Work

Matching Linear Algebra and Tensor Code to Specialized Hardware Accelerators

Year: 2023

Type: conference-paper

Abstract: Dedicated tensor accelerators demonstrate the importance of linear algebra in modern applications. Such accelerators have the potential for impressive performance gains, but require programmers to rewrite code using vendor APIs - a barrier to wider scale adoption. Recent work overcomes this by matching and replacing patterns within code, but such approaches are fragile and fail to cope with the diversity of real-world codes.

Source: ACM SIGPLAN International Conference on Compiler Construction (CC)

DOI: 10.1145/3578360.3580262

Authors: Pablo Antonio Martínez, Jackson Woodruff, Jordi Armengol-Estapé, Gregorio Bernabé, José Manuel García, Michael F. P. O’Boyle

Institutions: Universidad de Murcia, University of Edinburgh

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: bronze

Funders: European Commission, Agencia Estatal de Investigación