Numerical Kernels on a Spatial Accelerator: A Study of Tenstorrent Wormhole

2026-03-24

AI Summary

This work presents the first systematic evaluation of AI accelerators for traditional scientific computing, with a focus on efficiently executing sparse numerical algorithms on the Tenstorrent Wormhole spatial architecture. By implementing three core sparse numerical kernels and constructing a conjugate gradient solver, the study proposes optimization strategies tailored to spatial architectures for sparse computations. Experimental results demonstrate that the optimized implementations achieve performance comparable to or exceeding that of NVIDIA GPUs, thereby validating the potential of AI accelerators in scientific computing and significantly broadening their applicability beyond conventional machine learning workloads.

Technology Category

• Machine Learning: Hardware-aware ML
• Humans and AI: Other Foundations of Human Computation & AI
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

As AI accelerators gain prominence, their potential for traditional scientific computing workloads remains unclear. This paper explores Tenstorrent's Wormhole architecture, a spatial computing platform designed for neural network acceleration, by implementing three numerical kernels and composing them into a conjugate gradient solver. We present architecture-specific optimizations for sparse numerical algorithms, evaluate their performance against Nvidia GPUs, and expose both challenges and opportunities in porting numerical methods to spatial architectures. Our results demonstrate that AI accelerators merit consideration for workloads traditionally dominated by CPUs and GPUs, and more work should be invested in understanding the capabilities of these architectures and making them accessible to the scientific computing community.

Problem

• AI accelerators
• scientific computing
• numerical kernels
• spatial architectures
• sparse numerical algorithms

Innovation

• spatial accelerator
• numerical kernels
• conjugate gradient
• sparse algorithms
• AI accelerators