{"id":{"repo_id":"cornell","oai_identifier":"oai:ecommons.cornell.edu:1813/120653"},"canonical_url":"https://search.ndltd.org/etd/cornell/oai:ecommons.cornell.edu:1813/120653","repository":{"repo_id":"cornell","name":"Cornell University","base_url":"https://ecommons.cornell.edu/server/oai/request"},"display":{"title":"Designing Machine Learning Accelerators via High-Level Synthesis Through Calyx","abstract":"As applications grow in complexity and performance demand, general-purpose processors are no longer optimal for many domains due to their limited efficiency and high energy overhead in computation-intensive tasks. This is particularly evident in the field of machine learning (ML), where model sizes and usage have grown exponentially, making CPUs ill-suited for inference and training workloads. Although custom hardware accelerators offer significant performance and energy efficiency gains, designing them at the register-transfer level (RTL) is time-consuming and error-prone. Furthermore, the vast majority of ML programs are written in Python, a high-level language that is far removed from RTL design. To bridge this semantic gap, we present a complete open-source compiler toolchain that translates ML models written in Python into synthesizable SystemVerilog, targeting FPGAs as the hardware backend. Our toolchain leverages Calyx, a structured intermediate representation designed for hardware accelerators, and is integrated into the CIRCT project for extensibility and analysis. In addition to building the end-to-end flow, we design and implement a set of compiler passes for memory partitioning, enabling effective parallelism in memory-intensive ML workloads. Experimental results demonstrate that our compiler can effectively generate hardware from high-level ML models and achieve performance gains through static memory banking optimizations.","abstract_html":"As applications grow in complexity and performance demand, general-purpose processors are no longer optimal for many domains due to their limited efficiency and high energy overhead in computation-intensive tasks. This is particularly evident in the field of machine learning (ML), where model sizes and usage have grown exponentially, making CPUs ill-suited for inference and training workloads. Although custom hardware accelerators offer significant performance and energy efficiency gains, designing them at the register-transfer level (RTL) is time-consuming and error-prone. Furthermore, the vast majority of ML programs are written in Python, a high-level language that is far removed from RTL design. To bridge this semantic gap, we present a complete open-source compiler toolchain that translates ML models written in Python into synthesizable SystemVerilog, targeting FPGAs as the hardware backend. Our toolchain leverages Calyx, a structured intermediate representation designed for hardware accelerators, and is integrated into the CIRCT project for extensibility and analysis. In addition to building the end-to-end flow, we design and implement a set of compiler passes for memory partitioning, enabling effective parallelism in memory-intensive ML workloads. Experimental results demonstrate that our compiler can effectively generate hardware from high-level ML models and achieve performance gains through static memory banking optimizations.","abstract_has_math":false,"creators":["Xie, Jiahan"],"institution":"Cornell University","degree_name":"M.S., Computer Science","degree_level":"Master of Science","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":["Sampson, Adrian"],"committee_members":["Zhang, Zhiru"],"year":2025,"date_issued":"2025-08","date_published":"2025-08","updated_at":"2026-11-19T03:12:35Z","subjects":[],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.7298/4mfp-yy52"],"render_values":[{"text":"https://doi.org/10.7298/4mfp-yy52","href":"https://doi.org/10.7298/4mfp-yy52","code":true}]},{"key":"dc:identifier.bibid","label":"Dc Identifier Bibid","values":["17422417"],"render_values":[{"text":"17422417","href":null,"code":true}]},{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["ProQuest Submission ID: 12584","ProQuest Publication ID: 32171305"],"render_values":[{"text":"ProQuest Submission ID: 12584","href":null,"code":true},{"text":"ProQuest Publication ID: 32171305","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/1813/120653","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/1813/120653","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":"embargoed","embargo_until":"2026-09-09","access_note":null},"source_record":{"url":"https://ecommons.cornell.edu/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Aecommons.cornell.edu%3A1813%2F120653","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.chair","label":"Chair","values":["Sampson, Adrian"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Zhang, Zhiru"]},{"key":"dc:creator","label":"Author","values":["Xie, Jiahan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-04-02T18:33:35Z"]},{"key":"dc:date.embargo","label":"Dc Date Embargo","values":["2026-09-09"]},{"key":"dc:date.issued","label":"Date","values":["2025-08"]},{"key":"dc:type","label":"Dc Type","values":["dissertation or thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Master of Science"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S., Computer Science"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Cornell University"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.bibid","label":"Dc Identifier Bibid","values":["17422417"]},{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.7298/4mfp-yy52"]},{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["ProQuest Submission ID: 12584","ProQuest Publication ID: 32171305"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/1813/120653"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"cerif:openaire","label":"Cerif Openaire","values":["<Publication xmlns=\"https://www.openaire.eu/cerif-profile/1.1/\" id=\"b82574aa-955e-4677-8ffb-db0845c5e8df\"> <Type xmlns=\"https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types\">http://purl.org/coar/resource_type/c_1843</Type> <Language>en</Language> <Title>Designing Machine Learning Accelerators via High-Level Synthesis Through Calyx</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2025-08</PublicationDate> <DOI>https://doi.org/10.7298/4mfp-yy52</DOI> <Authors> <Author> <DisplayName>Xie, Jiahan</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <OrgUnit /> </Publisher> </Publishers> <Abstract>As applications grow in complexity and performance demand, general-purpose processors are no longer optimal for many domains due to their limited efficiency and high energy overhead in computation-intensive tasks. This is particularly evident in the field of machine learning (ML), where model sizes and usage have grown exponentially, making CPUs ill-suited for inference and training workloads. Although custom hardware accelerators offer significant performance and energy efficiency gains, designing them at the register-transfer level (RTL) is time-consuming and error-prone. Furthermore, the vast majority of ML programs are written in Python, a high-level language that is far removed from RTL design. To bridge this semantic gap, we present a complete open-source compiler toolchain that translates ML models written in Python into synthesizable SystemVerilog, targeting FPGAs as the hardware backend. Our toolchain leverages Calyx, a structured intermediate representation designed for hardware accelerators, and is integrated into the CIRCT project for extensibility and analysis. In addition to building the end-to-end flow, we design and implement a set of compiler passes for memory partitioning, enabling effective parallelism in memory-intensive ML workloads. Experimental results demonstrate that our compiler can effectively generate hardware from high-level ML models and achieve performance gains through static memory banking optimizations.</Abstract> <Access xmlns=\"http://purl.org/coar/access_right\" > </Access> </Publication>"]},{"key":"cris:virtual.author","label":"Cris Virtual Author","values":["Xie, Jiahan"]},{"key":"cris:virtualsource.collection","label":"Cris Virtualsource Collection","values":["5893a6ea-7af3-41d7-abc6-04bcd26ab5df"]},{"key":"dc:description","label":"Description","values":["31 pages"]},{"key":"dc:description.abstract","label":"Abstract","values":["As applications grow in complexity and performance demand, general-purpose processors are no longer optimal for many domains due to their limited efficiency and high energy overhead in computation-intensive tasks. This is particularly evident in the field of machine learning (ML), where model sizes and usage have grown exponentially, making CPUs ill-suited for inference and training workloads. Although custom hardware accelerators offer significant performance and energy efficiency gains, designing them at the register-transfer level (RTL) is time-consuming and error-prone. Furthermore, the vast majority of ML programs are written in Python, a high-level language that is far removed from RTL design. To bridge this semantic gap, we present a complete open-source compiler toolchain that translates ML models written in Python into synthesizable SystemVerilog, targeting FPGAs as the hardware backend. Our toolchain leverages Calyx, a structured intermediate representation designed for hardware accelerators, and is integrated into the CIRCT project for extensibility and analysis. In addition to building the end-to-end flow, we design and implement a set of compiler passes for memory partitioning, enabling effective parallelism in memory-intensive ML workloads. Experimental results demonstrate that our compiler can effectively generate hardware from high-level ML models and achieve performance gains through static memory banking optimizations."]},{"key":"dc:description.collection","label":"Dc Description Collection","values":["Cornell Theses and Dissertations"]},{"key":"dc:format.mimetype","label":"Dc Format Mimetype","values":["application/pdf"]},{"key":"dcterms:license","label":"Dcterms License","values":["https://hdl.handle.net/1813/59810.2"]},{"key":"dc:title","label":"Title","values":["Designing Machine Learning Accelerators via High-Level Synthesis Through Calyx"]},{"key":"dspace:entity.type","label":"Dspace Entity Type","values":["Publication"]},{"key":"others:access_status","label":"Others Access Status","values":["embargo"]}]}],"canonical_facts":{"cerif:openaire":["<Publication xmlns=\"https://www.openaire.eu/cerif-profile/1.1/\" id=\"b82574aa-955e-4677-8ffb-db0845c5e8df\"> <Type xmlns=\"https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types\">http://purl.org/coar/resource_type/c_1843</Type> <Language>en</Language> <Title>Designing Machine Learning Accelerators via High-Level Synthesis Through Calyx</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2025-08</PublicationDate> <DOI>https://doi.org/10.7298/4mfp-yy52</DOI> <Authors> <Author> <DisplayName>Xie, Jiahan</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <OrgUnit /> </Publisher> </Publishers> <Abstract>As applications grow in complexity and performance demand, general-purpose processors are no longer optimal for many domains due to their limited efficiency and high energy overhead in computation-intensive tasks. This is particularly evident in the field of machine learning (ML), where model sizes and usage have grown exponentially, making CPUs ill-suited for inference and training workloads. Although custom hardware accelerators offer significant performance and energy efficiency gains, designing them at the register-transfer level (RTL) is time-consuming and error-prone. Furthermore, the vast majority of ML programs are written in Python, a high-level language that is far removed from RTL design. To bridge this semantic gap, we present a complete open-source compiler toolchain that translates ML models written in Python into synthesizable SystemVerilog, targeting FPGAs as the hardware backend. Our toolchain leverages Calyx, a structured intermediate representation designed for hardware accelerators, and is integrated into the CIRCT project for extensibility and analysis. In addition to building the end-to-end flow, we design and implement a set of compiler passes for memory partitioning, enabling effective parallelism in memory-intensive ML workloads. Experimental results demonstrate that our compiler can effectively generate hardware from high-level ML models and achieve performance gains through static memory banking optimizations.</Abstract> <Access xmlns=\"http://purl.org/coar/access_right\" > </Access> </Publication>"],"cris:virtual.author":["Xie, Jiahan"],"cris:virtualsource.collection":["5893a6ea-7af3-41d7-abc6-04bcd26ab5df"],"dc:contributor.chair":["Sampson, Adrian"],"dc:contributor.committeemember":["Zhang, Zhiru"],"dc:creator":["Xie, Jiahan"],"dc:date.accessioned":["2026-04-02T18:33:35Z"],"dc:date.embargo":["2026-09-09"],"dc:date.issued":["2025-08"],"dc:description":["31 pages"],"dc:description.abstract":["As applications grow in complexity and performance demand, general-purpose processors are no longer optimal for many domains due to their limited efficiency and high energy overhead in computation-intensive tasks. This is particularly evident in the field of machine learning (ML), where model sizes and usage have grown exponentially, making CPUs ill-suited for inference and training workloads. Although custom hardware accelerators offer significant performance and energy efficiency gains, designing them at the register-transfer level (RTL) is time-consuming and error-prone. Furthermore, the vast majority of ML programs are written in Python, a high-level language that is far removed from RTL design. To bridge this semantic gap, we present a complete open-source compiler toolchain that translates ML models written in Python into synthesizable SystemVerilog, targeting FPGAs as the hardware backend. Our toolchain leverages Calyx, a structured intermediate representation designed for hardware accelerators, and is integrated into the CIRCT project for extensibility and analysis. In addition to building the end-to-end flow, we design and implement a set of compiler passes for memory partitioning, enabling effective parallelism in memory-intensive ML workloads. Experimental results demonstrate that our compiler can effectively generate hardware from high-level ML models and achieve performance gains through static memory banking optimizations."],"dc:description.collection":["Cornell Theses and Dissertations"],"dc:format.mimetype":["application/pdf"],"dc:identifier.bibid":["17422417"],"dc:identifier.doi":["https://doi.org/10.7298/4mfp-yy52"],"dc:identifier.other":["ProQuest Submission ID: 12584","ProQuest Publication ID: 32171305"],"dc:identifier.uri":["https://hdl.handle.net/1813/120653"],"dc:language.iso":["en"],"dcterms:license":["https://hdl.handle.net/1813/59810.2"],"dc:title":["Designing Machine Learning Accelerators via High-Level Synthesis Through Calyx"],"dc:type":["dissertation or thesis"],"dspace:entity.type":["Publication"],"others:access_status":["embargo"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Master of Science"],"thesis:degree_name":["M.S., Computer Science"],"thesis:institution_name":["Cornell University"]},"updated_at":"2026-11-19T03:12:35Z"}