Accelerating Time and Depth Seismic Migration by CPU and GPU Cooperation

• Published in: International journal of parallel programming Vol. 40; no. 3; pp. 290 - 312
• Main Authors: Panetta, Jairo, Teixeira, Thiago, de Souza Filho, Paulo R. P., da Cunha Filho, Carlos A., Sotelo, David, Roxo da Motta, Fernando M., Pinheiro, Silvio Sinedino, Romanelli Rosa, Andre L., Monnerat, Luiz R., Carneiro, Leandro T., de Albrecht, Carlos H. B.
• Format: Journal Article
• Language: English
• Published: Boston Springer US 01.06.2012; Springer Nature B.V
• Subjects: Algorithms; Architecture; Central processing units; Clusters; Codes; Computer Science; Cooperation; CPUs; Medical imaging; Medical research; Migration; Overloading; Parallel processing; Petroleum industry; Processor Architectures; Seismic engineering; Seismic phenomena; Software Engineering/Programming and Operating Systems; Strategy; Studies; Theory of Computation; Parallel processing; Seismic migration; Dynamic load balancing; GPGPU; CPU-GPU cooperation
• ISSN: 0885-7458; 1573-7640
• DOI: 10.1007/s10766-011-0185-2

In this paper, we discuss the porting strategy and the achieved performance of Petrobras production code that implements Kirchhoff prestack time and depth seismic migration algorithms on a cluster of 64 GPUs and 256 CPU cores. The time migration of a real-world survey was 20 times faster in the full cluster with GPUs than in the full cluster without GPUs. The introduction of GPUs accelerated the depth migration of another real-world survey by a factor of 5. In both cases, a large fraction of the achieved gains were due to the porting strategy that privileges CPU-GPU cooperation instead of pure GPU speed. Another strategy that contributed to the high gains was GPU overloading by multiple MPI processes. To cope with the heterogeneous load of real-world data, we developed a dynamic load balancing scheme that significantly speeds up computation.