Research Papers
Research Papers news — Monday 12 October 2026
28 stories from 2 sources, every one in English. Each links to its publisher.
-
Optimizing Frequent Checkpointing via Low-Cost Differential for Distributed Training Systems
arXiv cs.DC (Distributed)
arXiv:2509.04084v5 Announce Type: replace Abstract: Distributed training of large deep-learning models often leads to failures, so checkpointing is commonly employed for recovery. State-of-the-art studies focus on frequent checkpointing…
-
SnapShare: Isolation-Preserving Deduplication for Serverless Snapshots
arXiv cs.DC (Distributed)
arXiv:2610.13318v1 Announce Type: new Abstract: The efficiency of serverless computing fundamentally depends on rapid function startup, a challenge increasingly addressed by restoring from virtual machine snapshots. However, because…
-
Equal-Budget Monte Carlo Tree Search versus Randomized-Restart List Scheduling for DAG Workflows: A Controlled Equal-Budget Computational Study
arXiv cs.DC (Distributed)
arXiv:2610.12542v1 Announce Type: cross Abstract: Monte Carlo Tree Search (MCTS) is often proposed for scheduling directed-acyclic-graph (DAG) workflows, yet most studies never run the comparison that could refute it - giving a much…
-
Sustainable Cloud-Fog-Edge Systems for Smart Buildings: Algorithmic Adaptation under Memory Constraints and Dynamic Orchestration Strategies
arXiv cs.DC (Distributed)
arXiv:2610.13307v1 Announce Type: cross Abstract: The escalating environmental footprint of the Information and Communication Technol- ogy (ICT) sectorwith global Data Center electric- ity usage potentially reaching 3% by 2030 and the…
-
P-PAS: Prefill-Pressure Adaptive Scheduling for Long-Context LLM Serving
arXiv cs.DC (Distributed)
arXiv:2608.15171v2 Announce Type: replace Abstract: Long-context LLM applications such as retrieval-augmented generation (RAG) and agentic systems often process tens of thousands of input tokens to produce short outputs, making end-to-end…
-
Coexponential Exploration in the Clef Programming Language
arXiv cs.DC (Distributed)
arXiv:2610.12911v1 Announce Type: cross Abstract: Clef was created as a concurrent programming language whose compiler tracks state ownership, waiting computations and request dependencies. We develop a coexponential account of stateful…
-
Deploying RDMA for Large-Scale Inference and Training Workloads
arXiv cs.DC (Distributed)
arXiv:2610.12815v1 Announce Type: cross Abstract: RDMA can reduce communication overhead in distributed machine learning systems, but realizing these benefits in production requires more than replacing TCP. Variable-size messages…
-
Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices
arXiv cs.DC (Distributed)
arXiv:2610.12970v1 Announce Type: new Abstract: The increasing demand for GPU-accelerated machine learning inference has exposed limitations in centralized GPU infrastructures, including high cost, limited elasticity, and long wait times…
-
CompileMask: Bridging the Gap Between Expressiveness and Efficiency in Attention Masking
arXiv cs.DC (Distributed)
arXiv:2610.13474v1 Announce Type: new Abstract: Modern attention workloads require diverse masking patterns whose boundaries depend on runtime metadata, yet efficient attention kernels natively support only a small set of patterns…
-
SlackSwap: Bandwidth-Aware Activation Swapping and Recomputation for Memory-Efficient DNN Training
arXiv cs.DC (Distributed)
arXiv:2610.13003v1 Announce Type: new Abstract: Swapping saved activations to host memory once enabled GPUs to train models that exceeded their device-memory capacity when compute throughput and host-link bandwidth were roughly balanced…
-
Democratizing MoE inference on commodity GPUs with CoMoE
arXiv cs.DC (Distributed)
arXiv:2610.09424v2 Announce Type: replace Abstract: Deploying Mixture-of-Experts (MoE) models relies heavily on Expert Parallelism, which generates intense inter-GPU communication. Consequently, state-of-the-art inference systems require…
-
Execution-Feasible Dispatch of LLM Pre-training Loads for Renewable-Dominant AI Data Centers
arXiv cs.DC (Distributed)
arXiv:2610.13047v1 Announce Type: new Abstract: AI data centers (AIDCs) are emerging as large power-system loads, making their operational flexibility increasingly valuable. Existing studies often treat AIDC demand as continuously…
-
Ember: Making Unified Virtual Memory Practical for Discrete-Memory GPUs
arXiv cs.DC (Distributed)
arXiv:2610.13408v1 Announce Type: cross Abstract: GPU working sets have outgrown GPU memory, so applications must oversubscribe it, keeping only hot data on the GPU. NVIDIA's Unified Virtual Memory (UVM) supports oversubscription, but its…
-
Benchmarking Content-Defined Chunking Algorithms for Data Deduplication
arXiv cs.DC (Distributed)
arXiv:2610.13704v1 Announce Type: cross Abstract: Data deduplication relies on content-defined chunking (CDC) algorithms to decide where data is split. We observe that most CDC algorithms are modular: a base chunking algorithm fixes the…
-
Efficient Fault Localization in a Cloud Stack Using End-to-End Application Service Topology
arXiv cs.DC (Distributed)
arXiv:2509.05511v2 Announce Type: replace-cross Abstract: Cloud application services are distributed and have components across the stack working together to deliver the end-user experience. The widespread adoption of microservice…
-
Capturing Electrical Windows Through Reconfigurable Optical Networks: Transfer-aware Scheduling of Multi-Region AI Data Centers
arXiv cs.DC (Distributed)
arXiv:2610.13061v1 Announce Type: new Abstract: Artificial-intelligence data centers (AIDCs) can provide spatiotemporal load flexibility by migrating training workloads in response to regional electricity prices and renewable-energy (RE)…
-
Multi-Source and Multi-Dimensional Information Fusion in Resource-Constrained Edge Networks
arXiv cs.DC (Distributed)
arXiv:2610.13298v1 Announce Type: new Abstract: Multi-source semantic information fusion enables edge networks to support global inference using observations collected by distributed terminals. However, limited communication and…
-
Zepp: Accelerating Distributed MoE Serving under Relaxed Balance Constraints
arXiv cs.DC (Distributed)
arXiv:2610.11158v2 Announce Type: replace Abstract: As Mixture-of-Experts (MoE) models continue to scale, serving them increasingly relies on expert parallelism (EP) across a growing number of devices. Yet skewed expert workloads create…
-
Finding Nemo-Nemo: CFT DAG-based Consensus in the WAN
arXiv cs.DC (Distributed)
arXiv:2604.08914v2 Announce Type: replace Abstract: This paper introduces Nemo-Nemo, a practical crash-fault tolerant (CFT) consensus protocol designed to outperform existing protocols in wide-area networks by bridging design principles…
-
Adaptive DHiSparse: Speculative Decoding over a Load-Adaptive KV Hierarchy
arXiv cs.DC (Distributed)
arXiv:2610.13708v1 Announce Type: cross Abstract: Long-context serving runs out of KV memory before the hardware reaches its useful throughput. HiSparse stores history in host memory and fetches only the selected entries, but it uses one…
-
GPU-resident Execution for Timing-predictable Sensor Processing Systems
arXiv cs.DC (Distributed)
arXiv:2610.13809v1 Announce Type: new Abstract: Real-time embedded systems increasingly place GPU-accelerated processing of high-bandwidth sensor streams inside closed-loop systems, but the CPU (the host) remains the primary orchestrator…
-
Stacked Intelligent Metasurface-Based MultiModal Feature Fusion for Collaborative Inference in Edge Networks
arXiv cs.DC (Distributed)
arXiv:2610.13364v1 Announce Type: new Abstract: In navigating the challenges of computational and communication overhead within the 6G-era Collaborative Inference (CI), this paper explores the feasibility of employing Stacked Intelligent…
-
Taking Tokenization off the Host for Agentic LLM Serving
arXiv cs.DC (Distributed)
arXiv:2610.13325v1 Announce Type: new Abstract: Agentic workloads resend long contexts that barely change between calls. Prefix caching now skips most of their prefill, yet every request must still be tokenized in full before the cache can…
-
Scheduling in Cloud,Fog, and Edge Computing: A Comparative Analysis and Decision Framework for Technique Selection
arXiv cs.DC (Distributed)
arXiv:2610.13830v1 Announce Type: new Abstract: The increasing heterogeneity and dynamism of cloud, fog, and edge computing make scheduling a challenging problem involving diverse resource capabilities, workload characteristics, latency…
-
ShoreKeeper: Accelerating Iterative Data Analytics in the Cloud via Distributed Recovery-Cost-Aware Caching
arXiv cs.DC (Distributed)
arXiv:2610.13396v1 Announce Type: new Abstract: Large-scale data-intensive workloads powered by iterative graph or machine learning processing have an insatiable demand for memory due to the continuous generation of new intermediate data…
-
Designing dielectrics for microchip capacitors
Nature Technology
-
Hidden signals in sleep can predict future health
Nature Technology
-
Private, offline and yours: how to search the literature with local AI systems
Nature Technology