@inproceedings{zhang2021hyperwavve,
author = {Zhang, Qie and Iordanescu, George and Tok, Wee Hyong and Brandsberg-Dahl, Sverre and Srinivasan, Hari Krishnan and Chandra, Ranveer and Kukreja, Navjot and Gorman, Gerard},
title = {Hyperwavve: A cloud-native solution for hyperscale seismic imaging on Azure},
organization = {SEG},
booktitle = {First International Meeting for Applied Geoscience & Energy},
year = {2021},
month = {September},
abstract = {As cloud-computing becomes more and more popular lately, we explore its potential for hyperscale seismic imaging workloads on Azure. We introduce our cloud-native fault-tolerant solution named Hyperwavve which is based on advanced cloud technologies including Docker/Container, Kubernetes and Dask. We demonstrate a large-scale 3D FWI using 1000 VMs/nodes on Azure, where Hyperwavve uses distributed containerized processes to successfully invert for the full 3D (20x20x5 km3) overthrust velocity model. We also further validate that our Hyperwavve can distribute FWI work onto 6000 (or more) VMs/nodes concurrently. Last, we show that our Python-based FWI runs on both Azure CPUs and GPUs including various architectures.},
url = {http://approjects.co.za/?big=en-us/research/publication/hyperwavve-a-cloud-native-solution-for-hyperscale-seismic-imaging-on-azure/},
}