Unlock instant, AI-driven research and patent intelligence for your innovation.

A multi-GPU fast forward modeling method for seismic data based on asynchronous flow

A technology of seismic data and forward calculation, applied in the field of geophysical exploration, can solve problems such as insufficient memory of a single GPU, insufficient computing power of a PC cluster, etc., and achieve the effect of fine large-scale data calculation, low cost, and high calculation efficiency

Active Publication Date: 2018-11-23
CHINA UNIV OF GEOSCIENCES (BEIJING)
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the problems of insufficient computing power of PC clusters and insufficient memory of a single GPU in large-scale seismic data deduction calculation requirements, the present invention provides a multi-GPU fast forward calculation method for seismic data based on asynchronous flow, which can be Guaranteed high-efficiency implementation of large data volume forward calculation, and the calculation efficiency of dual GPU can reach 2:1 compared with that of single GPU

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A multi-GPU fast forward modeling method for seismic data based on asynchronous flow
  • A multi-GPU fast forward modeling method for seismic data based on asynchronous flow
  • A multi-GPU fast forward modeling method for seismic data based on asynchronous flow

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0052] See attached Figure 3-6 , according to the above steps and methods, we use the data for testing. The tested GPU is TeslaK80, and the main frequency of the CPU is 2.6GHz. The test results are attached Figure 3-6 shown. The abscissa indicates the size of the forward modeling model, with image 3 The vertical axis of represents the calculation time. Thick lines represent the time consumed by dual-card GPUs, and thin lines represent the time consumed by single-card GPUs. attached Figure 4 The comparison chart showing the acceleration ratio, using the single card time as a reference, obtains the ratio of the calculation time of the single card and the dual card. It can be seen from the figure that the acceleration ratio of the dual card exceeds 2:1, which proves that our asynchronous stream method is effective for Efficiency improvement effect. Finally, compare the calculation and simulation results of single card and dual card. attached Figure 5 is the forward ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a seismic data multi-GPU fast forward computation method based on asynchronous flow. According to the method, seismic data are segmented according to slow dimensional domains, the number of the segmented subdomains is the same with the number of GPU, and the data of each subdomain is allocated to one GPU to be computed; bidirectional real-time communication between every two GPU is realized; and the computational process is controlled by adopting the technology of multiple asynchronous flows. With application of the method, high-efficiency implementation of big data forward computation can be guaranteed, and the ratio of the computational efficiency of double GPU to that of single GPU is enabled to be 2:1.

Description

technical field [0001] The invention belongs to the technical field of geophysical exploration, and in particular relates to a multi-GPU fast forward calculation method for seismic data based on asynchronous flow. Background technique [0002] Wave equation forward modeling plays an important role in the inversion interpretation of seismic data. The processing and interpretation of 3D seismic data requires an effective forward modeling model to verify. Simultaneously, wave equation forward modeling is the basis of reverse time migration imaging technology and full waveform inversion technology. The most commonly used wave equation forward modeling method is based on the full wave equation, such as finite difference, pseudospectral method, and finite element method, which can simulate the propagation characteristics of seismic waves in the ground and are suitable for complex media models, but the amount of calculation and memory overhead are large , especially for 3D wave e...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G01V1/28
CPCG01V1/282
Inventor 刘国峰李春
Owner CHINA UNIV OF GEOSCIENCES (BEIJING)