Time‐Domain Parallelization for Accelerating Cloth Simulation

Abstract

Cloth simulations, widely used in computer animation and apparel design, can be computationally expensive for real‐time applications. Some parallelization techniques have been proposed for visual simulation of cloth using CPU or GPU clusters and often rely on parallelization using spatial domain decomposition techniques that have a large communication overhead. In this paper, we propose a novel time‐domain parallelization technique that makes use of the two‐level mesh representation to resolve the time‐dependency issue and develop a practical algorithm to smooth the state transition from the corresponding coarse to fine meshes. A load estimation and a load balancing technique used in online partitioning are also proposed to maximize the performance acceleration. Our method achieves a nearly linear performance scaling on manycore clusters and outperforms spatial‐domain parallelization on a diverse set of benchmarks.

P-Cloth: Interactive Complex Cloth Simulation on Multi-GPU Systems using Dynamic Matrix Assembly and Pipelined Implicit Integrators

“…Comparing to the parallel cloth simulation algorithms on a CPU cluster [Liang and Lin 2018;Ni et al 2015;Selle et al 2009] or a hybrid combination of CPU and GPU [Pabst et al 2010], our multi-GPU based algorithm is about 10X faster on performance for complex benchmarks. Recent cloth simulation algorithm proposed by takes about 2 minutes per frame for complex benchmarks with 1.4 − 1.8M triangles on an Intel Xeon system with multiple CPU cores.…”

“…up to a few hundred thousand triangles). Other techniques have been proposed to parallelize the cloth simulation on a CPU cluster [Liang and Lin 2018;Ni et al 2015;Selle et al 2009] or a hybrid combination of CPU and GPU [Pabst et al 2010]. Our approach for parallel cloth simulation is designed to exploit the computational capability and high memory bandwidth of GPUs and is complementary to prior parallel simulation algorithms.…”

“…Many parallel techniques have also been proposed to utilize a large number of CPUs on a cluster [Liang and Lin 2018;Ni et al 2015;Selle et al 2009]. However, current CPUs have a lower memory bandwidth and a smaller number of cores than commodity GPUs.…”