SGI and Intel boost oil and gas exploration
DownUnder GeoSolutions (DUG), a geoscience specialist providing exploration and production services to the global oil and gas industry, has boosted its HPC facilities with a new supercomputer from SGI that uses 3,800 Intel Xeon Phi coprocessors.
DUG and SGI worked closely with Intel in developing the system, which is one of the largest commercial deployments of Intel Xeon Phi coprocessors, and the largest such deployment intended for use in the petroleum geosciences industry.
‘We’ve already started to see dramatic improvements in turn-around times when we compare our upgraded machines to those without coprocessors. Our time migration now runs more than ten times faster. Our depth migration runs six times faster. DUG has also seen its Reverse Time Migration (RTM) run significantly faster using this new technology,’ said Dr Matt Lamont, DUG’s managing director.
Dr Lamont explained that in addition to his top researchers being more familiar with the many-core programming model, he stated that programming on GPUs was roughly four times more difficult than on the Intel Xeon Phi for Kirchhoff Time Migration, a key algorithm for oil and gas exploration.
He also stated that overall total cost of ownership is a top consideration for DUG, and that the performance, programming environment and end price all play a role in this. For HPC workloads at DownUnder GeoSolutions, consisting of a full suite of seismic processing and imaging algorithms, the Intel Xeon Phi coprocessors were selected as the best option.
Charles Wuischpard, vice president and general manager of Workstations and HPC at Intel said: ‘The innovative use of Intel Xeon Phi coprocessors by DownUnder GeoSolutions is enabling their geophysicists to work with large seismic data sets interactively. In an industry where time is invaluable, the Intel Xeon Phi-based SGI system allows DUG to test more and faster, leading to better results in a much shorter period of time. Their integration of Intel Xeon Phi coprocessors has enabled them to quickly adapt their existing code and immediately pass this value on to their customers.’