{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,17]],"date-time":"2026-04-17T06:30:16Z","timestamp":1776407416563,"version":"3.51.2"},"reference-count":40,"publisher":"IEEE","license":[{"start":{"date-parts":[[2019,10,1]],"date-time":"2019-10-01T00:00:00Z","timestamp":1569888000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2019,10,1]],"date-time":"2019-10-01T00:00:00Z","timestamp":1569888000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2019,10]]},"DOI":"10.1109\/ipccc47392.2019.8958738","type":"proceedings-article","created":{"date-parts":[[2020,1,16]],"date-time":"2020-01-16T21:04:32Z","timestamp":1579208672000},"page":"1-8","source":"Crossref","is-referenced-by-count":8,"title":["Topology-aware Sparse Allreduce for Large-scale Deep Learning"],"prefix":"10.1109","author":[{"given":"Truong Thao","family":"Nguyen","sequence":"first","affiliation":[{"name":"National Institute of Advanced Industrial Science and Technology (AIST) 2-12-1 W8-92 Ookayama,AIST-Tokyo Tech Real World Big-Data Computation Open Innovation Laboratory (RWBC-OIL),Meguro-ku,Tokyo,Japan,152-8550"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mohamed","family":"Wahib","sequence":"additional","affiliation":[{"name":"National Institute of Advanced Industrial Science and Technology (AIST) 2-12-1 W8-92 Ookayama,AIST-Tokyo Tech Real World Big-Data Computation Open Innovation Laboratory (RWBC-OIL),Meguro-ku,Tokyo,Japan,152-8550"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ryousei","family":"Takano","sequence":"additional","affiliation":[{"name":"National Institute of Advanced Industrial Science and Technology (AIST) 2-12-1 W8-92 Ookayama,AIST-Tokyo Tech Real World Big-Data Computation Open Innovation Laboratory (RWBC-OIL),Meguro-ku,Tokyo,Japan,152-8550"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref39","year":"2019","journal-title":"CS7500 InfiniBand EDR 100Gb\/s Switch System"},{"key":"ref38","year":"2019","journal-title":"SB7800 InfiniBand EDR 100Gb\/s Switch System"},{"key":"ref33","first-page":"1709","article-title":"QSGD: Communication-efficient SGD via gradient quantization and encoding in","author":"alistarh","year":"2017","journal-title":"Advances in neural information processing systems"},{"key":"ref32","article-title":"MXNET-MPI: Embedding MPI parallelism in Parameter Server Task Model for scaling Deep Learning","author":"mamidala","year":"2018","journal-title":"arXiv preprint arXiv 1801 03186"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1145\/3155284.3018769"},{"key":"ref30","year":"0"},{"key":"ref37","year":"2019"},{"key":"ref36","first-page":"5977","article-title":"The convergence of sparsified gradient methods in","author":"alistarh","year":"2018","journal-title":"Proceedings of the 32Nd International Conference on Neural Information Processing Systems"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D17-1045"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/MLHPC.2016.004"},{"key":"ref10","article-title":"ImageNet\/ResNet-50 Training in 224 Seconds","volume":"abs 1811 5233","author":"mikami","year":"2018","journal-title":"CoRR"},{"key":"ref40","article-title":"Evaluating Modern GPU Interconnect: PCIe, NVLink, NV-SLI, NVSwitch and GPUDirect","volume":"abs 1903 4611","author":"li","year":"2019","journal-title":"CoRR"},{"key":"ref11","article-title":"Yet Another Accelerated SGD: ResNet-50 Training on ImageNet in 74.7 seconds","volume":"abs 1903 12650","author":"yamazaki","year":"2019","journal-title":"CoRR"},{"key":"ref12","first-page":"770","article-title":"Deep residual learning for image recognition ser","author":"he","year":"2016","journal-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"key":"ref14","first-page":"64","article-title":"Scalable reduction collectives with data partitioning-based multi-leader design ser","author":"bayatpour","year":"2017","journal-title":"Proc Int Conf High-Performance Computing"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTER.2018.00014"},{"key":"ref16","first-page":"216","article-title":"Hierarchical distributed-memory multi-leader mpi-allreduce for deep learning workloads","author":"nguyen","year":"2018","journal-title":"Sixth International Symposium on Computing and Networking CAN-DAR Workshops 2018"},{"key":"ref17","article-title":"1-Bit Stochastic Gradient Descent and Application to Data-Parallel Distributed Training of Speech DNNs ser","author":"seide","year":"2014","journal-title":"Fifteenth Annual Conference of the International Speech Communication Association (Interspeech)"},{"key":"ref18","first-page":"1509","article-title":"TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning in","author":"wen","year":"2017","journal-title":"Advances in Neural IInformation Processing Systems"},{"key":"ref19","article-title":"Scalable distributed DNN training using commodity GPU cloud computing ser","author":"strom","year":"2015","journal-title":"Sixteenth Annual Conference of the International Speech Communication Association (INTERSPEECH)"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1016\/j.jpdc.2008.09.002"},{"key":"ref4","first-page":"9","article-title":"Scaling deep learning on GPU and knights landing clusters ser","author":"you","year":"2017","journal-title":"Proceedings of the International Conference for High Performance Computing Networking Storage and Analysis"},{"key":"ref27","first-page":"1","article-title":"Bandwidth efficient all-reduce operation on tree topologies ser","author":"patarasuk","year":"2007","journal-title":"Parallel and Distributed Processing Symposium 2007 IPDPS 2007"},{"key":"ref3","article-title":"GPipe: Efficient Training of Giant Neural Networks using Pipeline Parallelism","volume":"abs 1811 6965","author":"huang","year":"2018","journal-title":"CoRR"},{"key":"ref6","article-title":"Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour","volume":"abs 1706 2677","author":"goyal","year":"2017","journal-title":"CoRR"},{"key":"ref29","year":"2019"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2014.6853593"},{"key":"ref8","article-title":"Extremely Large Minibatch SGD: Training ResNet-50 on ImageNet in 15 Minutes","volume":"abs 1711 4325","author":"akiba","year":"2017","journal-title":"CoRR"},{"key":"ref7","article-title":"Don’t Decay the Learning Rate, Increase the Batch Size in","author":"smith","year":"2018","journal-title":"6th International Conference on Learning Representations ICLR 2018"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1145\/3225058.3225069"},{"key":"ref9","article-title":"Highly Scalable Deep Learning Training System with Mixed-Precision: Training ImageNet in Four Minutes","volume":"abs 1807 11205","author":"jia","year":"2018","journal-title":"CoRR"},{"key":"ref1","article-title":"Demystifying parallel and distributed deep learning: An in-depth concurrency analysis","volume":"abs 1802 9941","author":"ben-nun","year":"2018","journal-title":"CoRR"},{"key":"ref20","article-title":"Deep gradient compression: Reducing the communication bandwidth for distributed training in","author":"lin","year":"0","journal-title":"6th International Conference on Learning Representations ICLR 2018"},{"key":"ref22","volume":"abs 1808 4357","year":"2018","journal-title":"CoRR"},{"key":"ref21","article-title":"Sparse binary compression: Towards distributed deep learning with minimal communication","volume":"abs 1805 8768","author":"sattler","year":"2018","journal-title":"CoRR"},{"key":"ref24","year":"0","journal-title":"White Paper NVIDIA DGX-l System Architecture The Fastest Platform for Deep Learning"},{"key":"ref23","article-title":"SparCML: High-Performance Sparse Communication for Machine Learning","volume":"abs 1802 8021","author":"renggli","year":"2018","journal-title":"CoRR"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1177\/1094342005051521"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1002\/cpe.1206"}],"event":{"name":"2019 IEEE 38th International Performance Computing and Communications Conference (IPCCC)","location":"London, UK","start":{"date-parts":[[2019,10,29]]},"end":{"date-parts":[[2019,10,31]]}},"container-title":["2019 IEEE 38th International Performance Computing and Communications Conference (IPCCC)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/8955479\/8958711\/08958738.pdf?arnumber=8958738","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,7,31]],"date-time":"2025-07-31T18:25:38Z","timestamp":1753986338000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/8958738\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,10]]},"references-count":40,"URL":"https:\/\/doi.org\/10.1109\/ipccc47392.2019.8958738","relation":{},"subject":[],"published":{"date-parts":[[2019,10]]}}}