O'Boyle, Michael

Ainsworth, Sam

Woodruff, Jackson

2024-08-06T11:33:35Z

2024-08-06T11:33:35Z

2024-08-06

Although specialized hardware promises orders of magnitude performance gains, their uptake has been limited by how challenging it is to program them. Hardware accelerators present challenges programmers are not used to, exposing details of the hardware that are often hidden and requiring new programming styles to use them effectively. Existing programming models often involve learning complex and hardware-specific APIs, using Domain Specific Languages (DSLs), or programming in customized assembly languages. These programming models for hardware accelerators present a significant challenge to uptake: a steep, unforgiving, and untransferable learning curve. However, programming hardware accelerators using traditional programming models presents a challenge: mapping code not written with hardware accelerators in mind to accelerators with restricted behaviour. This thesis presents these challenges in the context of the acceleration equation, and it presents solutions to it in three different contexts: for regular expression accelerators, for API-programmable accelerators (with Fourier Transforms as a key case-study) and for heterogeneous coarse-grained reconfigurable arrays (CGRAs). This thesis shows that automatically morphing software written in traditional manners to fit hardware accelerators is possible with no programmer effort and that huge potential speedups are available.

https://hdl.handle.net/1842/42059

http://dx.doi.org/10.7488/era/4781

en

The University of Edinburgh

Jackson Woodruff, Michael F.P. O’Boyle, New Regular Expressions on Old Accelerators. 58th Design Automation Conference, 2021.

Jackson Woodruff, Jordi Armengol-Estape, Sam Ainsworth, Michael F.P. O’Boyle, ´ Bind the Gap: Compiling Real Software to Hardware FFT Accelerators. PLDI 2022

Jackson Woodruff, Thomas Kœhler, Alexander Brauckmann, Chris Cummins, Sam Ainsworth, Michael F.P. O’Boyle, Rewriting History: Repurposing Domain-Specific CGRAs. CoRR, 2023. Available at https://arxiv.org/pdf/2309. 09112.pdf.

Jordi Armengol-Estape, Jackson Woodruff, Alexander Bruackmann, Jos ´ e Wesley ´ de Souza Magalhaes, and Michael F.P. O’Boyle. Exebench: An ML-scale dataset ˜ of executable C functions. International Symposium on Machine Programming, 2022

Bruce Collie, Philip Ginsbach, Jackson Woodruff, Ajitha Rajan, and Michael F.P. O’Boyle. M3: Semantic API migration. ASE, 2020

Bruce Collie, Jackson Woodruff, and Michael FP O’Boyle. Modeling black-box components with probabilitic synthesis. GPCE, 2020.

Pablo Antonio Martinez, Jackson Woodruff, Jordi Armengol-Estape, Gregorio ´ Bernabe, Jos ´ e Manuel Garc ´ ´ıa, and Michael O’Boyle. Matching linear algebra and tensor code to specialized hardware accelerators. CC, 2023

Jackson Woodruff, Sam Ainsworth, and Michael F.P. O’Boyle. Secco: Codesign for resource sharing in regular expression accelerators. ASP-DAC, 2024.

Jackson Woodruff and Michael F P O’Boyle. New regular expressions on old accelerators. DAC 2021, 2021

programming hardware accelerators

acceleration equation

API-programmable accelerators

Fourier Transforms

CGRAs

Acceleration for the many, not the few

Thesis or Dissertation

Doctoral

PhD Doctor of Philosophy