Customized Computing and Machine Learning - A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy in Computer Science

Aug 2, 2026

Upload by Hazel

Customized computing and machine learning addresses the limits of general-purpose scaling and the programmability gap of domain-specific accelerators (DSAs). The dissertation combines FPGA-based customized computing with machine learning to automate microarchitecture optimization. It introduces architecture templates for accelerating CNN and GCN workloads and develops an automated optimization workflow covering design space exploration and performance/area modeling. A bottleneck optimizer and multiple learning models are used to build accurate, robust performance predictors, enabling fast microarchitecture optimization and broader FPGA adoption.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages232

Reading Time9 h 45 min

AccessFree to download

Introduction

Dissertation Overview

Customized Computing for Machine Learning Acceleration

Machine learning for Designing Customized Accelerators

Background

High-Level Synthesis (HLS)

Convolutional Neural Networks

Graph Neural Networks

Related Works

UCLA UCLA Electronic Theses and Dissertations Title Customized Computing and Machine Learning Permalink [https://escholarship.org/uc/item/4kw0h4nz](https://escholarship.org/uc/item/4kw0h4nz) Author Sohrabizadeh, Atefeh Publication Date 2024 Peer reviewed|Thesis/dissertation [eScholarship.org](eScholarship.org) Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA Los Angeles Customized Computing and Machine Learning A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy in Computer Science by Atefeh Sohrabizadeh © Copyright by Atefeh Sohrabizadeh 2024 ABSTRACT OF THE DISSERTATION Customized Computing and Machine Learning by Atefeh Sohrabizadeh Doctor of Philosophy in Computer Science University of California, Los Angeles, 2024 Professor Jingsheng Jason Cong, Chair Nowadays, abundant data across various domains necessitate high-performance computing capabilities. While we used to be able to answer this need by scaling the frequency, the breakdown of Dennard’s scaling has rendered this approach obsolete. On the other hand, Domain-specific Accelerators (DSAs) have gained a growing interest since they can offer high performance while being energy efficient. This stems from several factors, such as, 1) they support utilizing special data types and operations, 2) they offer massive parallelism, 3) one can customize the memory access, 4) customizing the control/data path helps with amortizing the overhead of fixed instructions, and 5) one has the option of co-designing the algorithm with the hardware. Unfortunately, despite the huge speedups that DSAs can deliver compared to generalpurpose processors, their programmability has not caught up. In the past few decades, High-Level Synthesis (HLS) tools were introduced to raise the abstraction level and free designers from delving into architecture details at the circuit level. While HLS can significantly reduce the efforts involved in the hardware architecture design, not every HLS code yields optimal performance, requiring designers to articulate the most suitable microarchitecture for the target application. This can affect the design turnaround times as there are more choices to explore at a higher level. Moreover, this limitation has confined the DSA community primarily to hardware designers, impeding widespread adoption. This dissertation endeavors to alleviate this problem by combining customized computing and machine learn- ing. Consequently, this dissertation consists of two core parts: 1) customized computing tailored for machine learning applications, and 2) machine learning employed to automate the optimization process of customized computing. Our focus will be on FPGAs as their cost-effective nature and rapid prototyping capabilities make them especially suitable for our research. The large amounts of data available in data centers have motivated researchers to develop machine learning algorithms for processing them. Given that a significant portion of data stored in these centers exists in the form of images or graphs, our attention is directed towards two prominent algorithms designed for such tasks: Convolutional Neural Network (CNN) and Graph Convolutional Network (GCN) . In the first part of the dissertation, we develop architecture templates for accelerating these applications. This approach facilitatesa reduction in the development cycle, allowing the instantiation of module templates with customizable parameters based on the specific target application. In the second part of the dissertation, we move our focus to general applications and work on automating their optimization steps including design space exploration and performance/area modeling. Therefore, we structure our problem in a way that can be fed into the learning algorithms. We develop a highly efficient bottleneck optimizer to explore the search space. We also e

FAQs

Why does the dissertation focus on domain-specific accelerators (DSAs) and FPGAs?

DSAs provide high performance with energy efficiency, but their programmability lags behind. The dissertation targets FPGAs because they enable cost-effective implementation and rapid prototyping for customized designs.

What are the two core parts of the dissertation?

The first part builds customized computing for machine learning acceleration. The second part uses machine learning to automate the optimization process for customized computing.

How does the dissertation accelerate CNN and GCN workloads?

It develops architecture templates for these applications, allowing module instantiation with customizable parameters tailored to specific target requirements.

Aug 2, 2026

Upload by Hazel