Back to search

Cornell University

Modular Abstractions for Efficient Hardware Design

Abstract

dc:description.abstract

Hardware design is primarily concerned with efficiency: the need to implement the fastest circuit using the least amount of resources and power. Coupled with the staggering amounts of resources poured into designing, manufacturing, and deploying hardware, optimization decisions dominate the design of tools for hardware design. Modularity, or the separation of concerns, allows for design of reusable components and has been a primary driver of the software revolution. However, in hardware design, it has taken a backseat; modular design obfuscates key properties of circuits which may lead to inefficient implementation. In the specialization era, where performance gains are driven by designing hardware for specific computations, the need for modular and efficient abstractions for hardware design is dire. This thesis identifies explicit reasoning about time as a key ingredient for the design of such abstractions and embodies them in three systems. First, Dahlia, an imperative language that compiles to hardware and uses time-sensitive reasoning to ensure that surface programs compile to efficient hardware. Second, Calyx, a compiler and an intermediate language for transforming Dahlia-like languages into hardware descriptions. Calyx bridges the gap between computational descriptions and circuit implementations using a novel intermediate language that mixes software-like control flow and hardware-like structural constructs. Calyx further resolves the tension between precise modeling of cycle-level time and scalable compiler optimizations by exploiting the observation that time-sensitive execution schedules are a refinement of time-insensitive ones. Finally, Filament, a new hardware description language that directly models cycle-level constraints in the interfaces of modules and ensures, at compile-time, that designs do not have any structural hazards. Together, these systems explore the importance of reasoning about time in the context of hardware design and beckon towards a future where modular abstractions do not need to compromise on efficiency.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Computer Science
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Computer Science
Grantor
Cornell University
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Nigam, Rachit
Chair dc:contributor.chair
  • Sampson, Adrian
Committee members dc:contributor.committeemember
  • Foster, John
  • De Sa, Christopher
  • Zhang, Zhiru

Subjects

dc:subject × 4

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
16938436
Dc Identifier Other
ProQuest Submission ID: 14808
ProQuest Publication ID: 31768008

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Nigam, Rachit. Modular Abstractions for Efficient Hardware Design. Doctor of Philosophy thesis, Cornell University, 2025. https://hdl.handle.net/1813/117609