Unprecedented Seismic Processing Performance
Cray systems have been enabling the most demanding seismic processing, exploration and production challenges for decades. Our technology has been tried, tested and successfully used in the world's most compute- and data-intensive environments.
Cray® XC™ series supercomputers provide scalable capability computing for the intense data and compute requirements of seismic processing and seismic imaging. The XC system's intelligent Aries™ interconnect helps deliver unprecedented performance for spectral algorithms like reverse time migration (RTM) and full waveform inversion (FWI) by accelerating many-to-many communications.
A standout feature for the oil & gas industry is the XC system's flexibility for using GPUs and CPUs. This allows exploration organizations to handle seismic workflows that incorporate optimized codes for both compute infrastructures.
Accelerate Application Runtime up to 9x
Landmark and Cray offer a tested solution with SeisSpace® seismic processing software on Cray® CS™ series cluster supercomputers. In addition, the Cray® CS-Storm™ GPU-based solution for RTM provides geophysicists with a fast, efficient and reliable way to run their most complex seismic processing workflows.
SeisSpace® ProMAX® software is a comprehensive seismic processing framework that includes the entire suite of processing technologies and algorithms designed by Halliburton Landmark for use in large volume 2-D or 3-D marine and land seismic acquisitions. The software facilitates construction of the optimum seismic imaging of the most difficult geologic targets and allows O&G companies to better understand their seismic data.
Cray conducted tests and found that running SeisSpace on a Cray CS400 system led to substantial improvements in application runtime performance over traditional clustered infrastructures. Testing revealed up to 9x application runtime improvements for RTM (LDI and Tierra), Kirchhoff prestack depth migration, merge and sort, and acoustic and elastic forward modeling over similar infrastructures.
The CS-Storm solution for RTM provides GPU-centric performance for seismic exploration, with a 4:1 GPU-to-processor ratio and proprietary I/O design to eliminate communication bottlenecks that impede the performance of other clusters. It provides GPU capacity of up to 1/3 PF per rack, leveraging the latest K80 accelerators.