Matches in SemOpenAlex for { <https://semopenalex.org/work/W3103606447> ?p ?o ?g. }
- W3103606447 endingPage "207111" @default.
- W3103606447 startingPage "207097" @default.
- W3103606447 abstract "Distributed processing using high-performance computing resources is essential for developers to train large-scale deep neural networks (DNNs). The major impediment to distributed DNN training is the communication bottleneck during the parameter exchange among the distributed DNN training workers. The communication bottleneck increases training time and decreases the utilization of the computational resources. Our previous study, SoftMemoryBox (SMB1) presented considerably superior performance compared to message passing interface (MPI) in the parameter communication of distributed DNN training. However, SMB1 had disadvantages such as the limited scalability of the distributed DNN training due to the restricted communication bandwidth from a single memory server, inability to provide a synchronization function for the shared memory buffer, and low portability/usability as a consequence of the kernel-level implementation. This paper proposes a scalable, shared memory buffer framework, called SoftMemoryBox II (SMB2), which overcomes the shortcomings of SMB1. With SMB2, distributed training processes can easily share virtually unified shared memory buffers composed of memory segments provided from remote memory servers and can exchange DNN parameters at high speed through the shared memory buffer. The scalable communication bandwidth of the SMB2 framework facilitates the reduction of DNN distributed training times compared to SMB1. According to intensive evaluation results, the communication bandwidth of the proposed SMB2 is 6.3 times greater than that of SMB1 when the SMB2 framework is scaled out to use eight memory servers. Moreover, the training time of SMB2-based asynchronous distributed training of five DNN models is up to 2.4 times faster than SMB1-based training." @default.
- W3103606447 created "2020-11-23" @default.
- W3103606447 creator A5052536261 @default.
- W3103606447 creator A5063953385 @default.
- W3103606447 date "2020-01-01" @default.
- W3103606447 modified "2023-09-24" @default.
- W3103606447 title "SoftMemoryBox II: A Scalable, Shared Memory Buffer Framework for Accelerating Distributed Training of Large-Scale Deep Neural Networks" @default.
- W3103606447 cites W1442374986 @default.
- W3103606447 cites W1547840952 @default.
- W3103606447 cites W1598866093 @default.
- W3103606447 cites W1686810756 @default.
- W3103606447 cites W1903676646 @default.
- W3103606447 cites W2060393849 @default.
- W3103606447 cites W2062022900 @default.
- W3103606447 cites W2086161653 @default.
- W3103606447 cites W2097117768 @default.
- W3103606447 cites W2102017903 @default.
- W3103606447 cites W2120432001 @default.
- W3103606447 cites W2138243089 @default.
- W3103606447 cites W2147768505 @default.
- W3103606447 cites W2155893237 @default.
- W3103606447 cites W2163605009 @default.
- W3103606447 cites W2168231600 @default.
- W3103606447 cites W2183341477 @default.
- W3103606447 cites W2184045248 @default.
- W3103606447 cites W2194775991 @default.
- W3103606447 cites W2241510067 @default.
- W3103606447 cites W2274162699 @default.
- W3103606447 cites W2336650964 @default.
- W3103606447 cites W2398934890 @default.
- W3103606447 cites W2405578611 @default.
- W3103606447 cites W2622263826 @default.
- W3103606447 cites W2739720758 @default.
- W3103606447 cites W2762776925 @default.
- W3103606447 cites W2800893679 @default.
- W3103606447 cites W2807970597 @default.
- W3103606447 cites W2808949243 @default.
- W3103606447 cites W2884711234 @default.
- W3103606447 cites W2888429796 @default.
- W3103606447 cites W2893813411 @default.
- W3103606447 cites W2901541570 @default.
- W3103606447 cites W2920668770 @default.
- W3103606447 cites W2955425717 @default.
- W3103606447 cites W2962747323 @default.
- W3103606447 cites W2962758826 @default.
- W3103606447 cites W2963804082 @default.
- W3103606447 cites W2963831937 @default.
- W3103606447 cites W2964081807 @default.
- W3103606447 cites W2964350391 @default.
- W3103606447 cites W2972087877 @default.
- W3103606447 cites W2997937417 @default.
- W3103606447 cites W3037288590 @default.
- W3103606447 doi "https://doi.org/10.1109/access.2020.3038112" @default.
- W3103606447 hasPublicationYear "2020" @default.
- W3103606447 type Work @default.
- W3103606447 sameAs 3103606447 @default.
- W3103606447 citedByCount "1" @default.
- W3103606447 countsByYear W31036064472021 @default.
- W3103606447 crossrefType "journal-article" @default.
- W3103606447 hasAuthorship W3103606447A5052536261 @default.
- W3103606447 hasAuthorship W3103606447A5063953385 @default.
- W3103606447 hasBestOaLocation W31036064471 @default.
- W3103606447 hasConcept C111919701 @default.
- W3103606447 hasConcept C120314980 @default.
- W3103606447 hasConcept C12186640 @default.
- W3103606447 hasConcept C133875982 @default.
- W3103606447 hasConcept C136085584 @default.
- W3103606447 hasConcept C149635348 @default.
- W3103606447 hasConcept C151319957 @default.
- W3103606447 hasConcept C173608175 @default.
- W3103606447 hasConcept C176649486 @default.
- W3103606447 hasConcept C204156049 @default.
- W3103606447 hasConcept C2776257435 @default.
- W3103606447 hasConcept C2780513914 @default.
- W3103606447 hasConcept C31258907 @default.
- W3103606447 hasConcept C39528615 @default.
- W3103606447 hasConcept C41008148 @default.
- W3103606447 hasConcept C48044578 @default.
- W3103606447 hasConcept C51290061 @default.
- W3103606447 hasConcept C91481028 @default.
- W3103606447 hasConcept C93996380 @default.
- W3103606447 hasConceptScore W3103606447C111919701 @default.
- W3103606447 hasConceptScore W3103606447C120314980 @default.
- W3103606447 hasConceptScore W3103606447C12186640 @default.
- W3103606447 hasConceptScore W3103606447C133875982 @default.
- W3103606447 hasConceptScore W3103606447C136085584 @default.
- W3103606447 hasConceptScore W3103606447C149635348 @default.
- W3103606447 hasConceptScore W3103606447C151319957 @default.
- W3103606447 hasConceptScore W3103606447C173608175 @default.
- W3103606447 hasConceptScore W3103606447C176649486 @default.
- W3103606447 hasConceptScore W3103606447C204156049 @default.
- W3103606447 hasConceptScore W3103606447C2776257435 @default.
- W3103606447 hasConceptScore W3103606447C2780513914 @default.
- W3103606447 hasConceptScore W3103606447C31258907 @default.
- W3103606447 hasConceptScore W3103606447C39528615 @default.
- W3103606447 hasConceptScore W3103606447C41008148 @default.
- W3103606447 hasConceptScore W3103606447C48044578 @default.
- W3103606447 hasConceptScore W3103606447C51290061 @default.