Assessing Accelerator-Based HPC Reverse Time Migration

TLDR

Reverse Time Migration (RTM) is a critical seismic imaging technique for oil and gas exploration, valued at roughly $10^13, yet its practical adoption is limited by the enormous computational resources it demands. This study aims to pinpoint the main limitations and bottlenecks of various accelerators for RTM and to propose a set of features that future accelerator designs should include. The authors map an RTM algorithm onto an IBM Cell/B.E., an NVIDIA Tesla GPU, and an FPGA platform modeled after the Convey HC‑1. All three accelerator implementations achieve about tenfold speedups over an Intel Harpertown processor, but the gains come at the cost of substantial development effort due to immature frameworks and inadequate programming models, confirming accelerators as promising platforms for high‑order finite‑difference workloads.

Abstract

Oil and gas companies trust Reverse Time Migration (RTM), the most advanced seismic imaging technique, with crucial decisions on drilling investments. The economic value of the oil reserves that require RTM to be localized is in the order of 10^{13} dollars. But RTM requires vast computational power, which somewhat hindered its practical success. Although, accelerator-based architectures deliver enormous computational power, little attention has been devoted to assess the RTM implementations effort. The aim of this paper is to identify the major limitations imposed by different accelerators during RTM implementations, and potential bottlenecks regarding architecture features. Moreover, we suggest a wish list, that from our experience, should be included as features in the next generation of accelerators, to cope with the requirements of applications like RTM. We present an RTM algorithm mapping to the IBM Cell/B.E., NVIDIA Tesla and an FPGA platform modeled after the Convey HC-1. All three implementations outperform a traditional processor (Intel Harpertown) in terms of performance (10x), but at the cost of huge development effort, mainly due to immature development frameworks and lack of well-suited programming models. These results show that accelerators are well positioned platforms for this kind of workload. Due to the fact that our RTM implementation is based on an explicit high order finite difference scheme, some of the conclusions of this work can be extrapolated to applications with similar numerical scheme, for instance, magneto-hydrodynamics or atmospheric flow simulations.