Strategies for accelerating combustion simulations with GPUs

2014-08-05T20:49:53Z (GMT) by Kyle Niemeyer Chih-Jen Sung
<p>Presented at the 35th International Symposium on Combustion, San Francisco, CA, USA. 4-8 August 2014.</p> <p> </p> <p>Combustion simulations with detailed chemical kinetics require the integration of a large number of ordinary differential equations (ODEs), with at least one ODE system per spatial location solved every time step. This task is well-suited to the massively parallel processing capabilities of graphics processing units (GPUs), where individual GPU threads concurrently integrate independent ODE systems for different spatial locations. However, the typical high-order implicit algorithms used in combustion modeling applications (e.g., VODE, LSODE) to handle stiffness involve complex logical flow that causes severe thread divergence when implemented on GPUs, thus limiting the performance. Alternate algorithms are therefore needed.<br> </p> <p>We will demonstrate that standard explicit integrators such as the fifth-order Runge–Kutta–Cash–Karp (RKCK) algorithm can be used in the case of nonstiff chemical kinetics. When implemented on GPUs, we demonstrate a performance speedup of up to 126 times over the same algorithm executed on a single CPU with a nine-species hydrogen oxidation mechanism. In the case of moderate stiffness, the second-order stabilized Runge–Kutta–Chebyshev (RKC) method can be used. The GPU-based RKC algorithm performed 64 times faster than the same algorithm executed a CPU for an ethanol oxidation mechanism with 57 species; in addition, GPU-RKC outperformed a six-core parallel CPU version of VODE by a factor of 57 with the well-known GRI-Mech 3.0 methane oxidation mechanism (53 species). However, in the case of more severe stiffness, the performance of GPU-RKC degraded below that of VODE on six CPUs, demonstrating the need for a stiff integrator appropriate for GPU acceleration.</p>