assessors whose errors are presumed independent and benign. On the…
-
rush: Scalable Asynchronous Distributed Computing via Shared State in R
arXiv cs.DC (Distributed)
arXiv:2606.21430v4 Announce Type: replace Abstract: Many algorithms in statistics and machine learning can be parallelized in an asynchronous manner where workers need to communicate through shared state rather than execute independent…
-
LLMs in Cloud: A Discussion on Architecture, Orchestration, and Inference
arXiv cs.DC (Distributed)
arXiv:2610.33475v1 Announce Type: new Abstract: The widespread deployment of foundational Large Language Models (LLM) across cloud datacenters marks a fundamental paradigm shift from stateless deep learning toward distributed…
-
Prefix Keys and Values Matter Late: RIPE Caching for Self-Speculative Decoding in Looped Language Models
arXiv cs.DC (Distributed)
arXiv:2610.34436v1 Announce Type: new Abstract: A looped language model runs one block of layers $R$ times per token, so its key-value (KV) cache holds $R$ planes per recurrent layer, and self-speculative decoding (draft with a few loops…
-
NovaFabric: A Black-Box Recorder for AI Agents with Sealed, Replayable, Diffable and Auditable Runs
arXiv cs.DC (Distributed)
arXiv:2610.34594v1 Announce Type: new Abstract: An AI agent run is no longer a function call. It is a cascade of prompts, model calls, tool invocations, file and network effects and sub-agents. When such a run misbehaves, or an auditor…
-
When and Why to Use TMA: A Workload-Grounded Characterization of Tensor Memory Acceleration on NVIDIA Hopper GPUs
arXiv cs.DC (Distributed)
arXiv:2610.34907v1 Announce Type: new Abstract: The Tensor Memory Accelerator (TMA), introduced on NVIDIA Hopper, is used pervasively in production GPU kernels yet its usage is conspicuously selective: expert kernel developers in CUTLASS…
-
FrontGI: Architecture-Aware Parallel GNN Inference on Chiplet-Based Multicore Systems
arXiv cs.DC (Distributed)
arXiv:2610.34911v1 Announce Type: new Abstract: The realm of high-performance computing has expanded to include machine learning tasks of many forms. Graph Neural Networks (GNNs) are becoming increasingly popular, with their applications…
-
Context Is Not Free: Carbon-Aware LLM Routing for Multi-Turn LLM Web Services across Geo-Distributed Data Centers
arXiv cs.DC (Distributed)
arXiv:2610.35230v1 Announce Type: new Abstract: The sustainability of the Web has drawn growing attention, as large language model (LLM)-based web services leave a substantial carbon footprint. These services typically run in…
-
How Numerical Errors Scale with Hardware
arXiv cs.DC (Distributed)
arXiv:2610.35413v1 Announce Type: new Abstract: In this short technical note, we outline how to bound numerical errors with respect to computational resources. We consider algorithms with known complexity estimates that can be executed in…
-
Governing Enterprise AI Traffic at Scale: Design and Operation of a Policy-Driven, Multi-Vendor LLM Gateway on Envoy
arXiv cs.DC (Distributed)
arXiv:2610.35509v1 Announce Type: new Abstract: Large organizations consume large language models (LLMs) from several vendors at once, each with its own API dialect, credential model, streaming protocol and pricing. Integrated directly…
-
A Unified Queueing-Based Framework for Comparative Evaluation of Serverless Autoscaling Strategies under Bursty Workloads
arXiv cs.DC (Distributed)
arXiv:2610.35564v1 Announce Type: new Abstract: Serverless computing platforms must preserve quality of service (QoS) while adapting to highly bursty and unpredictable workloads. To cope with workload fluctuations while satisfying QoS…
-
Making the Telemetry Pipeline Measurable: An Edge IoT Testbed and Dataset for End-to-End Latency, Provenance, and Energy-Aware Benchmarking
arXiv cs.DC (Distributed)
arXiv:2610.35591v1 Announce Type: new Abstract: Edge-centric IoT systems are increasingly evaluated through sensor values or platform features, yet the telemetry pipeline itself, the sequence of sensing, serialization, message mediation…
-
Nemo: Efficient Caching with Index Pushdown for Disaggregated OLTP Databases
arXiv cs.DC (Distributed)
arXiv:2610.33306v1 Announce Type: cross Abstract: Storage-disaggregated databases decouple compute from storage, improving elasticity but suffering from network bottlenecks. Most cloud databases use page-based B+trees and buffer pools…
-
Operating System Support for Accelerators as First-Class Compute Elements in Heterogeneous Systems
arXiv cs.DC (Distributed)
arXiv:2610.33380v1 Announce Type: cross Abstract: Heterogeneous systems-on-chip (SoCs) increasingly rely on domain-specific hardware accelerators (HAs) to deliver performance and energy efficiency gains beyond what general-purpose…
-
Accelerating SO(2)-Equivariant Materials Models through Low-Precision Emulation
arXiv cs.DC (Distributed)
arXiv:2610.35164v1 Announce Type: cross Abstract: Rotationally equivariant architectures improve the physical fidelity of machine-learned (ML) universal materials models based on graph neural networks (GNNs). However, their scalability is…
-
Hapax Locks : Value-Based Mutual Exclusion
arXiv cs.DC (Distributed)
arXiv:2511.14608v4 Announce Type: replace Abstract: We present Hapax Locks, a novel locking algorithm that is simple, enjoys constant-time arrival and unlock paths, provides FIFO admission order, and which is also space efficient and…
-
From GPU to TPU with Coding Agents: What Works, What Fails, and What It Costs?
arXiv cs.DC (Distributed)
arXiv:2610.34422v1 Announce Type: cross Abstract: A coding agent can migrate a machine learning workload, report success, and still change how the model trains. Rising compute costs, limited accelerator capacity, and vendor dependence are…
-
Modeling Liquid-Cooled Data Centers in EnergyPlus
arXiv cs.DC (Distributed)
arXiv:2610.33148v1 Announce Type: cross Abstract: As AI workloads accelerate data center transition toward high-density racks with direct-to-chip (DTC) liquid cooling, existing simulation tools remain limited in evaluating whole-building…
-
Cross-Domain Sequential Recommendation with Representation-Level User Transfer
arXiv cs.DC (Distributed)
arXiv:2610.35550v1 Announce Type: cross Abstract: Cross-domain sequential recommendation aims to improve next-item prediction in a target domain by leveraging user behavior from auxiliary source domains. However, real-world deployment…
-
HiLO: Towards Product-Level Graceful Degradation with a Quality-Aware, Load-Aware Control Framework
arXiv cs.DC (Distributed)
arXiv:2610.34844v1 Announce Type: new Abstract: Graceful degradation is a key mechanism employed by large-scale systems to handle overload conditions. Its importance has grown further as GPU-intensive LLM inference introduces persistent…
-
One-Shot Federated Model Editing for Device Dynamics at the Edge
arXiv cs.DC (Distributed)
arXiv:2610.35212v1 Announce Type: cross Abstract: Standard federated learning (FL) requires persistent device connectivity across multiple communication rounds, making it unsuitable for dynamic environments such as vehicular networks…
-
Live Adaptation of Dataflow Stream Processing for Hybrid Edge-Cloud Analytics
arXiv cs.DC (Distributed)
arXiv:2610.34658v1 Announce Type: new Abstract: Dataflow streaming systems, like Apache Flink and Spark Streaming, are integral to modern business analytics, providing real-time insights across industries. However, for emerging hybrid…
-
Speculative Decoding for Mixture-of-Experts Models: Expert Bytes, Kernel Bandwidth and the Best Draft Length
arXiv cs.DC (Distributed)
arXiv:2610.34440v1 Announce Type: cross Abstract: Speculative decoding checks several drafted tokens in one verification pass. On a mixture-of-experts (MoE) model, consecutive tokens often select the same experts. A deduplicating engine…
-
Waypoint: Analytically Modeling GEMM for CPUs
arXiv cs.DC (Distributed)
arXiv:2610.34914v1 Announce Type: new Abstract: High-performance General Matrix Multiplication (GEMM) implementations rely on breaking down GEMM into carefully selected tiles to achieve peak performance. We present Waypoint, a new…
-
Peer-Relative Trust-Guided Update Screening for Robust Semi-Decentralized Federated Learning under Label-Flipping Attacks
arXiv cs.DC (Distributed)
arXiv:2610.25714v2 Announce Type: replace-cross Abstract: Trust and reputation mechanisms can reduce the influence of malicious clients in federated learning (FL), but strict update filtering is difficult under non-independent and…
-
MANTA: Unlocking DAG Flexibility in Asynchronous BFT
arXiv cs.DC (Distributed)
arXiv:2610.34419v1 Announce Type: new Abstract: DAG-based BFT protocols marked a shift in consensus design by replacing single-leader proposal paths with concurrent block proposals linked through a shared graph, where references encode…
-
Merge Hulls Without Copying Vertices: GPU Isotonic Regression
arXiv cs.DC (Distributed)
arXiv:2610.24038v2 Announce Type: replace Abstract: Isotonic regression finds the monotone sequence closest to the data in weighted least squares. It is used to calibrate probabilities and inside differentiable ranking. Its standard…
-
Cross-Modal Data Co-Locality for Heterogeneous Storage Systems
arXiv cs.DC (Distributed)
arXiv:2610.34595v1 Announce Type: new Abstract: Distributed vector stores increasingly serve multimodal workloads whose representations are jointly accessed but independently partitioned and placed. As a result, cross-modal physical…
-
LooseLeaf: Enabling Serverless GPUs for Computational Notebooks
arXiv cs.DC (Distributed)
arXiv:2610.33275v1 Announce Type: new Abstract: Computational notebooks reserve a GPU for the whole session, yet the GPU computes only while a cell executes and sits idle otherwise. Serverless execution would release the GPU between cells…
-
Failure-Resilient and Verifiable Distributed Approximate Nearest-Neighbor Retrieval
arXiv cs.DC (Distributed)
arXiv:2610.34608v1 Announce Type: new Abstract: Distributed vector databases increasingly support security-sensitive applications, yet existing systems primarily protect index integrity or replica availability rather than the…
-
Multi-Agentic AI for Fairness-Aware and Accelerated Multi-modal Large Model Inference in Real-world Mobile Edge Networks
arXiv cs.DC (Distributed)
arXiv:2602.07215v2 Announce Type: replace-cross Abstract: Generative AI (GenAI) has transformed applications in natural language processing and content creation, yet centralized inference remains hindered by high latency, limited…
-
Efficient Exact Unlearning for Federated Zeroth-Order Optimization
arXiv cs.DC (Distributed)
arXiv:2610.35211v1 Announce Type: cross Abstract: Federated zeroth-order optimization (FedZOO) reduces the memory and communication costs of collaborative training for edge intelligence by avoiding backpropagation while communicating only…
-
One Megakernel is Not Enough: Predicting Where to Cut for Streaming Inference
arXiv cs.DC (Distributed)
arXiv:2610.35336v1 Announce Type: cross Abstract: We address a fundamental question: How far should kernel fusion extend and which boundaries are worth preserving? Unlike the general convention of using one monolith megakernel at lower…
-
Do Bio-Inspired Algorithms Save Energy in Data Centres? Separating Server Packing from Migration Search in Virtual Machine Consolidation
arXiv cs.DC (Distributed)
arXiv:2610.34102v1 Announce Type: new Abstract: Energy-efficient virtual machine (VM) consolidation matters as data centre electricity demand rises. Bio-inspired metaheuristics, including ant colony optimisation (ACO), particle swarm…
-
A Deterministic Security Runtime for LLM Agents on MPI and Batch-Scheduled HPC Systems: Consensus-Governed, Hardware-Rooted Supervision of Every Rank
arXiv cs.DC (Distributed)
arXiv:2610.33318v1 Announce Type: cross Abstract: Large language model agents launch simulations on high-performance computing (HPC) systems, where one sbatch or mpirun can commit thousands of node-hours and a forged result can cost lives…
-
FaaSpot: Fault-Tolerant Serverless Workflow Execution on Spot Instances
arXiv cs.DC (Distributed)
arXiv:2610.26709v2 Announce Type: replace Abstract: Serverless computing is increasingly used to run complex DAG workflows across diverse applications, and individual function execution times have grown to support them, pushing workflow…
-
WorldPager: Zero-Stall Tiered Memory for Long-Horizon World Models on a Single Consumer GPU
arXiv cs.DC (Distributed)
arXiv:2610.35529v1 Announce Type: new Abstract: Interactive world models keep the KV of past chunks as retrieval memory so that revisited places stay consistent. On a single consumer machine (one RTX 4090, 32 GiB of host memory, one NVMe…
-
AI watermarking efforts must encourage independent public scrutiny
Nature Technology