Research Papers
Research Papers news — Friday 02 October 2026
29 stories from 1 sources, every one in English. Each links to its publisher.
-
Nalar: Workflow-Aware Management of Agentic Applications
arXiv cs.DC (Distributed)
arXiv:2601.05109v2 Announce Type: replace Abstract: LLM-driven agentic applications automate complex, multi-step tasks, but serving them efficiently remains difficult due to heterogeneous components, dynamic model-driven control flow…
-
LBFAST: A Lightweight Moment-Represented Lattice Boltzmann Solver for Multi-GPU Architectures
arXiv cs.DC (Distributed)
arXiv:2609.09160v2 Announce Type: replace Abstract: We present LBFAST, a GPU-oriented lattice Boltzmann solver based on a lightweight moment-represented formulation, in which post-collision populations are reconstructed on the fly from a…
-
MANTA: Machine Learning Augmented Tiering Advisor
arXiv cs.DC (Distributed)
arXiv:2610.00714v1 Announce Type: new Abstract: Memory tiering has been used to expand memory capacity, particularly in datacenters, by combining fast DRAM with slower tiers, including CXL-attached memory. Its effectiveness depends on…
-
GPU-Initiated Communication: Dissecting Down to the Bone
arXiv cs.DC (Distributed)
arXiv:2610.01380v1 Announce Type: new Abstract: GPU-initiated communication lets GPU threads post RDMA operations directly to the NIC. It underpins NVSHMEM, NCCL GIN, and DeepEP, which serve the fine-grained, latency-critical communication…
-
Fault-Tolerant Budget Conservation in Distributed Multi-Agent Delegation
arXiv cs.DC (Distributed)
arXiv:2610.00349v1 Announce Type: cross Abstract: Resource limits are becoming an authorization boundary for AI agents that delegate work across concurrent and failure-prone workers. Parent-child allocation constraints, affine objects, and…
-
Echo: Merging Host Device Buffers to Avoid Redundant Data Movement on Unified Memory SoCs
arXiv cs.DC (Distributed)
arXiv:2609.05635v2 Announce Type: replace Abstract: GPU applications on unified-memory (UMA) edge platforms often inherit a discrete-GPU memory abstraction in which they allocate one buffer for the CPU, another for the GPU, and copy data…
-
Fast Stencil Computations on a Single Arbitrarily Moving Interval
arXiv cs.DC (Distributed)
arXiv:2609.14879v2 Announce Type: replace-cross Abstract: A stencil computation repeatedly updates every cell of a grid from its neighbours' values at the previous timestep. Simulating T steps on N cells directly costs Theta(NT), and a…
-
Blockchain Lifecycle Prediction - Dead Coins
arXiv cs.DC (Distributed)
arXiv:2610.01379v1 Announce Type: new Abstract: The cryptocurrency ecosystem has experienced extraordinary growth alongside an equally remarkable rate of failure, with over 52 percent of all tokens launched since 2021 ceasing to trade by…
-
Progressive-Resolution Secure Aggregation for Federated Learning
arXiv cs.DC (Distributed)
arXiv:2610.00695v1 Announce Type: cross Abstract: Secure aggregation lets a server recover an aggregate of client updates without observing any individual update, but conventional protocols fix the aggregate precision when clients upload…
-
RapidMoE: Exploiting Cross-Asymmetry via Adaptive Residual Offloading for Large-Scale MoE Inference
arXiv cs.DC (Distributed)
arXiv:2610.01265v1 Announce Type: new Abstract: The widespread adoption of Mixture-of-Experts (MoE) has created a growing need for deployment on heterogeneous platforms. However, it exposes a fundamental mismatch between the algorithmic…
-
PaxosLease Revisited: A Checked Model of Diskless Distributed Leases
arXiv cs.DC (Distributed)
arXiv:2609.14640v2 Announce Type: replace Abstract: PaxosLease is a protocol by which a quorum of acceptors grants time-bounded exclusive ownership with no durable acceptor lease state and no disk write on the lease acquisition path. This…
-
Impossibility of One-Way One-Round Quantum 4-Coloring via Matrix-Space Stability
arXiv cs.DC (Distributed)
arXiv:2609.09091v2 Announce Type: replace-cross Abstract: We show that one-way one-round quantum-LOCAL algorithms cannot $4$-color directed cycles with high probability. This is the first lower bound in the high-probability quantum LOCAL…
-
Serving a Revisable World: Versioned Execution for Interruptible Agents
arXiv cs.DC (Distributed)
arXiv:2610.01160v1 Announce Type: new Abstract: LLM agents revise running tasks when users change instructions, tools fail, or new information changes a plan. Today's servers express a revision as aborting old requests and submitting…
-
MoE-CORE: Coordinated Expert Offloading and Residency for Memory-Constrained MoE Inference
arXiv cs.DC (Distributed)
arXiv:2610.01950v1 Announce Type: new Abstract: Sparse expert activation reduces MoE models' computation, yet expert weights can exceed limited device memory. Offloading makes inference feasible on a compact AI appliance but exposes…
-
MegaFlux: Skew-Resilient MoE Megakernels via Pipelined Expert Replication
arXiv cs.DC (Distributed)
arXiv:2610.00671v1 Announce Type: new Abstract: Mixture-of-experts (MoE) megakernels fuse expert-parallel communication with expert computation. However, under fixed expert placement, routing skew creates GPU stragglers: overloaded GPUs…
-
Redundancy Meets Synergy: Dependency-aware Expert Selection for MoE via Submodular Optimization
arXiv cs.DC (Distributed)
arXiv:2610.00558v1 Announce Type: cross Abstract: While Mixture-of-Experts (MoE) models effectively scale model capacity through sparse activation, their deployment is often bottlenecked by prohibitive memory requirements. Extracting a…
-
Towards a Cloud Fog Edge System for Smart Building
arXiv cs.DC (Distributed)
arXiv:2610.01647v1 Announce Type: new Abstract: In this article, we present our vision and recent advancements toward creating a decentralized system capable of learning from real-time data within buildings to support sustainable and…
-
Leto: Fast In-Place Recovery for LLM Training on Surviving Hardware
arXiv cs.DC (Distributed)
arXiv:2610.00687v1 Announce Type: new Abstract: Hardware-operable failures (HOFs) interrupt large language model (LLM) training but permit recovery on the same hardware without reset, repair, or replacement. Existing recovery systems…
-
HakiCC: LLM-Driven Multi-Agent Design and Optimization of Concurrency Control Protocols
arXiv cs.DC (Distributed)
arXiv:2610.00889v1 Announce Type: cross Abstract: Large language models (LLMs) have recently been applied in systems research as a tool to reduce human-intensive engineering effort through cost-efficient automation. Decades of research…
-
Ditto: Generalized Reconfigurable Linearizable Reads
arXiv cs.DC (Distributed)
arXiv:2610.00702v1 Announce Type: new Abstract: Linearizability gives developers the illusion that operations on a distributed datastore execute sequentially on a single machine. Providing it is expensive, so numerous specialized reads…
-
ePACT: Energy-Performance-Aware Commitment Tracking for LLM Serving
arXiv cs.DC (Distributed)
arXiv:2610.01784v1 Announce Type: new Abstract: Reducing LLM serving energy does not by itself guarantee lower deployment cost when electricity procurement exposes operators to unfavorable deviations from preset commitments. We study…
-
When Fancy Eviction Fails: Rethinking Cache Replacement For LLM Prefix Reuse
arXiv cs.DC (Distributed)
arXiv:2609.28870v2 Announce Type: replace Abstract: Long-running LLM applications repeatedly send growing context, making prefix caching critical for reducing prefill cost. Yet prefix-cache behavior under agentic workloads remains poorly…
-
Pushing CPU Speech Synthesis to the Wall: Extreme Inference Tuning under Serverless Architecture and Billing
arXiv cs.DC (Distributed)
arXiv:2610.00063v1 Announce Type: new Abstract: Instance-billed serverless platforms charge for CPU and memory over the lifetime of a warm instance, making idle inference state a direct serving cost. We present billing-aware neural…
-
Seamless Reconfiguration for DAG-BFT
arXiv cs.DC (Distributed)
arXiv:2610.00896v1 Announce Type: new Abstract: Byzantine Atomic Broadcast in Asynchronous networks has been studied extensively for decades. The FLP impossibility result rules out deterministic consensus in a fully asynchronous setting…
-
The Other Half of Workflow Portability: Evidence-Backed HPC Site Profiles with Agentic Discovery
arXiv cs.DC (Distributed)
arXiv:2610.00971v1 Announce Type: new Abstract: Moving a workflow developed and tested at one HPC site to another rarely succeeds without some amount of trial and error. Package managers rebuild software environments, containers ship whole…
-
Exploiting the Interplay of Compute- and Memory-Bound kernels in MPI Applications
arXiv cs.DC (Distributed)
arXiv:2610.01587v1 Announce Type: new Abstract: Parallel applications are often designed for synchronous, lock-step execution, treating communication stalls as performance hazards. Yet, in a communication-light application without frequent…
-
UNIQ: Communication-Efficient Distributed Quantum Computing via Unified Nonlinear Integer Programming
arXiv cs.DC (Distributed)
arXiv:2512.00401v2 Announce Type: replace-cross Abstract: Distributed quantum computing (DQC) is widely regarded as a promising approach to overcome quantum hardware limitations. A major challenge in DQC lies in reducing the communication…
-
Emergence-as-Code as a Foundation for Reliable Self-Governance
arXiv cs.DC (Distributed)
arXiv:2602.05458v3 Announce Type: replace-cross Abstract: Local adaptations can preserve component health while changing whether a system meets its reliability requirements. The system-level consequence depends on the interaction affected…
-
GridSMR: Causal Compression for Sharded Blockchains
arXiv cs.DC (Distributed)
arXiv:2610.01443v1 Announce Type: new Abstract: We present GridSMR, a sharded blockchain that scales execution horizontally while allowing dependent cross-shard operations to progress within a single block. Existing sharded systems…