Accel-GCN: High-Performance GPU Accelerator Design for Graph Convolution Networks
Graph Convolutional Networks (GCNs) are pivotal in extracting latent information from graph data across various domains, yet their acceleration on mainstream GPUs is challenged by workload imbalance and memory access irregularity. To address these challenges, we present Accel-GCN, a GPU accelerator...
Uloženo v:
| Vydáno v: | Digest of technical papers - IEEE/ACM International Conference on Computer-Aided Design s. 01 - 09 |
|---|---|
| Hlavní autoři: | , , , , , , , , , |
| Médium: | Konferenční příspěvek |
| Jazyk: | angličtina |
| Vydáno: |
IEEE
28.10.2023
|
| Témata: | |
| ISSN: | 1558-2434 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Abstract | Graph Convolutional Networks (GCNs) are pivotal in extracting latent information from graph data across various domains, yet their acceleration on mainstream GPUs is challenged by workload imbalance and memory access irregularity. To address these challenges, we present Accel-GCN, a GPU accelerator architecture for GCNs. The design of Accel-GCN encompasses: (i) a lightweight degree sorting stage to group nodes with similar degree; (ii) a block-level partition strategy that dynamically adjusts warp workload sizes, enhancing shared memory locality and workload balance, and reducing metadata overhead compared to designs like GNNAdvisor; (iii) a combined warp strategy that improves memory coalescing and computational parallelism in the column dimension of dense matrices. Utilizing these principles, we formulate a kernel for SpMM in GCNs that employs block-level partitioning and combined warp strategy. This approach augments performance and multi-level memory efficiency and optimizes memory bandwidth by exploiting memory coalescing and alignment. Evaluation of Accel-GCN across 18 benchmark graphs reveals that it outperforms cuSPARSE, GNNAdvisor, and graph-BLAST by factors of 1.17×, 1.86×, and 2.94× respectively. The results underscore Accel-GCN as an effective solution for enhancing GCN computational efficiency. The implementation can be found on Github*. |
|---|---|
| AbstractList | Graph Convolutional Networks (GCNs) are pivotal in extracting latent information from graph data across various domains, yet their acceleration on mainstream GPUs is challenged by workload imbalance and memory access irregularity. To address these challenges, we present Accel-GCN, a GPU accelerator architecture for GCNs. The design of Accel-GCN encompasses: (i) a lightweight degree sorting stage to group nodes with similar degree; (ii) a block-level partition strategy that dynamically adjusts warp workload sizes, enhancing shared memory locality and workload balance, and reducing metadata overhead compared to designs like GNNAdvisor; (iii) a combined warp strategy that improves memory coalescing and computational parallelism in the column dimension of dense matrices. Utilizing these principles, we formulate a kernel for SpMM in GCNs that employs block-level partitioning and combined warp strategy. This approach augments performance and multi-level memory efficiency and optimizes memory bandwidth by exploiting memory coalescing and alignment. Evaluation of Accel-GCN across 18 benchmark graphs reveals that it outperforms cuSPARSE, GNNAdvisor, and graph-BLAST by factors of 1.17×, 1.86×, and 2.94× respectively. The results underscore Accel-GCN as an effective solution for enhancing GCN computational efficiency. The implementation can be found on Github*. |
| Author | Zhang, Wei Fang, Haowen Khan, Omer Hasan, Amit Zhao, Jiahui Huang, Shaoyi Ding, Caiwen Xie, Xi Peng, Hongwu Geng, Tong |
| Author_xml | – sequence: 1 givenname: Xi surname: Xie fullname: Xie, Xi email: xi.xie@uconn.edu organization: University of Connecticut – sequence: 2 givenname: Hongwu surname: Peng fullname: Peng, Hongwu email: hongwu.peng@uconn.edu organization: University of Connecticut – sequence: 3 givenname: Amit surname: Hasan fullname: Hasan, Amit email: amit.hasan@uconn.edu organization: University of Connecticut – sequence: 4 givenname: Shaoyi surname: Huang fullname: Huang, Shaoyi email: shaoyi.huang@uconn.edu organization: University of Connecticut – sequence: 5 givenname: Jiahui surname: Zhao fullname: Zhao, Jiahui email: jiahui.zhao@uconn.edu organization: University of Connecticut – sequence: 6 givenname: Haowen surname: Fang fullname: Fang, Haowen email: haowfang@gmail.com organization: University of Connecticut – sequence: 7 givenname: Wei surname: Zhang fullname: Zhang, Wei email: wei.13.zhang@uconn.edu organization: University of Connecticut – sequence: 8 givenname: Tong surname: Geng fullname: Geng, Tong email: tgeng@ur.rochester.edu organization: University of Rochester – sequence: 9 givenname: Omer surname: Khan fullname: Khan, Omer email: khan@uconn.edu organization: University of Connecticut – sequence: 10 givenname: Caiwen surname: Ding fullname: Ding, Caiwen email: caiwen.ding@uconn.edu organization: University of Connecticut |
| BookMark | eNo1UNtOwkAUXI0mAvIHPuwPFM_e2q5vTdFCQhATeSbb7Smsli7ZVo1_b709TSZzyWTG5KL1LRJCGcwYA327zPNsrhKhYcaBixkDwUXC-RmZ6kSnQg2cc6XOyYgplUZcCnlFxl33AjAE0nhEnjJrsYmKfH1HF25_iDYYah-OprVIi82W_ugYTO8DnWPn9i0ddFoEczrQ3LfvvnnrnW_pGvsPH167a3JZm6bD6R9OyPbh_jlfRKvHYplnq8hxkH3EtapSURmwIFOrjTFCVjEzKEvUXIuy-l5uK4hLKflgsQaZrTWWEDMrQUzIzW-vQ8TdKbijCZ-7_wvEF9QQUkU |
| ContentType | Conference Proceeding |
| DBID | 6IE 6IH CBEJK RIE RIO |
| DOI | 10.1109/ICCAD57390.2023.10323722 |
| DatabaseName | IEEE Electronic Library (IEL) Conference Proceedings IEEE Proceedings Order Plan (POP) 1998-present by volume IEEE Xplore All Conference Proceedings IEEE Electronic Library (IEL) IEEE Proceedings Order Plans (POP) 1998-present |
| DatabaseTitleList | |
| Database_xml | – sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering |
| EISBN | 9798350322255 |
| EISSN | 1558-2434 |
| EndPage | 09 |
| ExternalDocumentID | 10323722 |
| Genre | orig-research |
| GrantInformation_xml | – fundername: Semiconductor Research Corporation (SRC) funderid: 10.13039/100000028 |
| GroupedDBID | 6IE 6IF 6IH 6IL 6IN AAWTH ABLEC ADZIZ ALMA_UNASSIGNED_HOLDINGS BEFXN BFFAM BGNUA BKEBE BPEOZ CBEJK CHZPO FEDTE IEGSK IJVOP M43 OCL RIE RIL RIO |
| ID | FETCH-LOGICAL-i204t-295d83da0c048c9aaa34d61ae4be9293bd2255cd06b44248ccae1cf9eb061c403 |
| IEDL.DBID | RIE |
| ISICitedReferencesCount | 12 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001116715100065&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| IngestDate | Wed Aug 27 02:22:16 EDT 2025 |
| IsPeerReviewed | false |
| IsScholarly | true |
| Language | English |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-i204t-295d83da0c048c9aaa34d61ae4be9293bd2255cd06b44248ccae1cf9eb061c403 |
| PageCount | 9 |
| ParticipantIDs | ieee_primary_10323722 |
| PublicationCentury | 2000 |
| PublicationDate | 2023-Oct.-28 |
| PublicationDateYYYYMMDD | 2023-10-28 |
| PublicationDate_xml | – month: 10 year: 2023 text: 2023-Oct.-28 day: 28 |
| PublicationDecade | 2020 |
| PublicationTitle | Digest of technical papers - IEEE/ACM International Conference on Computer-Aided Design |
| PublicationTitleAbbrev | ICCAD |
| PublicationYear | 2023 |
| Publisher | IEEE |
| Publisher_xml | – name: IEEE |
| SSID | ssj0020286 |
| Score | 2.4427855 |
| Snippet | Graph Convolutional Networks (GCNs) are pivotal in extracting latent information from graph data across various domains, yet their acceleration on mainstream... |
| SourceID | ieee |
| SourceType | Publisher |
| StartPage | 01 |
| SubjectTerms | Accelerator architectures Bandwidth Benchmark testing Convolution GPUs Graph Convolution Network Graphics processing units Memory management parallel computing Parallel processing sparse matrix multiplication (SpMM) |
| Title | Accel-GCN: High-Performance GPU Accelerator Design for Graph Convolution Networks |
| URI | https://ieeexplore.ieee.org/document/10323722 |
| WOSCitedRecordID | wos001116715100065&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07T8MwELagYoCFVxFveWB169quE7OhQgtLFCQqdav8uEiVUIv6-v2c3RcMDCxWFCeyYuf8ne3vuyPkgTsdkc4zJ7FQJqDNtVxg6Ll6V1XCClmlZBNZUeSDgSnXYvWkhQGARD6DRrxMZ_lh4hdxq6wZg7_JTOCMu59leiXW2q6uECj1hqrDTfOtg5_SznBJ34gZwhubd39lUUkg0j3-Z_MnpL6T49FyCzSnZA_GZ-ToRyTBc_L-5D18sl6neKSRusHKnSCA9so-TfWQztTpc2JtUKynvRivmmI7y_UvSIsVL3xWJ_3uy0fnla2zJbCR4GrOhGmHXAbLPRqlN9ZaqYJuWVAO0AeSLqDptn3g2ikl8BFvoeUrAw4h3SsuL0htPBnDJaE2d14bLCUHNHCFXoOTGn098A40ZFekHntn-LUKiDHcdMz1H_dvyGEcgzjli_yW1ObTBdyRA7-cj2bT-zSM30a6nZ0 |
| linkProvider | IEEE |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3JTsMwELVQQQIubEXs-MA1rWs7GzdU6CJKFKRW6q3yMpEqVS3q9v2M3Q0OHLhYUZwoyTiTN47fmyHkienIIZ0JtMBGphZ9rqZtgJGr0UXBFReFLzYRZ1nS76f5WqzutTAA4MlnUHGbfi3fTszC_SqruuRvIub4xd0PpeRsJdfazq8QKqMNWYel1XYdHyaMcVJfcTXCK5uzf9VR8TDSOPnnDZyS8k6QR_Mt1JyRPRifk-MfuQQvyOeLMTAKmvXsmTryRpDvJAG0mfeo7we_qk5fPW-DYj9tuozVFK-zXL-ENFsxw2dl0mu8deutYF0vIRhyJucBT0ObCKuYQbc0qVJKSBvVFEgNGAUJbdF5Q2NZpNF6eIhRUDNFChpB3UgmLklpPBnDFaEq0SZKsRUM0MUlxg1aRBjtgdEQQXxNys46g69VSozBxjA3f-x_JIet7kdn0Gln77fkyI2HAwCe3JHSfLqAe3JglvPhbPrgh_Qbqm2g5A |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=proceeding&rft.title=Digest+of+technical+papers+-+IEEE%2FACM+International+Conference+on+Computer-Aided+Design&rft.atitle=Accel-GCN%3A+High-Performance+GPU+Accelerator+Design+for+Graph+Convolution+Networks&rft.au=Xie%2C+Xi&rft.au=Peng%2C+Hongwu&rft.au=Hasan%2C+Amit&rft.au=Huang%2C+Shaoyi&rft.date=2023-10-28&rft.pub=IEEE&rft.eissn=1558-2434&rft.spage=01&rft.epage=09&rft_id=info:doi/10.1109%2FICCAD57390.2023.10323722&rft.externalDocID=10323722 |