FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4285264230> ?p ?o ?g. }

• W4285264230 endingPage "163" @default.
• W4285264230 startingPage "150" @default.
• W4285264230 abstract "Graph convolutional network (GCN) is a promising but computing- and memory-intensive learning model. Processing-in-memory (PIM) architecture based on the resistive random access memory-based crossbar (ReRAM crossbar) is a natural fit for GCN inference. It can reduce the data movements and compute the vector-matrix multiplication (VMM) in analog. However, it requires an unbearable crossbar cost to leverage the massive parallelism exhibited in GCNs. First, this article explores the design space for GCN inference on ReRAM crossbars and presents the first PIM-based GCN accelerator named PIMGCN, PIMGCN employs dense data mapping and a search-execute architecture to take full advantage of the intravertex parallelisms with acceptable crossbars cost. Two scheduling strategies for PIMGCN to maximize the intervertex parallelisms and optimize the pipeline are proposed. The optimal scheduling is reduced to a maximum independent set problem, which is solved by a novel node-grouping algorithm. Second, this article explores the task-irrelevant information in the graphs and proposes an adaptively sparsified GCN network targeted for PIMGCN, which is named as ASparGCN. ASparGCN exploits a multilayer perceptron (MLP)-based edge predictor to get edge selection strategies for each GCN layer separately and adaptively in the training stage, and only inferences with the selected edges in the test stage. We design two regularization terms to guide the selection strategies to achieve architecture-friendly sparse graphs for PIMGCN. The overall algorithm-architecture co-design is named as PASGCN. Compared to the state-of-the-art software framework running on Intel Xeon CPU and NVIDIA RTX8000 GPU, PASGCN achieves an average of <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$16455times $ </tex-math></inline-formula> and <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$110.7times $ </tex-math></inline-formula> speedup and 8.0E <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$+ 06times $ </tex-math></inline-formula> and 6.67E <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$+ 03times $ </tex-math></inline-formula> energy reduction, respectively. Compared with the ASIC accelerator HyGCN (Yan et al., 2020), PASGCN achieves <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$326.31times $ </tex-math></inline-formula> speedup and <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$124.8times $ </tex-math></inline-formula> energy reduction." @default.
• W4285264230 created "2022-07-14" @default.
• W4285264230 creator A5017670541 @default.
• W4285264230 creator A5022287798 @default.
• W4285264230 creator A5029153042 @default.
• W4285264230 creator A5041228669 @default.
• W4285264230 creator A5053801300 @default.
• W4285264230 creator A5067161373 @default.
• W4285264230 creator A5068919040 @default.
• W4285264230 creator A5087000964 @default.
• W4285264230 date "2023-01-01" @default.
• W4285264230 modified "2023-10-12" @default.
• W4285264230 title "PASGCN: An ReRAM-Based PIM Design for GCN With Adaptively Sparsified Graphs" @default.
• W4285264230 cites W2010202670 @default.
• W4285264230 cites W2013028205 @default.
• W4285264230 cites W2089247135 @default.
• W4285264230 cites W2399958287 @default.
• W4285264230 cites W2508602506 @default.
• W4285264230 cites W2518281301 @default.
• W4285264230 cites W2725159389 @default.
• W4285264230 cites W2794243109 @default.
• W4285264230 cites W2921653370 @default.
• W4285264230 cites W2962903741 @default.
• W4285264230 cites W2964051675 @default.
• W4285264230 cites W3004028296 @default.
• W4285264230 cites W3007272028 @default.
• W4285264230 cites W3017228913 @default.
• W4285264230 cites W3043072009 @default.
• W4285264230 cites W3043715351 @default.
• W4285264230 cites W3105753905 @default.
• W4285264230 cites W3211445324 @default.
• W4285264230 cites W4210257598 @default.
• W4285264230 cites W4233317171 @default.
• W4285264230 doi "https://doi.org/10.1109/tcad.2022.3175031" @default.
• W4285264230 hasPublicationYear "2023" @default.
• W4285264230 type Work @default.
• W4285264230 citedByCount "3" @default.
• W4285264230 countsByYear W42852642302022 @default.
• W4285264230 countsByYear W42852642302023 @default.
• W4285264230 crossrefType "journal-article" @default.
• W4285264230 hasAuthorship W4285264230A5017670541 @default.
• W4285264230 hasAuthorship W4285264230A5022287798 @default.
• W4285264230 hasAuthorship W4285264230A5029153042 @default.
• W4285264230 hasAuthorship W4285264230A5041228669 @default.
• W4285264230 hasAuthorship W4285264230A5053801300 @default.
• W4285264230 hasAuthorship W4285264230A5067161373 @default.
• W4285264230 hasAuthorship W4285264230A5068919040 @default.
• W4285264230 hasAuthorship W4285264230A5087000964 @default.
• W4285264230 hasConcept C11413529 @default.
• W4285264230 hasConcept C118524514 @default.
• W4285264230 hasConcept C147789679 @default.
• W4285264230 hasConcept C154945302 @default.
• W4285264230 hasConcept C156884757 @default.
• W4285264230 hasConcept C173608175 @default.
• W4285264230 hasConcept C17525397 @default.
• W4285264230 hasConcept C182019814 @default.
• W4285264230 hasConcept C185592680 @default.
• W4285264230 hasConcept C2776214188 @default.
• W4285264230 hasConcept C29984679 @default.
• W4285264230 hasConcept C41008148 @default.
• W4285264230 hasConcept C76155785 @default.
• W4285264230 hasConceptScore W4285264230C11413529 @default.
• W4285264230 hasConceptScore W4285264230C118524514 @default.
• W4285264230 hasConceptScore W4285264230C147789679 @default.
• W4285264230 hasConceptScore W4285264230C154945302 @default.
• W4285264230 hasConceptScore W4285264230C156884757 @default.
• W4285264230 hasConceptScore W4285264230C173608175 @default.
• W4285264230 hasConceptScore W4285264230C17525397 @default.
• W4285264230 hasConceptScore W4285264230C182019814 @default.
• W4285264230 hasConceptScore W4285264230C185592680 @default.
• W4285264230 hasConceptScore W4285264230C2776214188 @default.
• W4285264230 hasConceptScore W4285264230C29984679 @default.
• W4285264230 hasConceptScore W4285264230C41008148 @default.
• W4285264230 hasConceptScore W4285264230C76155785 @default.
• W4285264230 hasFunder F4320321001 @default.
• W4285264230 hasFunder F4320335777 @default.
• W4285264230 hasIssue "1" @default.
• W4285264230 hasLocation W42852642301 @default.
• W4285264230 hasOpenAccess W4285264230 @default.
• W4285264230 hasPrimaryLocation W42852642301 @default.
• W4285264230 hasRelatedWork W2536878608 @default.
• W4285264230 hasRelatedWork W2606690302 @default.
• W4285264230 hasRelatedWork W2889758018 @default.
• W4285264230 hasRelatedWork W2969116217 @default.
• W4285264230 hasRelatedWork W4254672563 @default.
• W4285264230 hasRelatedWork W4289541932 @default.
• W4285264230 hasRelatedWork W4308098692 @default.
• W4285264230 hasRelatedWork W4308870977 @default.
• W4285264230 hasRelatedWork W4315464822 @default.
• W4285264230 hasRelatedWork W4381328989 @default.
• W4285264230 hasVolume "42" @default.
• W4285264230 isParatext "false" @default.
• W4285264230 isRetracted "false" @default.
• W4285264230 workType "article" @default.