A gpgpu parallel computing method for sph fluid simulation

Abstract

The invention discloses a GPGPU (General Purpose Graphics Processing Unit) parallel computing method oriented to SPH (Smoothed Particle Hydrodynamics) fluid simulation. The method comprises the step of: grouping particle computing tasks in an SPH method to accord with a tread level structure in a GPGPU, wherein a thread in a cooperative thread array (CTA) can process logically similar work. The grouping method is developed for a CTA scheduling rule in the GPGPU and is realized on the GPGPU, thus the cache hit rate of a system can be improved while allocating tasks rapidly. An on-chip shared memory can be used for caching read data from a global memory on the basis of the allocating method; the memory bandwidth can be fully utilized according to the storage access characteristics of the global memory; and meanwhile through utilization of SIMD (Single Instruction Multiple Data) characteristics of the GPGPU, the thread synchronization overhead brought by use of the shared memory can be effectively avoided.

Description

technical field [0001] The invention belongs to the field of computer graphics and high-performance computing, specifically a new parallel method based on a general-purpose computing graphics processor (GPGPU), which is used for fluid simulation calculation based on the SPH algorithm. It involves computer graphics and simulation, computational fluid dynamics, single instruction stream multiple data stream (SIMD) architecture, GPGPU, etc. Background technique [0002] The SPH method is a grid-free Lagrangian numerical method. This method has been widely used in fluid simulation, astrodynamics, molecular dynamics and biological simulation in recent years. As a particle method, SPH requires a large number of particles to simulate large-scale scenes with high precision. And a large number of high-density particles will lead to massive computing requirements. So for users, they are always forced to make a trade-off between simulation effects and system performance. Therefore,...

AI Technical Summary

Problems solved by technology

However, existing shared memory solutions do not bring the desired performance improvement, and are even 20% to 70% slower than traditional methods

This solution uses the explicit thread synchronization method in CTA to ensure the correctness of data, but thread synchronization brings very high performance overhead; at the same time, this method occupies too much shared memory, making it run on most devices It has a low device occupancy rate; secondly, its inefficient global memory access mode does not make good use of the access characteristics of global memory; and its random task allocation strategy will lead to a lower cache hit rate

[0004] All in all, although the performance of SPH has been significantly improved by taking advantage of the high parallel processing capability of GPGPU devices, its potential is still far from being fully exploited.

Images