Fast seismic modeling and reverse time migration on a graphics processing unit cluster

2011 ◽  
Vol 24 (7) ◽  
pp. 739-750 ◽  
Author(s):  
Rached Abdelkhalek ◽  
Henri Calandra ◽  
Olivier Coulaud ◽  
Guillaume Latu ◽  
Jean Roman
Keyword(s):  
Graphics Processing Unit ◽  
Processing Unit ◽  
Seismic Modeling ◽  
Reverse Time ◽  
Reverse Time Migration ◽  
Time Migration ◽  
Graphics Processing

Leveraging the accelerated processing units for seismic imaging: A performance and power efficiency comparison against CPUs and GPUs

2017 ◽  
Vol 32 (6) ◽  
pp. 819-837 ◽  
Author(s):  
Issam Said ◽  
Pierre Fortin ◽  
Jean–Luc Lamotte ◽  
Henri Calandra
Keyword(s):  
Graphics Processing Units ◽  
Power Efficiency ◽  
Seismic Imaging ◽  
Graphics Processing Unit ◽  
Processing Unit ◽  
Reverse Time ◽  
Reverse Time Migration ◽  
Single Node ◽  
Time Migration ◽  
Graphics Processing

Oil and gas companies rely on high performance computing to process seismic imaging algorithms such as reverse time migration. Graphics processing units are used to accelerate reverse time migration, but these deployments suffer from limitations such as the lack of high graphics processing unit memory capacity, frequent CPU-GPU communications that may be bottlenecked by the PCI bus transfer rate, and high power consumptions. Recently, AMD has launched the Accelerated Processing Unit (APU): a processor that merges a CPU and a graphics processing unit on the same die featuring a unified CPU-GPU memory. In this paper, we explore how efficiently may the APU be applicable to reverse time migration. Using OpenCL (along with MPI and OpenMP), a CPU/APU/GPU comparative study is conducted on a single node for the 3D acoustic reverse time migration, and then extended on up to 16 nodes. We show the relevance of overlapping the I/O and MPI communications with the computations for the APU and graphics processing unit clusters, that performance results of APUs range between those of CPUs and those of graphics processing units, and that the APU power efficiency is greater than or equal to the graphics processing unit one.

CuQ-RTM: A CUDA-based code package for stable and efficient Q-compensated reverse time migration

Geophysics ◽  
2019 ◽  
Vol 84 (1) ◽  
pp. F1-F15 ◽  
Author(s):  
Yufeng Wang ◽  
Hui Zhou ◽  
Xuebin Zhao ◽  
Qingchen Zhang ◽  
Poru Zhao ◽  
...  
Keyword(s):  
Large Scale ◽  
Gpu Computing ◽  
Graphics Processing Unit ◽  
Processing Unit ◽  
Program Optimization ◽  
Adaptive Stabilization ◽  
Reverse Time ◽  
Reverse Time Migration ◽  
Data Set ◽  
Time Migration

Reverse time migration (RTM) in attenuating media should take absorption and dispersion effects into consideration. The latest proposed viscoacoustic wave equation with decoupled fractional Laplacians facilitates separate amplitude compensation and phase correction in [Formula: see text]-compensated RTM ([Formula: see text]-RTM). However, intensive computation and enormous storage requirements of [Formula: see text]-RTM prevent it from being extended into practical application, especially for large-scale 2D or 3D cases. The emerging graphics processing unit (GPU) computing technology, built around a scalable array of multithreaded streaming multiprocessors, presents an opportunity for greatly accelerating [Formula: see text]-RTM by appropriately exploiting GPUs architectural characteristics. We have developed the cu[Formula: see text]-RTM, a CUDA-based code package that implements [Formula: see text]-RTM based on a set of stable and efficient strategies, such as streamed CUDA fast Fourier transform, checkpointing-assisted time-reversal reconstruction, and adaptive stabilization. The cu[Formula: see text]-RTM code package can run in a multilevel parallelism fashion, either synchronously or asynchronously, to take advantages of all the CPUs and GPUs available, while maintaining impressively good stability and flexibility. We mainly outline the architecture of the cu[Formula: see text]-RTM code package and some program optimization schemes. The speedup ratio on a single GeForce GTX760 GPU card relative to a single core of Intel Core i5-4460 CPU can reach greater than 80 in a large-scale simulation. The strong scaling property of multi-GPU parallelism is demonstrated by performing [Formula: see text]-RTM on a Marmousi model with one to six GPU(s) involved. Finally, we further verified the feasibility and efficiency of the cu[Formula: see text]-RTM on a field data set.

An efficient local operator-based Q-compensated reverse time migration algorithm with multistage optimization

Geophysics ◽  
2018 ◽  
Vol 83 (3) ◽  
pp. S249-S259 ◽  
Author(s):  
Tong Zhou ◽  
Wenyi Hu ◽  
Jieyuan Ning
Keyword(s):  
Seismic Attenuation ◽  
Depth Information ◽  
Frequency Noise ◽  
Processing Unit ◽  
Reverse Time ◽  
Reverse Time Migration ◽  
Pass Filter ◽  
Low Pass Filter ◽  
Time Migration ◽  
Imaging Results

Most existing [Formula: see text]-compensated reverse time migration ([Formula: see text]-RTM) algorithms are based on pseudospectral methods. Because of the global nature of pseudospectral operators, these methods are not ideal for efficient parallelization, implying that they may suffer from high computational cost and inefficient memory usage for large-scale industrial problems. In this work, we reported a novel [Formula: see text]-RTM algorithm — the multistage optimized [Formula: see text]-RTM method. This [Formula: see text]-RTM algorithm uses a finite-difference method to compensate the amplitude and the phase simultaneously by uniquely combining two techniques: (1) a negative [Formula: see text] method for amplitude compensation and (2) a multistage dispersion optimization technique for phase correction. To prevent high-frequency noise from growing exponentially and ruining the imaging results, we apply a finite impulse response low-pass filter using the Kaiser window. The theoretical analyses and numerical experiments demonstrate that this [Formula: see text]-RTM algorithm precisely recovers the decayed amplitude and corrects the distorted phase caused by seismic attenuation effects, and hence produces higher resolution subsurface images with the correct structural depth information. This new method performs best in the frequency range of 10–70 Hz. Compared with pseudospectral [Formula: see text]-RTM methods, this [Formula: see text]-RTM approach offers nearly identical imaging quality. Based on local numerical differential operators, this [Formula: see text]-RTM method is very suitable for parallel computing and graphic processing unit implementation, an important feature for large 3D seismic surveys.

Reverse Time Migration Using the Pseudospectral Time-Domain Algorithm

2016 ◽  
Vol 24 (02) ◽  
pp. 1650005 ◽  
Author(s):  
Jiangang Xie ◽  
Zichao Guo ◽  
Hai Liu ◽  
Qing Huo Liu
Keyword(s):  
Time Domain ◽  
Acoustic Waves ◽  
Fdtd Method ◽  
Pseudospectral Method ◽  
Imaging Method ◽  
Processing Unit ◽  
Reverse Time ◽  
Reverse Time Migration ◽  
Eighth Order ◽  
Time Migration

We propose a pre-stack reverse time migration (RTM) seismic imaging method using the pseudospectral time-domain (PSTD) algorithm. Traditional pseudospectral method uses the fast Fourier transform (FFT) algorithm to calculate the spatial derivatives, but is limited by the wraparound effect due to the periodicity assumed in the FFT. The PSTD algorithm combines the pseudospectral method with a perfectly matched layer (PML) for acoustic waves. PML is a highly effective absorbing boundary condition that can eliminate the wraparound effect. It enables a wide application of the pseudospectral method to complex models. RTM based on the PSTD algorithm has advantages in the computational efficiency compared to traditional methods such as the second-order and high order finite difference time-domain (FDTD) methods. In this work, we implement the PSTD algorithm for acoustic wave equation based RTM. By applying the PSTD-RTM method to various seismic models and comparing it with RTM based on the eighth-order FDTD method, we find that PSTD-RTM method has better performance and saves more than 50% memory. The method is suitable for parallel computation, and has been accelerated by general purpose graphics processing unit.

scholarly journals A causal imaging condition for reverse time migration using the Discrete Hilbert transform and its efficient implementation on GPU

2019 ◽  
Vol 16 (5) ◽  
pp. 894-912
Author(s):  
Feipeng Li ◽  
Jinghuai Gao ◽  
Zhaoqi Gao ◽  
Xiudi Jiang ◽  
Wenbo Sun
Keyword(s):  
Image Quality ◽  
Hilbert Transform ◽  
Real Data ◽  
Processing Unit ◽  
Reverse Time ◽  
Reverse Time Migration ◽  
Discrete Hilbert Transform ◽  
Imaging Condition ◽  
Time Migration ◽  
Wavefield Decomposition

Abstract Reverse time migration (RTM) has shown a significant advantage over other imaging algorithms for imaging complex subsurface structures. However, low-wavenumber noise severely contaminates the image, which is one of the main issues in the RTM algorithm. To attenuate the undesired low-wavenumber noise, the causal imaging condition based on wavefield decomposition has been proposed. First, wavefield decompositions are performed to separate the wavefields as up-going and down-going wave components, respectively. Then, to preserve causality, it constructs images by correlating wave components that propagate in different directions. We build a causal imaging condition in this paper. Not only does it consider the up/down wavefield decomposition, but it also applies the decomposition on the horizontal direction to enhance the image quality especially for steeply dipping structures. The wavefield decomposition is conventionally achieved by the frequency-wavenumber (F-K) transform that is very computationally intensive compared with the wave propagation process of the RTM algorithm. To improve the efficiency of the algorithm, we propose a fast implementation to perform wavefield separation using the discrete Hilbert transform via the Graphics Processing Unit. Numerical tests on both the synthetic models and a real data example demonstrate the effectiveness of the proposed method and the efficiency of the optimized implementation scheme. This new imaging condition shows its ability to produce high image quality when applied to both the RTM stack image and also the angle domain common image gathers. The comparison of the total elapsed time for different methods verifies the efficiency of the optimized algorithm.

Sign in / Sign up

Export Citation Format

Share Document