Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912759934> ?p ?o ?g. }
- W2912759934 abstract "Graphics processing units (GPUs) have been widely adopted to accelerate the training of deep neural networks (DNNs). Although the computational performance of GPUs has been improving steadily, the memory size of modern GPUs is still quite limited, which restricts the sizes of DNNs that can be trained on GPUs, and hence raises serious challenges. This paper introduces a framework, referred to as moDNN (memory optimal DNN training on GPUs), to optimize the memory usage in DNN training. moDNN supports automatic tuning of DNN training code to match any given memory budget (not smaller than the theoretical lower bound). By taking full advantage of overlapping computations and data transfers, we develop new heuristics to judiciously schedule data offloading and prefetching transfers, together with convolution algorithm selection, to optimize memory usage. We further devise a new sub-batch size selection method which also greatly reduces memory usage. moDNN can save memory usage up to 59×, compared with an ideal case which assumes that the GPU memory is sufficient to hold all data. When executing moDNN on a GPU with 12 GB memory, the training time is increased by only 3 percent, which is much shorter than that incurred by the best known approach, vDNN. Furthermore, we propose an optimization strategy for moDNN on multiple GPUs again by utilizing the idea of overlapping data transfers and GPU computations. The results show that 3.7× speedup is attained on four GPUs." @default.
- W2912759934 created "2019-02-21" @default.
- W2912759934 creator A5016864694 @default.
- W2912759934 creator A5064527448 @default.
- W2912759934 creator A5087244134 @default.
- W2912759934 creator A5089356973 @default.
- W2912759934 date "2019-03-01" @default.
- W2912759934 modified "2023-10-16" @default.
- W2912759934 title "moDNN: Memory Optimal Deep Neural Network Training on Graphics Processing Units" @default.
- W2912759934 cites W128208790 @default.
- W2912759934 cites W1590012787 @default.
- W2912759934 cites W1667652561 @default.
- W2912759934 cites W1686810756 @default.
- W2912759934 cites W1901129140 @default.
- W2912759934 cites W1903029394 @default.
- W2912759934 cites W1935978687 @default.
- W2912759934 cites W1972501001 @default.
- W2912759934 cites W2016879831 @default.
- W2912759934 cites W2046441184 @default.
- W2912759934 cites W2097117768 @default.
- W2912759934 cites W2102017903 @default.
- W2912759934 cites W2102605133 @default.
- W2912759934 cites W2155893237 @default.
- W2912759934 cites W2157282184 @default.
- W2912759934 cites W2160815625 @default.
- W2912759934 cites W2163605009 @default.
- W2912759934 cites W2168231600 @default.
- W2912759934 cites W2172654076 @default.
- W2912759934 cites W2186615578 @default.
- W2912759934 cites W2194775991 @default.
- W2912759934 cites W2271840356 @default.
- W2912759934 cites W2286365479 @default.
- W2912759934 cites W2300242332 @default.
- W2912759934 cites W2319920447 @default.
- W2912759934 cites W2332506150 @default.
- W2912759934 cites W2562749854 @default.
- W2912759934 cites W2798707674 @default.
- W2912759934 cites W2884834742 @default.
- W2912759934 cites W2963042536 @default.
- W2912759934 cites W2963374099 @default.
- W2912759934 cites W2963456262 @default.
- W2912759934 cites W2963674932 @default.
- W2912759934 cites W2964174152 @default.
- W2912759934 cites W3145128584 @default.
- W2912759934 doi "https://doi.org/10.1109/tpds.2018.2866582" @default.
- W2912759934 hasPublicationYear "2019" @default.
- W2912759934 type Work @default.
- W2912759934 sameAs 2912759934 @default.
- W2912759934 citedByCount "7" @default.
- W2912759934 countsByYear W29127599342019 @default.
- W2912759934 countsByYear W29127599342020 @default.
- W2912759934 countsByYear W29127599342021 @default.
- W2912759934 countsByYear W29127599342023 @default.
- W2912759934 crossrefType "journal-article" @default.
- W2912759934 hasAuthorship W2912759934A5016864694 @default.
- W2912759934 hasAuthorship W2912759934A5064527448 @default.
- W2912759934 hasAuthorship W2912759934A5087244134 @default.
- W2912759934 hasAuthorship W2912759934A5089356973 @default.
- W2912759934 hasBestOaLocation W29127599341 @default.
- W2912759934 hasConcept C111919701 @default.
- W2912759934 hasConcept C113775141 @default.
- W2912759934 hasConcept C11413529 @default.
- W2912759934 hasConcept C127705205 @default.
- W2912759934 hasConcept C173608175 @default.
- W2912759934 hasConcept C176649486 @default.
- W2912759934 hasConcept C21442007 @default.
- W2912759934 hasConcept C2779851693 @default.
- W2912759934 hasConcept C2781357197 @default.
- W2912759934 hasConcept C41008148 @default.
- W2912759934 hasConcept C45374587 @default.
- W2912759934 hasConcept C68339613 @default.
- W2912759934 hasConcept C68387754 @default.
- W2912759934 hasConcept C9390403 @default.
- W2912759934 hasConcept C98986596 @default.
- W2912759934 hasConceptScore W2912759934C111919701 @default.
- W2912759934 hasConceptScore W2912759934C113775141 @default.
- W2912759934 hasConceptScore W2912759934C11413529 @default.
- W2912759934 hasConceptScore W2912759934C127705205 @default.
- W2912759934 hasConceptScore W2912759934C173608175 @default.
- W2912759934 hasConceptScore W2912759934C176649486 @default.
- W2912759934 hasConceptScore W2912759934C21442007 @default.
- W2912759934 hasConceptScore W2912759934C2779851693 @default.
- W2912759934 hasConceptScore W2912759934C2781357197 @default.
- W2912759934 hasConceptScore W2912759934C41008148 @default.
- W2912759934 hasConceptScore W2912759934C45374587 @default.
- W2912759934 hasConceptScore W2912759934C68339613 @default.
- W2912759934 hasConceptScore W2912759934C68387754 @default.
- W2912759934 hasConceptScore W2912759934C9390403 @default.
- W2912759934 hasConceptScore W2912759934C98986596 @default.
- W2912759934 hasFunder F4320306087 @default.
- W2912759934 hasFunder F4320335353 @default.
- W2912759934 hasLocation W29127599341 @default.
- W2912759934 hasOpenAccess W2912759934 @default.
- W2912759934 hasPrimaryLocation W29127599341 @default.
- W2912759934 hasRelatedWork W1988195213 @default.
- W2912759934 hasRelatedWork W2061677511 @default.
- W2912759934 hasRelatedWork W2141978748 @default.
- W2912759934 hasRelatedWork W2407656474 @default.
- W2912759934 hasRelatedWork W2533998425 @default.
- W2912759934 hasRelatedWork W2614413552 @default.