A report from the task force on high performance computing from the US secretary of energy’s advisory board (SEAB) makes interesting reading for oil and gas HPC users. While the fastest computer in the world, the Tianhe-2 at the national super computer center in Guangzhou, China, boasts a 34 teraflop performance, according to the SEAB report, some oil and gas market players ‘already run data centers with over 60 petaflops of compute capacity.’
The authors however nuance oil and gas HPC’s prowess, ‘Oil exploration workflows are composed of many high-level data processing stages with little automated integration. Stages are rerun by hand, in an ad hoc fashion, when the output does not satisfy subjective criteria dictated by the experience of the operators.’
SEAB is advocating exaflop class machines that will allow geophysicists to interact dynamically with ‘thousands or millions’ of scenarios and to better integrate other business areas such as reservoir simulation, oil field management and the supply chain.
Plans are in place through the Coral program to deliver 200 petaflop systems with around 10 petabytes of memory in a ‘data centric’ architectural context.
© Oil IT Journal - all rights reserved.