# Distributed Systems & Parallel Computing

A computing discipline focused on coordinating multiple computational entities across locations and performing computations beyond the capacity of a single CPU, including challenges such as concurrency control, fault tolerance, algorithmic efficiency, and communication.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## RoCE and NVIDIA Quantum InfiniBand, and their quantified tradeoffs

DevFeed: [RoCE and NVIDIA Quantum InfiniBand, and their quantified tradeoffs](<https://devfeed.tech/articles/roce-and-nvidia-quantum-infiniband-and-their-quantified-tradeoffs-146467.md>)

Original publisher: [Read original article](<https://lambda.com/blog/roce-and-nvidia-quantum-infiniband-and-their-quantified-tradeoffs>)

Author: Zach Mueller

Published: 2026-11-20T14:08:48Z

Content type: comparison

Language: en

Sources: [The Lambda deep learning blog](<https://devfeed.tech/sources/the-lambda-deep-learning-blog.md>)

Topics: [InfiniBand](<https://devfeed.tech/topics/infiniband.md>), [rdma](<https://devfeed.tech/topics/rdma.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [NCCL](<https://devfeed.tech/topics/nccl.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>)

Tags: [1-click-cluster](<https://devfeed.tech/tags/1-click-cluster.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [infiniband](<https://devfeed.tech/tags/infiniband.md>), [llm-inference](<https://devfeed.tech/tags/llm-inference.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [quantum](<https://devfeed.tech/tags/quantum.md>), [roce](<https://devfeed.tech/tags/roce.md>), [tradeoffs](<https://devfeed.tech/tags/tradeoffs.md>)

### AI overview

RoCEv2 uses Ethernet with UDP/IP and can span subnets, while NVIDIA Quantum InfiniBand uses a dedicated fabric that requires InfiniBand switches and infrastructure. The text gives typical latency ranges of 0.5-2 µs for InfiniBand and 1-3 µs for optimized local RoCE, noting RoCE is more sensitive to configuration, congestion, and topology. In a two-node HGX B200 inference test using GLM-5.2-NVFP4 and SGLang, RoCE showed roughly 70% lower throughput than InfiniBand at about 250 concurrent users and 2.5x slower time to first token. The authors caution that real-world performance varies and recommend tempering expectations for On-Demand nodes; the comparison describes one setup.

### Source excerpt

The GPUs you rent are just part of what determines your cluster's speed and efficiency. This comes into play through interconnect, the network that connects multiple nodes (machines) on the backend.

## Towards Scale-Up Networking: Challenges and Open Questions

DevFeed: [Towards Scale-Up Networking: Challenges and Open Questions](<https://devfeed.tech/articles/towards-scale-up-networking-challenges-and-open-questions-146131.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26156>)

Author: Tingting Xu, Zixun Su, Xinyang Huang, Bowen Liu, Kun Tan, Dong Lin, Hao Wang, Yinben Xia, Weifeng Zhang, Binzhang Fu, Shuihai Hu, Kai Chen

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [NVLink](<https://devfeed.tech/topics/nvlink.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [bottleneck](<https://devfeed.tech/tags/bottleneck.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [collective](<https://devfeed.tech/tags/collective.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compatibility](<https://devfeed.tech/tags/compatibility.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [design](<https://devfeed.tech/tags/design.md>), [device](<https://devfeed.tech/tags/device.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [distributed-training](<https://devfeed.tech/tags/distributed-training.md>), [domain](<https://devfeed.tech/tags/domain.md>), [experts](<https://devfeed.tech/tags/experts.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [framework](<https://devfeed.tech/tags/framework.md>), [networking](<https://devfeed.tech/tags/networking.md>), [open](<https://devfeed.tech/tags/open.md>), [open-questions](<https://devfeed.tech/tags/open-questions.md>), [scale-up](<https://devfeed.tech/tags/scale-up.md>)

### AI overview

The paper derives scale-up network requirements from critical-path communication in distributed training and inference, focusing on domain size, per-accelerator bandwidth, and collective completion latency. It uses those requirements to compare representative proprietary and open fabrics, describing trade-offs in performance, scalability, compatibility, and implementation cost across their interconnects, topologies, protocols, and communication semantics. It also identifies open challenges in switching and physical interconnects, efficient communication, semantics, resource sharing, and connecting heterogeneous devices. This is a requirement-driven analysis; the supplied excerpt does not provide specific fabric-by-fabric results.

### Source excerpt

arXiv:2611.26156v1 Announce Type: cross Abstract: The rapid growth of AI models and the widespread use of distributed training and inference have made inter-accelerator communication a critical system bottleneck. Scale-up fabrics address this bottleneck by combining high per-accelerator bandwidth with low and predictable latency over tightly coupled accelerator domains. However, the appropriate topology, protocol mechanisms, communication semantics, and system integration remain actively debated. We analyze the critical-path communication generated by parallelism strategies and derive the resulting requirements for domain scale, per-accelerator bandwidth, and collective completion latency. We then use these requirements as a framework to examine representative proprietary and open scale-up fabrics, revealing how their physical interconnects, topologies, protocols, and semantics make different trade-offs among performance, scalability, compatibility, and implementation cost. Moreover, based on this analysis and discussions with industry experts, we identify challenges and open problems in physical interconnects and switching, efficient communication, semantics, resource sharing and heterogeneous device interconnection. This paper provides a requirement-driven foundation for analyzing current scale-up fabrics and guiding the design of future systems.

## Matching Architecture to Workload: Comparing RISC-V Parallel Accelerators for Energy-Efficient Embedded Computing

DevFeed: [Matching Architecture to Workload: Comparing RISC-V Parallel Accelerators for Energy-Efficient Embedded Computing](<https://devfeed.tech/articles/matching-architecture-to-workload-comparing-risc-v-parallel-accelerators-for-energy-efficient-embedded-computing-146128.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25947>)

Author: Luca Colombo, Luigi Ghionda, Amirhossein Kiamarzi, Yvan Tortorella, Riccardo Giunti, Niccol\\\`o Giuliani, Samuele Colmi, Angelo Garofalo, Giuseppe Tagliavini, Davide Rossi

Published: 2026-11-20T05:00:00Z

Content type: comparison

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [RISC-V](<https://devfeed.tech/topics/riscv.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Embedded Software Dev](<https://devfeed.tech/topics/embedded-software-dev.md>), [Sustainable Computing](<https://devfeed.tech/topics/sustainable-computing.md>)

Tags: [accelerators](<https://devfeed.tech/tags/accelerators.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [compare](<https://devfeed.tech/tags/compare.md>), [comparison](<https://devfeed.tech/tags/comparison.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [embedded](<https://devfeed.tech/tags/embedded.md>), [energy](<https://devfeed.tech/tags/energy.md>), [energy-efficiency](<https://devfeed.tech/tags/energy-efficiency.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [high-performance](<https://devfeed.tech/tags/high-performance.md>), [memory](<https://devfeed.tech/tags/memory.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [power](<https://devfeed.tech/tags/power.md>), [regular](<https://devfeed.tech/tags/regular.md>), [risc-v](<https://devfeed.tech/tags/risc-v.md>), [scalar](<https://devfeed.tech/tags/scalar.md>), [stream](<https://devfeed.tech/tags/stream.md>), [vector](<https://devfeed.tech/tags/vector.md>)

### AI overview

A comparison using open-source RISC-V accelerators implemented in the same 12 nm technology and evaluated on a common benchmark suite finds that no execution model leads on every workload. Spatz offers the largest gains for regular, compute-intensive applications with substantial vectorization, but is especially sensitive to data parallelism and regular computation and memory access. Snitch Cluster performs well when parallelism is limited and memory access is streamable. PULP Cluster is strongest on irregular, control-dominated applications. The abstract reports peak performance and energy-efficiency gains of up to 4.29x and 5.17x for Spatz, 5.53x and 1.45x for Snitch, and 5.66x and 2.80x for PULP, respectively. It also notes that PULP and Snitch's greater flexibility than Spatz costs area and power.

### Source excerpt

arXiv:2611.25947v1 Announce Type: new Abstract: Embedded computing demands high performance and energy efficiency across workloads with widely different computational characteristics. Choosing the most suitable execution model therefore requires a quantitative understanding of how each model maps to different workload classes. In this paper, we compare three execution models for energy-efficient embedded computing: scalar multicore, stream ing multicore, and vector processing. We use three open-source RISC-V accelerators as representative implementations: the PULP Cluster, the Snitch Cluster, and Spatz, respectively. Their open-source hardware and software stacks enable a controlled comparison based on implementations in the same 12 nm technology and a common benchmark suite. Our results reveal that no single execution model performs best across all workload classes. Spatz delivers the greatest benefits for regular and highly vectorizable compute-intensive applications, achieving up to 4.29x higher performance and 5.17x greater energy efficiency than the other architectures. However, its effectiveness strongly depends on the availability of data-level parallelism and regular computation and memory-access patterns, making it the most sensitive to workload characteristics. The Snitch Cluster performs particularly well on applications with limited parallelism and streamable memory-access patterns, providing up to 5.53x higher performance and 1.45x greater energy efficiency. Conversely, the PULP Cluster is most effective on irregular and control-dominated applications, achieving improvements of up to 5.66x in performance and 2.80x in energy efficiency. Compared with Spatz, however, the greater flexibility offered by PULP and Snitch comes at a non-negligible cost in terms of both area and power.

## LoongForge-Embodied: A Unified Framework for Training Heterogeneous Embodied Models

DevFeed: [LoongForge-Embodied: A Unified Framework for Training Heterogeneous Embodied Models](<https://devfeed.tech/articles/loongforge-embodied-a-unified-framework-for-training-heterogeneous-embodied-models-145515.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25460>)

Author: LoongForge Team

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Advanced AI workflows](<https://devfeed.tech/topics/advanced-ai-workflows.md>), [Machine Perception](<https://devfeed.tech/topics/machine-perception.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [acceleration](<https://devfeed.tech/tags/acceleration.md>), [architectures](<https://devfeed.tech/tags/architectures.md>), [concurrency](<https://devfeed.tech/tags/concurrency.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [data](<https://devfeed.tech/tags/data.md>), [data-infrastructure](<https://devfeed.tech/tags/data-infrastructure.md>), [ddp](<https://devfeed.tech/tags/ddp.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [execution](<https://devfeed.tech/tags/execution.md>), [fp8](<https://devfeed.tech/tags/fp8.md>), [framework](<https://devfeed.tech/tags/framework.md>), [fsdp](<https://devfeed.tech/tags/fsdp.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [i-o](<https://devfeed.tech/tags/i-o.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [layer](<https://devfeed.tech/tags/layer.md>), [memory](<https://devfeed.tech/tags/memory.md>), [native](<https://devfeed.tech/tags/native.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

LoongForge-Embodied uses a shared abstraction layer to make training infrastructure reusable across heterogeneous VLA and WAM architectures without constraining model-specific execution needs. The native-PyTorch framework combines shared and model- or hardware-specific optimizations for computation, memory, data I/O, execution, and distributed communication, including FP8 communication for DDP and FSDP. It also includes infrastructure for synthesizing robot data from egocentric data and concurrent simulator evaluation. Across representative models, the paper reports 1.47x-4.38x higher end-to-end training throughput than corresponding open-source baselines and 4.1x-23.2x faster evaluation under the same protocols.

### Source excerpt

arXiv:2611.25460v1 Announce Type: new Abstract: Training Vision-Language-Action (VLA) models and World-Action Models (WAMs) is computationally expensive, while their heterogeneous architectures and fragmented training stacks make system optimizations difficult to reuse across models. To address these challenges, we present LoongForge-Embodied, a native-PyTorch framework that unifies and accelerates the training of heterogeneous embodied models. At its core, LoongForge-Embodied introduces a common abstraction layer that decouples model-specific implementations from shared training infrastructure, enabling cross-model reuse without constraining individual architectures or execution requirements. Buildingon this unified infrastructure, the framework combines reusable acceleration mechanisms with model-specific and hardware-aware optimization strategies across computation, execution, memory, data I/O, and distributed communication, including FP8-based communication optimizations for DDP and FSDP. Beyond training acceleration, LoongForge-Embodied integrates a unified data infrastructure with a pipeline for synthesizing robot data from egocentric data, and an efficient evaluation framework that supports concurrency and optimized computation in simulator. Experiments across representative VLA and WAM models demonstrate 1.47x-4.38x end-to-end training throughput improvements over their corresponding open-source baselines, as well as 4.1x-23.2x faster evaluation under the same protocols.

## StarMPI-SVD: Distributed-Memory Parallel Star-M Tensor SVD

DevFeed: [StarMPI-SVD: Distributed-Memory Parallel Star-M Tensor SVD](<https://devfeed.tech/articles/starmpi-svd-distributed-memory-parallel-star-m-tensor-svd-146136.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26451>)

Author: Naman Pesricha, Grey Ballard, Srinivas Eswar, Md Taufique Hussain, Vishwas Rao, Phani Motamarri

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>), [Compression](<https://devfeed.tech/topics/compression.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [data](<https://devfeed.tech/tags/data.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [kernels](<https://devfeed.tech/tags/kernels.md>), [matrix](<https://devfeed.tech/tags/matrix.md>), [nodes](<https://devfeed.tech/tags/nodes.md>), [order](<https://devfeed.tech/tags/order.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [scalability](<https://devfeed.tech/tags/scalability.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [singular-value-decomposition](<https://devfeed.tech/tags/singular-value-decomposition.md>), [size](<https://devfeed.tech/tags/size.md>), [star](<https://devfeed.tech/tags/star.md>), [svd](<https://devfeed.tech/tags/svd.md>)

### AI overview

The work presents the first distributed-memory implementations of t-SVDM-I and adaptive t-SVDM-II for N-way tensors, extending star-M compression beyond shared-memory settings. t-SVDM-I uses invertible transformations and slice-wise truncated SVDs at a prescribed rank; t-SVDM-II selects slice-wise ranks to meet an error tolerance. Asynchronous, non-blocking communication is used for data redistribution to address computational and communication bottlenecks. Strong-scaling experiments report at least 70% parallel efficiency on up to 128 nodes (6,144 cores); the methods are applied to datasets up to 768 GB.

### Source excerpt

arXiv:2611.26451v1 Announce Type: new Abstract: The star-M tensor algebra generalizes the optimal compression guarantees of the matrix singular value decomposition (SVD) to higher-order tensors. The t-SVDM-I decomposition is computed using a set of invertible transformations and slice-wise truncated SVDs with a prescribed rank, and its adaptive, multirank variant t-SVDM-II can determine slice-wise ranks to satisfy a prescribed error tolerance. Existing implementations of these algorithms are limited to shared-memory environments, restricting their applicability to modern large-scale datasets. In this work, we present the first distributed-memory implementations of t-SVDM-I and t-SVDM-II, which are generalized to N-way tensors. To address computational and communication bottlenecks, we design asynchronous, non-blocking communication strategies for efficient data redistribution to obtain highly efficient key computational kernels. Strong-scaling experiments demonstrate the efficiency and scalability of our approach (at least 70% parallel efficiency on up to 128 nodes, or 6144 cores), and we apply the star-M compression techniques to a set of application datasets of size up to 768 GB.

## Provable Authority: A Protocol for Verifiable Authority Resolution and Control Across Human, Organizational, and Agentic Transactions

DevFeed: [Provable Authority: A Protocol for Verifiable Authority Resolution and Control Across Human, Organizational, and Agentic Transactions](<https://devfeed.tech/articles/provable-authority-a-protocol-for-verifiable-authority-resolution-and-control-across-human-organizational-and-agentic-transactions-146009.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.22899>)

Author: Jeremy Blackburn, Tim Hines, Matthew Koepp, Russell Lessard, Jason Patrick

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [control-plane](<https://devfeed.tech/topics/control-plane.md>)

Tags: [ach](<https://devfeed.tech/tags/ach.md>), [adapter](<https://devfeed.tech/tags/adapter.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-runtime](<https://devfeed.tech/tags/agent-runtime.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-commerce](<https://devfeed.tech/tags/agentic-commerce.md>), [agentic-transactions](<https://devfeed.tech/tags/agentic-transactions.md>), [ai](<https://devfeed.tech/tags/ai.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [assertions](<https://devfeed.tech/tags/assertions.md>), [authorization](<https://devfeed.tech/tags/authorization.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [budget](<https://devfeed.tech/tags/budget.md>), [canonical](<https://devfeed.tech/tags/canonical.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-ma](<https://devfeed.tech/tags/cs-ma.md>), [organizational](<https://devfeed.tech/tags/organizational.md>), [protocol](<https://devfeed.tech/tags/protocol.md>), [transactions](<https://devfeed.tech/tags/transactions.md>)

### AI overview

The ChainIT Authority Protocol proposes a pre-execution control plane to let relying parties verify that a person, organization, or AI agent is authorized for the exact transaction being submitted. It keeps identity, versioned organizational authority, evidence, policy checks, and configured dispositions distinct; binds decisions and human approvals to a canonical transaction digest; reserves capacity atomically; and gates commit on a single-use, sender-constrained credential consumed at the rail adapter. For AI agents, an Agent Subject Profile adds workload registration, attested keys, and a sovereignty boundary. The abstract says it specifies bounded-freshness status rules and enforcement profiles for fiat and stablecoin rails, while also identifying residual risks and non-goals.

### Source excerpt

arXiv:2610.22899v2 Announce Type: replace Abstract: Payment systems authenticate users, devices, accounts, and cryptographic credentials, but a relying party, such as a counterparty, payment rail, marketplace, or autonomous agent runtime, generally cannot obtain compact, independently verifiable evidence that the party presenting one exact transaction holds current organizational authority to execute it under the applicable evidence, policy, budget, and approval state. Agentic commerce makes this gap operationally urgent: a software agent can produce a valid signature while acting outside its delegated purpose, after its sponsor's authority has changed, or on transaction details that differ from those a human approved. We describe the ChainIT Authority Protocol, a subject-neutral and rail-neutral pre-execution authority and evidence control plane. The protocol separates durable identity, a versioned organizational authority record (the Authority Resolution Pactvera), sourced evidence assertions, deterministic reason-coded policy predicates, and organization-configured dispositions; binds each decision and human approval to a canonical transaction digest; reserves capacity atomically so that limits hold under concurrency; and authorizes commit only through a single-use, sender-constrained Execution Authorization Credential consumed at the rail adapter, with append-only lifecycle evidence sealed after the terminal outcome. An Agent Subject Profile adds the workload registration, attested key hierarchy, and sovereignty boundary required when the acting subject is an AI agent, without introducing a separate authority concept. We specify the commit-time execution predicate and bounded-freshness status rules, give enforcement profiles for fiat (ACH, wire, card, instant payment) and stablecoin rails, and state residual risks and non-goals.

## Topology-Driven Anti-Entanglement Control for Soft Robots

DevFeed: [Topology-Driven Anti-Entanglement Control for Soft Robots](<https://devfeed.tech/articles/topology-driven-anti-entanglement-control-for-soft-robots-145326.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.05236>)

Author: Haoyang Le, Shengxuan Wang, Mohan Chen, Shuo Feng

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [Physical AI](<https://devfeed.tech/topics/physical-ai.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [control](<https://devfeed.tech/tags/control.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [entanglement](<https://devfeed.tech/tags/entanglement.md>), [robots](<https://devfeed.tech/tags/robots.md>), [topology](<https://devfeed.tech/tags/topology.md>)

### AI overview

The paper proposes topology-driven multi-agent reinforcement learning (TD-MARL) to coordinate soft robots unwinding in constrained environments. It combines centralized learning, where agents share topological state to account for one another's strategies and reduce training instability, with distributed execution that removes the need for inter-robot communication. A topological safety layer uses invariants to assess and mitigate entanglement risk. The abstract reports that full simulation experiments outperform current deep reinforcement learning methods in convergence and anti-entanglement performance; it provides no quantitative results or experimental details.

### Source excerpt

arXiv:2605.05236v3 Announce Type: replace-cross Abstract: In the field of precision manufacturing in complex constrained environments, the role of soft robots is increasingly prominent, and the realization of anti-winding control based on multi-intelligent body reinforcement learning has become a research hotspot. One of the core problems at present is to coordinate multiple robots to complete the unwinding operation in a highly constrained environment. The existing distributed training framework faces some observability challenges in high-density barrier and unstable environments, resulting in poor learning results. This paper proposes a topology-driven Multi-Agent Reinforcement Learning (TD-MARL) framework to coordinate multi-robot systems to avoid entanglement. Specifically, the critical network adopts centralized learning, so that each intelligent body can perceive the strategies of other intelligent bodies by sharing the topological state, thus alleviating the training instability caused by complex interactions; eliminating the demand for communication resources between robots through distributed execution, Upgrade system reliability; the integrated topological security layer uses topological invariants to accurately assess and mitigate the risk of entanglement to avoid the strategy from falling into local difficulties. Finally, the full simulation experiments carried out in the real simulation environment show that the method is better than the current advanced deep reinforcement learning (DRL) method in terms of convergence and anti-winding effect.

## Introducing Auto Scaling in YugabyteDB Aeon

DevFeed: [Introducing Auto Scaling in YugabyteDB Aeon](<https://devfeed.tech/articles/introducing-auto-scaling-in-yugabytedb-aeon-144173.md>)

Original publisher: [Read original article](<https://www.yugabyte.com/blog/introducing-auto-scaling-in-yugabytedb-aeon/>)

Author: Kyle Hailey

Published: 2026-11-19T13:00:54Z

Content type: release

Language: en

Sources: [The Distributed SQL Blog](<https://devfeed.tech/sources/the-distributed-sql-blog.md>)

Topics: [Deployment & Scaling](<https://devfeed.tech/topics/deployment-scaling.md>), [Cloud Architecture](<https://devfeed.tech/topics/cloud-architecture.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [internal developer platform automation](<https://devfeed.tech/topics/internal-developer-platform-automation.md>)

Tags: [aeon](<https://devfeed.tech/tags/aeon.md>), [auto-scaling](<https://devfeed.tech/tags/auto-scaling.md>), [autoscaling](<https://devfeed.tech/tags/autoscaling.md>), [blog](<https://devfeed.tech/tags/blog.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [falls](<https://devfeed.tech/tags/falls.md>), [first](<https://devfeed.tech/tags/first.md>), [guardrails](<https://devfeed.tech/tags/guardrails.md>), [load](<https://devfeed.tech/tags/load.md>), [management](<https://devfeed.tech/tags/management.md>), [more](<https://devfeed.tech/tags/more.md>), [online](<https://devfeed.tech/tags/online.md>), [plate](<https://devfeed.tech/tags/plate.md>), [release-announcements](<https://devfeed.tech/tags/release-announcements.md>), [routine](<https://devfeed.tech/tags/routine.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [set](<https://devfeed.tech/tags/set.md>), [takes](<https://devfeed.tech/tags/takes.md>), [yugabytedb](<https://devfeed.tech/tags/yugabytedb.md>), [yugabytedb-aeon](<https://devfeed.tech/tags/yugabytedb-aeon.md>)

### AI overview

YugabyteDB Aeon Autoscaling is in Early Access and automatically adjusts node counts within configured minimum and maximum limits using average CPU and YSQL connection usage. Scale-out occurs when any rule fires; scale-in requires all configured conditions to hold, with cooldowns and maintenance pauses limiting repeated changes. Scaling proceeds in availability-zone-sized increments, and data rebalances online. Setup requires Support enablement and a single-region cluster; Terraform users can define policies with the Aeon provider and must configure node-count drift handling for active policies. The feature supports horizontal scaling only; vertical and disk scaling, scheduled policies, multi-region clusters, and QPS or latency triggers are listed as roadmap items.

### Source excerpt

Introducing Autoscaling for YugabyteDB Aeon! Autoscaling takes routine capacity management off your team's plate. Set the bounds once, and YugabyteDB Aeon scales the cluster up when load rises and down when load falls, with online rebalancing and zone-aware guardrails at every step. Read this blog to find out more. The post Introducing Auto Scaling in YugabyteDB Aeon appeared first on Yugabyte.

## Multilevel Graph Partitioning on GPU Supercomputers

DevFeed: [Multilevel Graph Partitioning on GPU Supercomputers](<https://devfeed.tech/articles/multilevel-graph-partitioning-on-gpu-supercomputers-142978.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24264>)

Author: Michael S. Gilbert, Erik G. Boman, Kamesh Madduri

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Multi-GPU](<https://devfeed.tech/topics/multi-gpu.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [GPU](<https://devfeed.tech/topics/gpu.md>), [graph theory](<https://devfeed.tech/topics/graph-theory.md>)

Tags: [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [graph](<https://devfeed.tech/tags/graph.md>), [multi-gpu](<https://devfeed.tech/tags/multi-gpu.md>), [optimize](<https://devfeed.tech/tags/optimize.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [partitioning](<https://devfeed.tech/tags/partitioning.md>), [scalability](<https://devfeed.tech/tags/scalability.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [supercomputers](<https://devfeed.tech/tags/supercomputers.md>)

### AI overview

The work presents a distributed multi-GPU graph partitioner based on the single-GPU Jet partitioner, aiming to improve scalability while preserving solution quality. It uses CUDA-aware MPI, one-sided communication, and a bulk-synchronous parallel structure. The authors report solution quality comparable to Jet and weak scaling up to 256 GPUs, allowing it to partition instances larger than a single-GPU partitioner can handle.

### Source excerpt

arXiv:2611.24264v1 Announce Type: new Abstract: Graph partitioning is a key enabler in computational science, with applications to sparse linear algebra, load-balanced parallel computation, and graph computations. In recent years, graphics processing units (GPUs) have become the preferred accelerator on the world's top supercomputers. Despite this, most existing distributed graph partitioners do not utilize GPUs. In this work, we develop a new multi-GPU distributed partitioner based on Jet, a prior single-GPU partitioner. We seek to optimize scalability without sacrificing quality. Our partitioner utilizes CUDA-aware MPI, one-sided communication routines, and an overall bulk-synchronous parallel structure. We demonstrate solution quality comparable to Jet. Our experiments show excellent weak scaling up to 256 GPUs, enabling partitioning of instances significantly larger than a single-GPU partitioner can handle.

## Cap-and-Track: Optimizing LLM Inference Request Dispatch Across Shifting Capacity

DevFeed: [Cap-and-Track: Optimizing LLM Inference Request Dispatch Across Shifting Capacity](<https://devfeed.tech/articles/cap-and-track-optimizing-llm-inference-request-dispatch-across-shifting-capacity-143577.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24891>)

Author: Benjamin Novotny, Doug Szeto, Chang Wang, Shicheng Chen, George Zhihua Yuan, Bob Kamma, Michael Ge, John Xin Zhou, Andrea Olgiati

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [backpressure](<https://devfeed.tech/topics/backpressure.md>), [rate-limiting](<https://devfeed.tech/topics/rate-limiting.md>), [async](<https://devfeed.tech/topics/async.md>)

Tags: [about](<https://devfeed.tech/tags/about.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [admission](<https://devfeed.tech/tags/admission.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [backend](<https://devfeed.tech/tags/backend.md>), [batches](<https://devfeed.tech/tags/batches.md>), [can](<https://devfeed.tech/tags/can.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [change](<https://devfeed.tech/tags/change.md>), [classifier](<https://devfeed.tech/tags/classifier.md>), [concurrent](<https://devfeed.tech/tags/concurrent.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [design](<https://devfeed.tech/tags/design.md>), [dispatch](<https://devfeed.tech/tags/dispatch.md>), [dispatcher](<https://devfeed.tech/tags/dispatcher.md>), [inference](<https://devfeed.tech/tags/inference.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-inference](<https://devfeed.tech/tags/llm-inference.md>), [optimizing](<https://devfeed.tech/tags/optimizing.md>), [request](<https://devfeed.tech/tags/request.md>)

### AI overview

Cap-and-Track dispatches asynchronous LLM inference using backend refusals as signals of current capacity. Cap rejects work before it incurs a long wait; durable priority queues hold excess and refused requests for retry. Track probes upward and backs off using those refusals, avoiding output-length prediction and a fixed fleet-wide rate. In a same-traffic production A/B, the system served a classifier with 43% fewer H100s than its queued, fixed-rate predecessor, though lower-priority work waited longer during backfill peaks. In steady-capacity engine replay, it achieved within 1% of a fixed rate tuned above capacity, with about one-quarter its refusal share. The authors also recommend keeping engine queues only long enough to fill batches, maintaining stateful dispatch, and sizing fleets according to tolerable waits for low-priority work.

### Source excerpt

arXiv:2611.24891v1 Announce Type: new Abstract: A fixed dispatch rate for LLM inference becomes stale as serving fleets resize and request sizes change. Generative output length is unknown at admission, and shared provider pools can lose capacity. We present Cap-and-Track, a design for asynchronous inference across provider capacity and internal GPUs. Meta's traffic shaper implements Cap-and-Track for post and live-video understanding. Refusals signal the backend's current admission ceiling. Durable priority queues hold excess work outside execution, including refused requests for retry. Cap is a fail-fast admission check at the saturated resource that refuses work before a long wait. Using only those refusals, the Track dispatcher probes upward and backs off. It follows the ceiling without output-length prediction or a fleet-level rate to retune. In a concurrent same-traffic production A/B, the shaper ran per-replica Cap plus Track. It served a classifier on 43% fewer H100s than its queued, fixed-rate precursor, though lower-priority work waited longer at backfill peaks. In steady-capacity real-engine replay, Track ran within 1% of the goodput of a fixed rate tuned above measured capacity, with about a quarter of that rate's refusal share. The shaper carries more than 300 million online requests and hundreds of terabytes of media a day. During a backfill, it handled 1.1 billion requests in its busiest 24 hours. Lessons include keeping queues inside inference engines just long enough to fill batches, steering dispatch by capacity refusals alone, keeping dispatch stateful, and sizing fleets by how long low-priority work can wait.

## ScaleBench: An Extensible and Reproducible Experimental Platform for Kubernetes Autoscaling Research

DevFeed: [ScaleBench: An Extensible and Reproducible Experimental Platform for Kubernetes Autoscaling Research](<https://devfeed.tech/articles/scalebench-an-extensible-and-reproducible-experimental-platform-for-kubernetes-autoscaling-research-142980.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24398>)

Author: Ze Wang, Sajid Alam, Amjad Ullah

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Kubernetes](<https://devfeed.tech/topics/kubernetes.md>), [Deployment & Scaling](<https://devfeed.tech/topics/deployment-scaling.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [autoscaling](<https://devfeed.tech/tags/autoscaling.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [experimental](<https://devfeed.tech/tags/experimental.md>), [integration](<https://devfeed.tech/tags/integration.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-autoscaling](<https://devfeed.tech/tags/kubernetes-autoscaling.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [optimisation](<https://devfeed.tech/tags/optimisation.md>), [performance-evaluation](<https://devfeed.tech/tags/performance-evaluation.md>), [platform](<https://devfeed.tech/tags/platform.md>), [research](<https://devfeed.tech/tags/research.md>)

### AI overview

ScaleBench is a native Kubernetes platform for evaluating autoscalers in a repeatable way while accounting for deployment and integration alongside scaling performance. The paper identifies a practical adoption gap: Kubernetes commonly uses the reactive, CPU-based Horizontal Pod Autoscaler, while research on newer methods has paid limited attention to their deployment complexity and operational overhead. ScaleBench integrates autoscalers as native Kubernetes components, and a case study evaluating multiple approaches demonstrates its use for systematic assessment in realistic cloud-native environments. The excerpt does not report specific comparative results.

### Source excerpt

arXiv:2611.24398v1 Announce Type: new Abstract: Autoscaling enables cloud-native applications to meet their SLOs by dynamically adjusting computational resources at runtime. Recent research has explored a broad range of advanced autoscaling approaches, including rule-based, optimisation-based, and machine learning-based techniques, to address the limitations of traditional resource management strategies in cloud-native environments. Despite significant advances, a considerable gap remains between academic innovations and their practical adoption in real-world environments. For instance, Kubernetes deployments commonly rely on the reactive, CPU-based Horizontal Pod Autoscaler, limiting the adoption of more advanced autoscaling approaches. Recent studies have investigated challenges such as autoscaler configuration and reproducible performance evaluation; however, the practical deployment, integration complexity, and operational overhead of research-based autoscalers in Kubernetes environments have received limited attention. To address this gap, we present ScaleBench, a native Kubernetes experimental platform to enable extensible and reproducible evaluation of autoscaling solutions. Unlike existing approaches, ScaleBench integrates autoscalers as native Kubernetes components, allowing researchers to assess not only scaling performance but also deployment characteristics and integration requirements within realistic cloud-native environments. We demonstrate the effectiveness of ScaleBench by evaluating multiple autoscaling approaches through a case study. The results highlight its capability to support systematic, repeatable, and practical assessment of Kubernetes autoscaling approaches.

## A Fast, Programmable GPU-to-CPU Invocation Layer for GPU-Initiated Communication

DevFeed: [A Fast, Programmable GPU-to-CPU Invocation Layer for GPU-Initiated Communication](<https://devfeed.tech/articles/a-fast-programmable-gpu-to-cpu-invocation-layer-for-gpu-initiated-communication-142971.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23848>)

Author: Jiakun Yan, Shiksha Patel, David Sidler, Gengyu Rao, Brandon Potter, Bradford Beckmann

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [gpu-initated](<https://devfeed.tech/topics/gpu-initated.md>)

Tags: [async](<https://devfeed.tech/tags/async.md>), [backend](<https://devfeed.tech/tags/backend.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [code](<https://devfeed.tech/tags/code.md>), [communication](<https://devfeed.tech/tags/communication.md>), [control](<https://devfeed.tech/tags/control.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [device](<https://devfeed.tech/tags/device.md>), [dispatch](<https://devfeed.tech/tags/dispatch.md>), [expert-parallel](<https://devfeed.tech/tags/expert-parallel.md>), [experts](<https://devfeed.tech/tags/experts.md>), [fallback](<https://devfeed.tech/tags/fallback.md>), [framework](<https://devfeed.tech/tags/framework.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [gpudirect](<https://devfeed.tech/tags/gpudirect.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [kernels](<https://devfeed.tech/tags/kernels.md>), [latency](<https://devfeed.tech/tags/latency.md>)

### AI overview

KIWI makes reverse offload to host proxy threads a programmable communication path for GPU workloads, rather than treating it as a slower fallback to GPUDirect Async (GDA). The framework targets shared-queue contention, polling overhead, and proxy load imbalance, and reports sub-microsecond one-way device-to-host latency and hundreds of millions of operations per second. Built on KIWI, KIWI-P2P outperforms state-of-the-art GDA implementations in latency and bandwidth. KIWI-EP uses workload-defined operations for mixture-of-experts dispatch and combine, reducing expert-parallel communication latency by up to 2.3x over the best existing backend on AMD GPUs and vLLM time per output token by up to 29%.

### Source excerpt

arXiv:2611.23848v1 Announce Type: new Abstract: Modern GPU workloads increasingly rely on device-initiated communication to let GPU kernels communicate directly without synchronizing with the host at kernel boundaries. Existing libraries commonly favor placing transport control on GPU threads, an approach known as GPUDirect Async (GDA), and treat reverse offload to host proxy threads as a slower fallback. We argue that reverse offload can outperform GDA and should be a first-class communication path. We further propose exposing reverse offload as a programming interface that lets users define workload-specific communication operations. We present KIWI, a programmable reverse-offload framework that achieves sub-microsecond one-way device-to-host latency and hundreds of millions of operations per second. We address practical challenges of scaling reverse offload, including contention on shared queues, polling overhead, and load imbalance across proxies. KIWI-P2P, a SHMEM-like library built on KIWI, outperforms state-of-the-art GDA implementations in both latency and bandwidth. More importantly, KIWI-EP redesigns low-latency mixture-of-experts dispatch and combine kernels around workload-defined operations, lowering expert-parallel communication latency by up to 2.3x over the best existing backend on AMD GPUs and time per output token in vLLM by up to 29%. The source code of KIWI is available at https://github.com/AMDResearch/kiwi.

## Accelerating Distributed Digital Ising Machines

DevFeed: [Accelerating Distributed Digital Ising Machines](<https://devfeed.tech/articles/accelerating-distributed-digital-ising-machines-142974.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23883>)

Author: Tong Wu, Ziyad Alswaidan, Shreya Mukherjee, Ting-Yu Lee, Daniel Xu, Atharva Raut, Harry Tye, Abhinav K. Puthran, Kemal Selcuk, Navid A. Aadit, Ken Mai, Prashanth Mohan, Kerem Y. Camsari, Tathagata Sri

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [performance-optimization](<https://devfeed.tech/topics/performance-optimization.md>)

Tags: [accelerators](<https://devfeed.tech/tags/accelerators.md>), [combinatorial-optimization](<https://devfeed.tech/tags/combinatorial-optimization.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compute](<https://devfeed.tech/tags/compute.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataflow](<https://devfeed.tech/tags/dataflow.md>), [devices](<https://devfeed.tech/tags/devices.md>), [digital](<https://devfeed.tech/tags/digital.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [error](<https://devfeed.tech/tags/error.md>), [fpga](<https://devfeed.tech/tags/fpga.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [ising](<https://devfeed.tech/tags/ising.md>), [latency](<https://devfeed.tech/tags/latency.md>), [markov-chain-monte-carlo](<https://devfeed.tech/tags/markov-chain-monte-carlo.md>), [mcmc](<https://devfeed.tech/tags/mcmc.md>), [memory](<https://devfeed.tech/tags/memory.md>), [mesh](<https://devfeed.tech/tags/mesh.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>)

### AI overview

The work models distributed digital Ising sweep time as compute, memory, synchronization, and inter-device communication, with 4.8% mean error across three GPU generations, FPGA, and ASIC configurations up to one million spins. Using that model, it reports up to 4.2x speedup from data placement and up to 3.16x lower time-to-solution on three benchmarks with multi-rate and bounded-staleness schedules that relax Gibbs sampling barriers. A phase-offset streaming design overlaps boundary exchange with interior computation when links meet a latency-transparency criterion; reported speedups are 1.25x on an H200 pair, 1.74x on a ZCU106 pair, and 2.01x on a four-board XEM8320 mesh. A fabricated four-chip 28 nm ASIC implementation sustains 28.1 Gflips/s at 4.81 pJ/flip.

### Source excerpt

arXiv:2611.23883v1 Announce Type: cross Abstract: Ising machines, dedicated accelerators built around the dynamics of Ising-like spin models, have emerged as a leading platform for combinatorial optimization and probabilistic sampling. The problems that motivate them require millions of spins, far beyond the on-chip capacity of any single device, so a practical Ising machine must distribute its computation across many devices. Prior work reports throughput at a single operating point, leaving the binding resource unidentified and platform crossovers measurable only by experiment; no calibrated cross-platform cost model of distributed Ising machines has been reported, nor have systematic techniques to accelerate their Markov chain Monte Carlo (MCMC) workloads. This work presents a calibrated cost model that decomposes digital Ising sweep time into compute, memory, synchronization, and inter-device communication terms, validated with 4.8% mean error across three GPU generations, FPGA and ASIC up to 1 million spins. Guided by the model, we address each term in turn and propose: (i) a data placement strategy, worth up to 4.2$\times$ speedup; (ii) multi-rate and bounded-staleness schedules that relax the barrier that Gibbs sampling imposes, cutting time-to-solution (TTS) by up to 3.16$\times$ on three benchmarks; and (iii) a phase-offset streaming dataflow that hides boundary exchange behind interior computation whenever a link satisfies a latency-transparency criterion, yielding up to 1.25$\times$/1.74$\times$/2.01$\times$ on an H200 pair, a ZCU106 pair, and a four-board XEM8320 $2{\times}2$ mesh. The streaming dataflow is also implemented on a fabricated four-chip 28 nm ASIC sustaining 28.1 Gflips/s at 4.81 pJ/flip.

## Efficient simulation of variational quantum circuits: representation, accuracy and CPU/GPU allocation

DevFeed: [Efficient simulation of variational quantum circuits: representation, accuracy and CPU/GPU allocation](<https://devfeed.tech/articles/efficient-simulation-of-variational-quantum-circuits-representation-accuracy-and-cpu-gpu-allocation-142982.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24444>)

Author: Vojt\\v{e}ch Nov\\'ak, Radovan Re\\v{c}ka, Ond\\v{r}ej Knebl, Ivan Zelinka

Published: 2026-11-19T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Simulation](<https://devfeed.tech/topics/simulation.md>), [Quantum Computing](<https://devfeed.tech/topics/quantum-computing.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerate](<https://devfeed.tech/tags/accelerate.md>), [acceleration](<https://devfeed.tech/tags/acceleration.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [benefits](<https://devfeed.tech/tags/benefits.md>), [can](<https://devfeed.tech/tags/can.md>), [change](<https://devfeed.tech/tags/change.md>), [circuit](<https://devfeed.tech/tags/circuit.md>), [circuit-simulation](<https://devfeed.tech/tags/circuit-simulation.md>), [circuits](<https://devfeed.tech/tags/circuits.md>), [compare](<https://devfeed.tech/tags/compare.md>), [complete](<https://devfeed.tech/tags/complete.md>), [complex](<https://devfeed.tech/tags/complex.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [quant-ph](<https://devfeed.tech/tags/quant-ph.md>), [quantum](<https://devfeed.tech/tags/quantum.md>), [quantum-circuits](<https://devfeed.tech/tags/quantum-circuits.md>), [simulation](<https://devfeed.tech/tags/simulation.md>)

### AI overview

The study compares statevector simulation, direct Pauli evolution, and matrix product states for molecular and spin variational Hamiltonian ansätze and hardware-efficient circuits across CPU/GPU systems. In the tested implementations, direct Pauli evolution benefits complex molecular circuits, while GPUs accelerate wider spin and hardware-efficient circuits. Matrix product states can reduce cost when entanglement is limited enough for accurate compression. Corrected CPU FP32 met the tested tolerance but cost more than FP64. Allocation choices depend on workload size, startup costs, worker memory, and profiling: matched NumPy Pauli updates sped up two-layer full LiH by 3.45x, and a GPU was about 30x faster than 18-thread Aer for a 28-qubit, four-layer TFIM, both in FP64. The proposed procedure jointly selects representation, precision, and worker allocation by comparing complete energy-evaluation costs under accuracy, memory, and core constraints.

### Source excerpt

arXiv:2611.24444v1 Announce Type: cross Abstract: Classical circuit simulation supports quantum software development and verification and provides a baseline for assessing quantum advantage. We compare gate-based statevector simulation, direct Pauli evolution and matrix product states for molecular and spin Variational Hamiltonian Ansatze and hardware-efficient ansatze on CPU/GPU systems, varying width, depth, precision and resource allocation. In the tested implementations, direct Pauli evolution benefits complex molecular VHA, whereas GPUs accelerate wider spin and hardware efficient circuits. Matrix product state is useful when limited entanglement permits accurate compression; corrected CPU FP32 can meet the tested tolerance while costing more than FP64. Workload size, startup, worker memory and task-specific profiling can change the preferred allocation. Matched NumPy Pauli updates accelerate two-layer full LiH by 3.45-fold; the GPU gives approximately 30-fold acceleration over 18-thread Aer for 28-qubit, four-layer TFIM, both in FP64. We provide a procedure for choosing representation, precision and worker allocation jointly, comparing the cost of complete energy evaluations under accuracy, memory and core constraints.

## Programmable Probabilistic Computer with 1,000,000 p-bits

DevFeed: [Programmable Probabilistic Computer with 1,000,000 p-bits](<https://devfeed.tech/articles/programmable-probabilistic-computer-with-1-000-000-p-bits-142961.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.25313>)

Author: Navid Anjum Aadit, Xiuqi Zhang, Shuvro Chowdhury, Kevin Callahan-Coray, Kyle Lee, Saleh Bunaiyan, Sanjay Seshan, Clayton Thomas, Jason Twigg, Andrew Seawright, Forrest Brewer, Tathagata Srimani, Kerem

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [fpga](<https://devfeed.tech/topics/fpga.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [Combinatorial optimization](<https://devfeed.tech/topics/combinatorial-optimization.md>), [Supercomputing](<https://devfeed.tech/topics/supercomputing.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [bits](<https://devfeed.tech/tags/bits.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [computer](<https://devfeed.tech/tags/computer.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [energy](<https://devfeed.tech/tags/energy.md>), [exchange](<https://devfeed.tech/tags/exchange.md>), [field](<https://devfeed.tech/tags/field.md>), [gate](<https://devfeed.tech/tags/gate.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [graph](<https://devfeed.tech/tags/graph.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [interconnect](<https://devfeed.tech/tags/interconnect.md>), [learning](<https://devfeed.tech/tags/learning.md>), [limit](<https://devfeed.tech/tags/limit.md>), [local](<https://devfeed.tech/tags/local.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [memory](<https://devfeed.tech/tags/memory.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [probabilistic](<https://devfeed.tech/tags/probabilistic.md>), [sampling](<https://devfeed.tech/tags/sampling.md>)

### AI overview

The computer scales to one million interacting p-bits across 18 FPGAs, sustaining 10¹² probabilistic updates per second. Coupling weights stay in on-chip memory beside the update circuits, while devices exchange only single-bit boundary states. In spin-glass experiments, the ratio of boundary-exchange frequency to local-update frequency determines how communication delays affect optimization. A graph-partition and interconnect bound identifies conditions that recover the residual-energy decay of an unpartitioned GPU reference. Infrequent exchange lowers the decay exponent, though faster local updates can still reduce the time to reach a target energy. A cluster mean-field model reproduces the dependence, and ferromagnetic sampling shows a corresponding effect on magnetization relaxation. The abstract presents this frequency ratio as a design principle for larger probabilistic computers.

### Source excerpt

arXiv:2606.25313v2 Announce Type: replace Abstract: Many problems in physics and machine learning require sampling the collective states of large numbers of interacting variables. On modern graphics processors, repeated movement of interaction data between memory and compute units can limit how efficiently these calculations can scale. Here we demonstrate a programmable probabilistic computer with one million interacting probabilistic bits (p-bits) across 18 field-programmable gate arrays, sustaining $10^{12}$ probabilistic updates per second. Coupling weights remain beside the update circuits in on-chip memory, and only single-bit boundary states pass between devices. Experiments on three-dimensional and quasi-two-dimensional spin glasses show that the ratio of boundary-exchange frequency to local update frequency controls the effect of communication delays on optimization. A bound derived from the graph partition and interconnect identifies operating conditions that recover the residual-energy decay of an unpartitioned GPU reference. Infrequent exchange reduces the decay exponent, although faster local updates can shorten the time to reach a target energy. A cluster mean-field model reproduces this dependence, while ferromagnetic sampling experiments reveal a corresponding dependence of magnetization relaxation on the same frequency ratio. These results establish a million-p-bit computing architecture in which processing and local memory scale together, and provide a quantitative design principle for exploiting tolerance to communication delays in larger probabilistic computers.

## DispersedBFT: Bottleneck-Free Primary-Backup Consensus using Dispersed Replication

DevFeed: [DispersedBFT: Bottleneck-Free Primary-Backup Consensus using Dispersed Replication](<https://devfeed.tech/articles/dispersedbft-bottleneck-free-primary-backup-consensus-using-dispersed-replication-142290.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24833>)

Author: Nancy Kansal, Celine Sana, Jelle Hellings

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Databases](<https://devfeed.tech/sources/arxiv-databases.md>), [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [consensus algorithm](<https://devfeed.tech/topics/consensus-algorithm.md>), [performance-engineering](<https://devfeed.tech/topics/performance-engineering.md>), [Blockchain](<https://devfeed.tech/topics/blockchain.md>)

Tags: [bottleneck](<https://devfeed.tech/tags/bottleneck.md>), [consensus](<https://devfeed.tech/tags/consensus.md>), [cs-db](<https://devfeed.tech/tags/cs-db.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [replication](<https://devfeed.tech/tags/replication.md>)

### AI overview

DispersedBFT aims to remove the primary bottleneck in PBFT-like consensus by using information dispersal to distribute request replication from the primary to backup replicas. Its variants trade CPU cost against communication cost: one uses Message Authentication Codes and Merkle proofs for lower CPU cost, while another uses digital signatures for lower communication cost. The abstract reports throughput of 3.4 million transactions per second with 43 replicas, up to 6.2 times faster than PBFT and 2.6 times faster than HotStuff. These are reported experimental results; the supplied text does not provide evaluation details.

### Source excerpt

arXiv:2611.24833v1 Announce Type: new Abstract: The advent of blockchains is fueling the development of novel resilient data management systems (RDSs). Typically, these RDSs are managed using a primary-backup design using a consensus protocol such as PBFT. In these primary-backup designs, a single participant, the primary, is responsible for coordinating the replication and execution of requests. Unfortunately, the central role of the primary makes it a bottleneck that drastically limits performance. In this paper, we propose DispersedBFT, a high performance PBFT-like consensus protocol that utilizes dispersed replication to eliminate the primary bottleneck of PBFT-like consensus. In specific, DispersedBFT uses information dispersal to distribute the task of replicating request from the primary to the backups. We propose several variants of DispersedBFT with either low CPU cost (by using Message Authentication Codes and Merkle proofs) or communication cost (by using digital signatures). We practically analyze the performance of DispersedBFT and show it greatly outperforms PBFT and HotStuff: DispersedBFT supports a throughput of 3.4Mtxn/sec with 43 replicas and is up-to 6.2x faster than PBFT and 2.6x faster than HotStuff.

## DexAttention: Distributed Sparse Attention through Rapid Online Planning and Persistent GPU Execution

DevFeed: [DexAttention: Distributed Sparse Attention through Rapid Online Planning and Persistent GPU Execution](<https://devfeed.tech/articles/dexattention-distributed-sparse-attention-through-rapid-online-planning-and-persistent-gpu-execution-142968.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23512>)

Author: Sirui Chen, Xuehai Qian

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [distributed-training](<https://devfeed.tech/topics/distributed-training.md>), [scheduling](<https://devfeed.tech/topics/scheduling.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [execution](<https://devfeed.tech/tags/execution.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [online](<https://devfeed.tech/tags/online.md>), [planning](<https://devfeed.tech/tags/planning.md>), [sparse-attention](<https://devfeed.tech/tags/sparse-attention.md>)

### AI overview

DexAttention co-designs online planning and persistent GPU execution to reduce communication and per-operation overhead in distributed sparse attention. Its GPU accelerated, topology aware assigner balances compute and communication; a greedy scheduler orders coarse operations, while persistent kernels enforce fine grained dependencies and dispatch ready tasks. This separates detailed synchronization from planning and supports overlap without repeated attention launches. The framework supports static and dynamic per head masks across multiple attention kernels. On a four node, 16 GPU cluster, it reports a 1.39x geometric mean end to end training speedup over the fastest applicable baseline across 11 configurations. For dynamic video attention, it reports 2.11-2.37x speedups over Ring Attention, including planning.

### Source excerpt

arXiv:2611.23512v1 Announce Type: new Abstract: Sparse attention reduces computation, but distributed execution may still transfer unnecessary data and incur substantial per-operation overhead. Mask-aware assignment and scheduling can reduce these costs, but replanning for dynamic masks must be cheaper than the execution time it saves. We present DexAttention, a distributed sparse attention system that co-designs fast online planning with persistent execution. Its assigner uses GPU-accelerated, topology-aware clustering to balance computational load and reduce communication. A lightweight greedy scheduler orders coarse communication and computation operations, while a persistent kernel on each GPU enforces fine-grained dependencies and dispatches ready tasks. This division keeps detailed synchronization out of planning while enabling overlap without repeated attention launches. The framework supports static and dynamic per-head masks across multiple attention compute kernels. On a four-node, 16-GPU cluster, DexAttention achieves a $1.39\times$ geometric-mean end-to-end training speedup over the fastest applicable baseline across 11 configurations. For dynamic video attention, it achieves $2.11$--$2.37\times$ speedups over Ring Attention including planning.

## Perforce TotalView adds Parallel Python Debugging

DevFeed: [Perforce TotalView adds Parallel Python Debugging](<https://devfeed.tech/articles/perforce-totalview-adds-parallel-python-debugging-141357.md>)

Original publisher: [Read original article](<https://sdtimes.com/debugging/perforce-totalview-adds-parallel-python-debugging/>)

Author: SD Times

Published: 2026-11-18T17:55:43Z

Content type: release

Language: en

Sources: [SD Times](<https://devfeed.tech/sources/sd-times.md>)

Topics: [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [Python](<https://devfeed.tech/topics/python.md>), [debug tools](<https://devfeed.tech/topics/debug-tools.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-ml](<https://devfeed.tech/tags/ai-ml.md>), [compute](<https://devfeed.tech/tags/compute.md>), [compute-resources](<https://devfeed.tech/tags/compute-resources.md>), [computing](<https://devfeed.tech/tags/computing.md>), [control](<https://devfeed.tech/tags/control.md>), [debug](<https://devfeed.tech/tags/debug.md>), [debugging](<https://devfeed.tech/tags/debugging.md>), [developers](<https://devfeed.tech/tags/developers.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [hpc](<https://devfeed.tech/tags/hpc.md>), [latest-news](<https://devfeed.tech/tags/latest-news.md>), [ml](<https://devfeed.tech/tags/ml.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [perforce](<https://devfeed.tech/tags/perforce.md>), [processes](<https://devfeed.tech/tags/processes.md>), [python](<https://devfeed.tech/tags/python.md>), [python-debugging](<https://devfeed.tech/tags/python-debugging.md>), [resources](<https://devfeed.tech/tags/resources.md>), [scientific-computing](<https://devfeed.tech/tags/scientific-computing.md>), [software](<https://devfeed.tech/tags/software.md>)

### AI overview

Perforce TotalView's new Parallel Python Debugging capability lets developers inspect every Python process in a distributed workload and control the processes together in one debugging session. It targets HPC, scientific computing, and AI/ML workloads, including mixed Python and C++ systems, where failures across nodes can consume costly compute time. Teams already using TotalView for MPI, OpenMP, C++, or Fortran can extend those workflows to Python without adding another specialized debugger. A planned demonstration uses a machine-learning simulation on four MPI ranks with fp16: without loss scaling, tiny gradients can underflow to zero, preventing affected weights from updating and causing training to stall or finish at lower accuracy without an error. The supplied text does not detail supported environments or setup requirements.

### Source excerpt

Perforce Software announced the general availability of Parallel Python Debugging in Perforce TotalView, with new capability that gives developers the ability to view and control multiple Python processes simultaneously as they debug distributed workloads without sacrificing high-cost compute resources. Python powers many HPC, AI/ML, and scientific computing workloads, with the GitHub Octoverse 2025 report showing... ... continue reading The post Perforce TotalView adds Parallel Python Debugging appeared first on SD Times.

## Agentic AI puts the CPU back at center stage as Dell and AMD chase every watt

DevFeed: [Agentic AI puts the CPU back at center stage as Dell and AMD chase every watt](<https://devfeed.tech/articles/agentic-ai-puts-the-cpu-back-at-center-stage-as-dell-and-amd-chase-every-watt-140918.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/11/18/business-development-agentic-ai-era-puts-cpus-first-dellatsc26/>)

Author: Jonathan Anthony

Published: 2026-11-18T17:06:36Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [cpu](<https://devfeed.tech/topics/cpu.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [dell](<https://devfeed.tech/topics/dell.md>), [clean energy data centers](<https://devfeed.tech/topics/clean-energy-data-centers.md>)

Tags: [accelerated-computing](<https://devfeed.tech/tags/accelerated-computing.md>), [accelerators](<https://devfeed.tech/tags/accelerators.md>), [advanced-micro-devices](<https://devfeed.tech/tags/advanced-micro-devices.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agentic-workloads](<https://devfeed.tech/tags/agentic-workloads.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-data-centers](<https://devfeed.tech/tags/ai-data-centers.md>), [ai-inference](<https://devfeed.tech/tags/ai-inference.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [ai-supercomputers](<https://devfeed.tech/tags/ai-supercomputers.md>), [amd](<https://devfeed.tech/tags/amd.md>), [amd-epyc-9006-processor](<https://devfeed.tech/tags/amd-epyc-9006-processor.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [business](<https://devfeed.tech/tags/business.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cube-event-coverage](<https://devfeed.tech/tags/cube-event-coverage.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [david-schmidt](<https://devfeed.tech/tags/david-schmidt.md>), [dell](<https://devfeed.tech/tags/dell.md>), [dell-poweredge-r9825](<https://devfeed.tech/tags/dell-poweredge-r9825.md>), [dellatsc26](<https://devfeed.tech/tags/dellatsc26.md>), [demand](<https://devfeed.tech/tags/demand.md>), [development](<https://devfeed.tech/tags/development.md>), [energy-efficiency](<https://devfeed.tech/tags/energy-efficiency.md>), [epyc](<https://devfeed.tech/tags/epyc.md>), [gemma-allen](<https://devfeed.tech/tags/gemma-allen.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [graphics](<https://devfeed.tech/tags/graphics.md>), [graphics-processing-units](<https://devfeed.tech/tags/graphics-processing-units.md>), [hpc-infrastructure](<https://devfeed.tech/tags/hpc-infrastructure.md>), [john-furrier](<https://devfeed.tech/tags/john-furrier.md>), [nasdaq-amd](<https://devfeed.tech/tags/nasdaq-amd.md>), [news](<https://devfeed.tech/tags/news.md>), [nyse-dell](<https://devfeed.tech/tags/nyse-dell.md>), [power](<https://devfeed.tech/tags/power.md>), [poweredge](<https://devfeed.tech/tags/poweredge.md>), [r9825](<https://devfeed.tech/tags/r9825.md>), [sc26](<https://devfeed.tech/tags/sc26.md>), [sc26-2026](<https://devfeed.tech/tags/sc26-2026.md>), [sc26eventpage](<https://devfeed.tech/tags/sc26eventpage.md>), [scott-tease](<https://devfeed.tech/tags/scott-tease.md>), [server](<https://devfeed.tech/tags/server.md>), [supercomputing](<https://devfeed.tech/tags/supercomputing.md>), [sustainable-computing](<https://devfeed.tech/tags/sustainable-computing.md>), [thecube](<https://devfeed.tech/tags/thecube.md>)

### AI overview

The article reports that agentic AI could increase demand for CPUs alongside GPUs: CPUs feed accelerator data and support the speed and concurrency agent workloads require. Dell and AMD say data center power limits make efficiency central. Dell's PowerEdge R9825, a 3U server using AMD EPYC 9006 processors, is designed to improve power and cooling efficiency; Dell says one system can replace 13 legacy machines while reducing cooling power by 80%. The companies see reclaimed power as a way to support more compute, though the excerpt offers no independent verification of these figures.

### Source excerpt

Agentic AI is rewriting the business development playbook for chipmakers and server builders. Graphics processing units still grab the headlines, but as supercomputing and AI converge, the next wave of demand is landing on central processing units that must fit inside the power budgets data centers already have. The shift has been building for years. [...] The post Agentic AI puts the CPU back at center stage as Dell and AMD chase every watt appeared first on SiliconANGLE.

## AInfer-PD: Communication-Safe In-Place Prefill-Decode Multiplexing for Distributed MoE Rollouts

DevFeed: [AInfer-PD: Communication-Safe In-Place Prefill-Decode Multiplexing for Distributed MoE Rollouts](<https://devfeed.tech/articles/ainfer-pd-communication-safe-in-place-prefill-decode-multiplexing-for-distributed-moe-rollouts-140376.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.00993>)

Author: Guowei Wang, Chaokun Yang, Zhenxuan Pan, Yipeng Wei, Yuhong Guo, Minghua Zhu, Zechuan Zhang, Shuo Wan, Xiaowei Zhu

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [async](<https://devfeed.tech/topics/async.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [distributed algorithm](<https://devfeed.tech/topics/distributed-algorithm.md>)

Tags: [ablation](<https://devfeed.tech/tags/ablation.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [batches](<https://devfeed.tech/tags/batches.md>), [collective](<https://devfeed.tech/tags/collective.md>), [communication](<https://devfeed.tech/tags/communication.md>), [completion](<https://devfeed.tech/tags/completion.md>), [continuation](<https://devfeed.tech/tags/continuation.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [d](<https://devfeed.tech/tags/d.md>), [decode](<https://devfeed.tech/tags/decode.md>), [deepep](<https://devfeed.tech/tags/deepep.md>), [devices](<https://devfeed.tech/tags/devices.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [engine](<https://devfeed.tech/tags/engine.md>), [moe](<https://devfeed.tech/tags/moe.md>), [multiplexing](<https://devfeed.tech/tags/multiplexing.md>), [pd](<https://devfeed.tech/tags/pd.md>), [prefill](<https://devfeed.tech/tags/prefill.md>), [safe](<https://devfeed.tech/tags/safe.md>)

### AI overview

AInfer-PD lets decode iterations progress during long prefills in distributed MoE rollouts by coordinating collective ordering across ranks at safe communication boundaries and separating prefill and decode communication buffers and protocol state within one DeepEP runtime. This allows prefills to use DeepEP's normal path while decode keeps its low-latency path, with model weights and KV storage shared on the same devices. In reported prefill-intensive workloads, fixed-workload completion time fell 7.1-22.5% versus the same AInfer engine without P/D multiplexing and 24.8-32.9% versus SGLang on one node; on two nodes, reductions were 18.0-35.3% and 18.3-31.8%, respectively. A same-engine ablation found fine-grained boundaries improved completion time by a further 8.6-19.8% over whole-prefill asynchronous enqueue.

### Source excerpt

arXiv:2609.00993v3 Announce Type: replace Abstract: In agentic reinforcement learning (RL), asynchronous multi-turn trajectories repeatedly introduce continuation prefills (P) while other trajectories remain in decode (D). Efficient rollout completion therefore requires D to advance during long P executions. In-place multiplexing shares devices, model weights, and KV state, but phase-specific communication in distributed MoE introduces two constraints: cross-rank collective ordering and ownership of mutable communication state. Whole-prefill coordination can ensure safe ordering but delays D. DeepEP's normal and low-latency paths suit large P and small D batches, respectively, but safely overlapping them requires separating their mutable communication state. We present AInfer-PD, which extends in-place P/D multiplexing to distributed MoE rollouts. AInfer-PD establishes rank-consistent collective ordering at safe communication boundaries, allowing successive D iterations to advance without waiting for an entire prefill execution to be enqueued or completed. It gives P and D separate communication buffers and protocol state within one DeepEP runtime, allowing P to use the normal path while D retains the low-latency path. Together, these mechanisms preserve useful P/D overlap while sharing model weights and KV storage on the same devices. Across repeated single-node prefill-intensive workloads, AInfer-PD reduces fixed-workload rollout completion time by 7.1-22.5% relative to the same AInfer engine with P/D multiplexing disabled and by 24.8-32.9% relative to SGLang. On two nodes, the reductions are 18.0-35.3% and 18.3-31.8%, respectively. In a same-engine ablation, fine-grained boundaries reduce completion time by a further 8.6-19.8% over whole-prefill asynchronous enqueue.

## Communicate the Change: Efficient Expert Parallelism via Activation Compression

DevFeed: [Communicate the Change: Efficient Expert Parallelism via Activation Compression](<https://devfeed.tech/articles/communicate-the-change-efficient-expert-parallelism-via-activation-compression-139983.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22215>)

Author: Fengyuan Liu, Jay Gala, Siva Reddy, Dzmitry Bahdanau

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Mixture of Experts (MoE)](<https://devfeed.tech/topics/mixture-of-experts-moe.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [vllm](<https://devfeed.tech/topics/vllm.md>), [quantization](<https://devfeed.tech/topics/quantization.md>)

Tags: [activation](<https://devfeed.tech/tags/activation.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [bottleneck](<https://devfeed.tech/tags/bottleneck.md>), [collective](<https://devfeed.tech/tags/collective.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [compute](<https://devfeed.tech/tags/compute.md>), [critical](<https://devfeed.tech/tags/critical.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [distributed-inference](<https://devfeed.tech/tags/distributed-inference.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [ethernet](<https://devfeed.tech/tags/ethernet.md>), [expert-parallelism](<https://devfeed.tech/tags/expert-parallelism.md>), [experts](<https://devfeed.tech/tags/experts.md>), [fewer](<https://devfeed.tech/tags/fewer.md>), [hidden](<https://devfeed.tech/tags/hidden.md>), [hidden-states](<https://devfeed.tech/tags/hidden-states.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [parallelism](<https://devfeed.tech/tags/parallelism.md>)

### AI overview

EchoMoE reduces communication overhead in expert-parallel MoE inference by sending a compressed difference between the current hidden state and a periodically refreshed earlier-layer reference, rather than sending each hidden state in full. The abstract reports 58% lower inter-node communication volume and 22.6% higher throughput, with average performance within 3.8 percentage points across 18 benchmarks. On bandwidth-constrained Ethernet, throughput improves by up to 2.38x. The modification is designed to retrofit pretrained MoE models, requiring about 8,500x fewer training tokens than full pretraining; the communicated representations can also be quantized without measurable quality loss. The authors report an implementation for pretrained models and an inference integration for vLLM.

### Source excerpt

arXiv:2611.22215v1 Announce Type: cross Abstract: Scaling has driven the success of large language models (LLM), with sparse mixture-of-experts (MoE) architectures enabling greater parameter capacity without a proportional increase in computation. At large scales, MoE models rely on expert parallelism to distribute experts across devices. Each MoE layer requires two synchronous collective communication steps during inference: dispatching hidden states to the activated experts and combining the resulting expert outputs. As compute throughput has outpaced interconnect bandwidth, these synchronous collectives have become a major bottleneck to MoE scaling and efficiency. We address this challenge with EchoMoE, a lightweight architecture modification that retrofits existing MoE models, requiring approximately 8,500x fewer training tokens than full pretraining. Motivated by the observation that transformer blocks update the residual stream incrementally, we use a hidden state from an earlier layer as a reference and communicate only the compressed difference between it and the current state, rather than communicating each hidden state in full. EchoMoE periodically refreshes these reference states and transmits only the compressed differences synchronously, reducing the traffic on latency-critical paths without significant computational overhead. EchoMoE reduces inter-node communication volume by 58% and improves throughput by 22.6%, while preserving average performance within 3.8 percentage points across 18 benchmarks. EchoMoE further improves inference throughput by up to 2.38x over bandwidth-constrained Ethernet, paving the way for wide-area distributed inference. EchoMoE also tolerates quantization of the communicated representations without measurable quality loss. We release an efficient implementation for applying EchoMoE to pretrained MoE models, together with an inference integration for vLLM.

## Angelfish: Leader, DAG, or Anywhere in Between

DevFeed: [Angelfish: Leader, DAG, or Anywhere in Between](<https://devfeed.tech/articles/angelfish-leader-dag-or-anywhere-in-between-140375.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2509.15847>)

Author: Qianyu Yu, Giuliano Losa, Nibesh Shrestha, Xuechao Wang

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>), [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [blockchain consensus protocol](<https://devfeed.tech/topics/blockchain-consensus-protocol.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [performance-engineering](<https://devfeed.tech/topics/performance-engineering.md>), [protocols](<https://devfeed.tech/topics/protocols.md>)

Tags: [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [bft-consensus](<https://devfeed.tech/tags/bft-consensus.md>), [blockchain](<https://devfeed.tech/tags/blockchain.md>), [broadcast](<https://devfeed.tech/tags/broadcast.md>), [broadcasting](<https://devfeed.tech/tags/broadcasting.md>), [communication](<https://devfeed.tech/tags/communication.md>), [consensus](<https://devfeed.tech/tags/consensus.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [dag](<https://devfeed.tech/tags/dag.md>), [data](<https://devfeed.tech/tags/data.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [latency](<https://devfeed.tech/tags/latency.md>), [leader](<https://devfeed.tech/tags/leader.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [load](<https://devfeed.tech/tags/load.md>), [nodes](<https://devfeed.tech/tags/nodes.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [performance](<https://devfeed.tech/tags/performance.md>), [protocol](<https://devfeed.tech/tags/protocol.md>), [protocols](<https://devfeed.tech/tags/protocols.md>)

### AI overview

Angelfish combines leader-based and DAG-based BFT consensus, adapting between them by dynamically selecting parties to send lightweight votes via best-effort broadcast instead of reliably broadcasting costlier DAG vertices. The authors report that this reduces communication, helps lagging nodes catch up, and lowers latency compared with prior DAG-based protocols. Their evaluation reports state-of-the-art peak throughput and significantly lower latency under moderate load; the supplied excerpt gives no benchmark figures or further evaluation details.

### Source excerpt

arXiv:2509.15847v4 Announce Type: replace Abstract: To maximize performance, many modern blockchain systems rely on eventually synchronous, Byzantine fault-tolerant (BFT) consensus protocols. Two protocol designs have emerged in this space: protocols that minimize latency using a leader that drives both data dissemination and consensus, and protocols that maximize throughput using a separate, asynchronous data dissemination layer. Recent protocols such as Partially Synchronous Bullshark and Sailfish combine elements of both approaches by using a DAG to enable parallel data dissemination and a leader that paces DAG formation. This improves latency while achieving state-of-the-art throughput. Yet leader-based protocols still achieve better latency under moderate loads, which are common in practice. We present Angelfish, a hybrid protocol that adapts smoothly across this design space, from leader-based to Sailfish-like DAG-based consensus. Angelfish lets a dynamically adjusted subset of parties use best-effort broadcast to issue lightweight votes instead of reliably broadcasting costlier DAG vertices. This reduces communication, helps lagging nodes catch up, and lowers latency in practice compared to prior DAG-based protocols. Our empirical evaluation shows that Angelfish attains state-of-the-art peak throughput while significantly lowering latency under moderate load, delivering the best of both worlds.

## Collecting Infrastructure-Level Measures for Energy Analysis in Modern Computing Infrastructures

DevFeed: [Collecting Infrastructure-Level Measures for Energy Analysis in Modern Computing Infrastructures](<https://devfeed.tech/articles/collecting-infrastructure-level-measures-for-energy-analysis-in-modern-computing-infrastructures-140383.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22160>)

Author: Isac Pasianotto, Stefano Cozzini, Niccolo Tosato

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Sustainable Computing](<https://devfeed.tech/topics/sustainable-computing.md>), [data centers](<https://devfeed.tech/topics/data-centers.md>), [Data pipelines](<https://devfeed.tech/topics/data-pipelines.md>), [infrastructure monitoring](<https://devfeed.tech/topics/infrastructure-monitoring.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [accounting](<https://devfeed.tech/tags/accounting.md>), [acquisition](<https://devfeed.tech/tags/acquisition.md>), [aggregates](<https://devfeed.tech/tags/aggregates.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [analyze](<https://devfeed.tech/tags/analyze.md>), [approach](<https://devfeed.tech/tags/approach.md>), [case-studies](<https://devfeed.tech/tags/case-studies.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [collecting](<https://devfeed.tech/tags/collecting.md>), [collection](<https://devfeed.tech/tags/collection.md>), [complex](<https://devfeed.tech/tags/complex.md>), [component](<https://devfeed.tech/tags/component.md>), [components](<https://devfeed.tech/tags/components.md>), [compute](<https://devfeed.tech/tags/compute.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [data](<https://devfeed.tech/tags/data.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [energy](<https://devfeed.tech/tags/energy.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [modern](<https://devfeed.tech/tags/modern.md>)

### AI overview

The paper presents a cluster-wide pipeline for energy analysis that synchronizes fine-grained measurements across infrastructure subsystems, addressing dynamics that measurements from individual compute nodes can miss. It reports a dataset sampled at 1 Hz for one month in a production research data center running HPC, Kubernetes services, and distributed storage. Case studies of distributed I/O and large-language-model inference illustrate the value of infrastructure-wide measurements. The dataset and method are intended to support future predictive modeling and energy-aware resource management.

### Source excerpt

arXiv:2611.22160v1 Announce Type: new Abstract: Modern HPC facilities are complex environments in which all components are tightly coupled; consequently, precisely modeling and accounting for energy consumption requires approaches that go beyond instrumenting individual compute nodes. Aiming for a cluster-wide vision, we designed a scalable data collection pipeline that aggregates fine-grained measurements from multiple subsystems-spanning component loads, software and hardware counters, sensors and energy-ensuring consistent sampling frequencies and synchronized data acquisition across the infrastructure. We present a dataset of comprehensive metrics collected at 1 Hz over one month from a production research data center supporting heterogeneous workloads, including HPC, Kubernetes services, and distributed storage. To illustrate what this approach makes possible in practice, we analyze two case studies-distributed I/O and large-language-model inference-showing how cluster-wide measurement reveals system-level dynamics missed by localized instrumentation. The proposed dataset and collection methodology provide a foundation for future work on predictive modeling, energy-aware resource management, and more efficient computing infrastructures.

## Databases Demystified Chapter 5 - Transactions Part 2 | Blog | Fivetran

DevFeed: [Databases Demystified Chapter 5 - Transactions Part 2 | Blog | Fivetran](<https://devfeed.tech/articles/databases-demystified-chapter-5-transactions-part-2-blog-fivetran-140212.md>)

Original publisher: [Read original article](<https://wf.fivetran.com/blog/databases-demystified-transactions-part-2>)

Author: Michael Kaminsky

Published: 2026-11-18T03:38:23Z

Content type: article

Language: en

Sources: [Fivetran Blog](<https://devfeed.tech/sources/fivetran-blog.md>)

Topics: [Transactions](<https://devfeed.tech/topics/transactions.md>), [race-condition](<https://devfeed.tech/topics/race-condition.md>), [acid](<https://devfeed.tech/topics/acid.md>), [Deadlock](<https://devfeed.tech/topics/deadlock.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [concurrency](<https://devfeed.tech/tags/concurrency.md>), [databases](<https://devfeed.tech/tags/databases.md>), [deadlock](<https://devfeed.tech/tags/deadlock.md>), [isolation](<https://devfeed.tech/tags/isolation.md>), [transactions](<https://devfeed.tech/tags/transactions.md>)

### AI overview

Database isolation reduces race conditions in concurrent transactions, but higher isolation usually requires more locks and can slow transactions or create contention and deadlocks. The article outlines dirty, non-repeatable, and phantom reads, and ranks common isolation levels from Serializable to Read Uncommitted. Locks can apply to rows, columns, or tables; locking less data can reduce blocking, while deadlocks can occur when transactions hold locks each other needs. Isolation behavior varies between databases, so the article advises checking documentation and testing parallel workloads, including under production-like contention, and balancing data correctness against performance.

### Source excerpt

Learn about the relationship between isolation and concurrency.

[Next page](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md?cursor=WyIyMDI2LTExLTE4VDAzOjM4OjIzKzAwOjAwIiwgImY4YTVlMmRlLTM0MmQtNGU1OS05OWRiLWRlYTU3MWE4ZWU4ZSJd>)