Performance and energy consumption analysis of a seismic application for three different architectures intended for oil and gas industry

Abstract

In the most cases, seismic migration applications demand considerable computing throughput, since this feature denotes a need to apply complex models that are continuously run to evaluate drilling of petroleum wells. Given to the inherent computational complexity and the immense amount of processed data, High-Performance Computing (HPC) solutions can be attractive to perform this kind of application. Nowadays, low energy consumption has been highly desirable in high-performance processing when applied to large clusters that continuously run certain applications. Thereby, the use of an architecture that combines high performance and low energy consumption is desirable. This work describes an analysis in terms of performance, energy consumption and cost for three different architectures (Multicore, FPGA and GPGPU) intended to process a seismic application based on RTM (Reverse Time Migration) algorithm for an industrial application with this objective. Results indicate that the GPGPU architecture achieved the best performance in terms of energy consumption. When compared to the Multicore architecture, an about 15 times higher efficiency/Joule was observed. This architecture performed the RTM algorithm about 32 times faster when compared with the non-optimized implementation on CPU.