English

A Dynamic Allocation Scheme for Adaptive Shared-Memory Mapping on Kilo-core RV Clusters for Attention-Based Model Deployment

Hardware Architecture 2025-08-05 v1

Abstract

Attention-based models demand flexible hardware to manage diverse kernels with varying arithmetic intensities and memory access patterns. Large clusters with shared L1 memory, a common architectural pattern, struggle to fully utilize their processing elements (PEs) when scaled up due to reduced throughput in the hierarchical PE-to-L1 intra-cluster interconnect. This paper presents Dynamic Allocation Scheme (DAS), a runtime programmable address remapping hardware unit coupled with a unified memory allocator, designed to minimize data access contention of PEs onto the multi-banked L1. We evaluated DAS on an aggressively scaled-up 1024-PE RISC-V cluster with Non-Uniform Memory Access (NUMA) PE-to-L1 interconnect to demonstrate its potential for improving data locality in large parallel machine learning workloads. For a Vision Transformer (ViT)-L/16 model, each encoder layer executes in 5.67 ms, achieving a 1.94x speedup over the fixed word-level interleaved baseline with 0.81 PE utilization. Implemented in 12nm FinFET technology, DAS incurs <0.1 % area overhead.

Keywords

Cite

@article{arxiv.2508.01180,
  title  = {A Dynamic Allocation Scheme for Adaptive Shared-Memory Mapping on Kilo-core RV Clusters for Attention-Based Model Deployment},
  author = {Bowen Wang and Marco Bertuletti and Yichao Zhang and Victor J. B. Jung and Luca Benini},
  journal= {arXiv preprint arXiv:2508.01180},
  year   = {2025}
}

Comments

8 pages, 9 figures, 36th IEEE International Conference on Application-specific Systems, Architectures and Processors

R2 v1 2026-07-01T04:30:33.752Z