Interconnection Networks and Systems

Derniers papiers

• T-Router: Learning Thalamic Routing for Reasoning with Parameter-Efficient Reinforcement Learning

  Liuxian Ma, Jiale Dai, Jiaqi Li, Lu Mi · 1 octobre 2026

  Parameter-efficient reinforcement learning aims to improve reasoning with a compact trainable interface to a pretrained model. We introduce the Thalamic Router (T-Router), which concentrates adaptation on the reuse of completed computations. A compressed, addressable bank preserves block changes; a …

• How Does mHC Use Its Residual Streams? Selective Routing and Near-Identity Mixing

  Pengxiang Zhao, Xing Li, Xianzhi Yu, Wei Guo, Zhenhua Dong · 7 septembre 2026

  Hyper-Connections and their manifold-constrained variant mHC widen a residual pathway from one stream to n, yet how trained models use this capacity remains unclear: how broadly blocks read and write, how strongly the residual pathway mixes streams, and whether the streams carry distinct representat…

• Belief-Space Perception Routing under Coupled Sensor Faults and Compute Contention

  Sparsh Roy, Vihan Aggarwal, Davin Yin · 4 août 2026

  A robot that has to see and react on a fixed clock runs into two problems at once. Its cameras degrade in rain, mud, fog, and darkness. And the single onboard processor it runs on is shared with planning and control, so the compute left over for perception moves around from second to second. Most sy…

• UBEP: Re-architecting Expert Parallelism Communication Library for Production Superpods

  Yipeng Liu, Chang Liu, Si Shen, Jiaqi Zheng, Mingfan Li, Yuyang Yang, Guanhua Li, Yuquan Zhang, Yimeng Xu, Zhongzhe Hu, Zhiyuan Huang, Qihang Duan, Junsong Wang, Wenkai Ling, Baochuan Yang, Xianzhi Yu, Han Bao, Yijie Chen, Guihai Chen · 8 juillet 2026

  The deployment of Mixture-of-Experts (MoE) models on production high-bandwidth superpods, such as NVIDIA's NVL72/576 and Huawei's CloudMatrix384, introduces critical challenges beyond raw interconnect bandwidth. While these systems provide unified global address spaces and high-bandwidth fabrics, th…

• Diagnosing Overhead in Dispatch Operations: Cross-architecture Observatory

  Bole Ma, Jan Eitzinger, Harald Koestler, Gerhard Wellein · 21 mai 2026

  AlltoAll dispatch is the dominant bottleneck of MoE expert parallelism, and the interconnect community has responded with four families of mitigations: predictive sample placement, adaptive expert relayout, hierarchical collectives, and EP-aware topology. All four rest on two assumptions about the w…

• The Symmetries of Three-Layer ReLU Networks

  Johanna Marie Gegenfurtner, Moritz Grillo, Guido Mont\'ufar · 19 mai 2026

  We develop a framework for analyzing parameter symmetries in deep ReLU networks and obtain a complete characterization of the generic parameter fibers for three-layer bottleneck architectures. Our approach provides explicit semi-algebraic descriptions of these fibers and yields a polynomial time alg…

• PiERN: Token-Level Routing for Integrating High-Precision Computation and Reasoning

  Hengbo Xiao, Jingyuan Fan, Xin Tong, Jingzhao Zhang, Chao Lu, Guannan He · 21 avril 2026

  Tasks on complex systems require high-precision numerical computation to support decisions, but current large language models (LLMs) cannot integrate such computations as an intrinsic and interpretable capability with existing architectures. Multi-agent approaches can leverage external experts, but …

• Improved Upper Bounds for Slicing the Hypercube

  Duncan Soiffer, Nathaniel Itty, Christopher D. Rosin, Blake Bruell, Mason DiCicco, G\'abor N. S\'ark\"ozy, Ryan Offstein, Daniel Reichman · 20 février 2026

  A collection of hyperplanes $\mathcal{H}$ slices all edges of the $n$-dimensional hypercube $Q_n$ with vertex set $\{-1,1\}^n$ if, for every edge $e$ in the hypercube, there exists a hyperplane in $\mathcal{H}$ intersecting $e$ in its interior. Let $S(n)$ be the minimum number of hyperplanes needed …

• Geometric Stratification for Singular Configurations of the P3P Problem via Local Dual Space

  Xueying Sun, Zijia Li, Nan Li · 16 février 2026

  This paper investigates singular configurations of the P3P problem. Using local dual space, a systematic algebraic-computational framework is proposed to give a complete geometric stratification for the P3P singular configurations with respect to the multiplicity $μ$ of the camera center $O$: for $μ…

• From Buffers to Registers: Unlocking Fine-Grained FlashAttention with Hybrid-Bonded 3D NPU Co-Design

  Jinxin Yu, Yudong Pan, Mengdi Wang, Huawei Li, Yinhe Han, Xiaowei Li, Ying Wang · 12 février 2026

  Transformer-based models dominate modern AI workloads but exacerbate memory bottlenecks due to their quadratic attention complexity and ever-growing model sizes. Existing accelerators, such as Groq and Cerebras, mitigate off-chip traffic with large on-chip caches, while algorithmic innovations such …

• FC-KAN: Function Combinations in Kolmogorov-Arnold Networks

  Hoang-Thang Ta, Duy-Quy Thai, Abu Bakar Siddiqur Rahman, Grigori Sidorov, Alexander Gelbukh · 2 février 2026

  In this paper, we introduce FC-KAN, a Kolmogorov-Arnold Network (KAN) that leverages combinations of popular mathematical functions such as B-splines, wavelets, and radial basis functions on low-dimensional data through element-wise operations. We explore several methods for combining the outputs of…

• Deep Reinforcement Learning for Fault-Adaptive Routing in Eisenstein-Jacobi Interconnection Topologies

  Mohammad Walid Charrwi, Zaid Hussain · 30 janvier 2026

  The increasing density of many-core architectures necessitates interconnection networks that are both high-performance and fault-resilient. Eisenstein-Jacobi (EJ) networks, with their symmetric 6-regular topology, offer superior topological properties but challenge traditional routing heuristics und…

• Artificial Intelligence-Driven Network-on-Chip Design Space Exploration: Neural Network Architectures for Design

  Amogh Anshu N, Harish BP · 10 décembre 2025

  Network-on-Chip (NoC) design requires exploring a high-dimensional configuration space to satisfy stringent throughput requirements and latency constraints.Traditional design space exploration techniques are often slow and struggle to handle complex, non-linear parameter interactions.This work prese…

• EquiTabPFN: A Target-Permutation Equivariant Prior Fitted Networks

  Michael Arbel, David Salinas, Frank Hutter · 3 décembre 2025

  Recent foundational models for tabular data, such as TabPFN, excel at adapting to new tasks via in-context learning, but remain constrained to a fixed, pre-defined number of target dimensions-often necessitating costly ensembling strategies. We trace this constraint to a deeper architectural shortco…

• Taming the Tail: NoI Topology Synthesis for Mixed DL Workloads on Chiplet-Based Accelerators

  Arnav Shukla, Harsh Sharma, Srikant Bharadwaj, Vinayak Abrol, Sujay Deb · 29 octobre 2025

  Heterogeneous chiplet-based systems improve scaling by disag-gregating CPUs/GPUs and emerging technologies (HBM/DRAM).However this on-package disaggregation introduces a latency inNetwork-on-Interposer(NoI). We observe that in modern large-modelinference, parameters and activations routinely move ba…