Download Now Free registration required
Parallel 3D FFT is a commonly used numerical method in scientific computing. P3DFFT is a recently proposed implementation of parallel 3D FFT that is designed to allow scalability to massively large systems such as Blue Gene. While there has been recent work that demonstrates such scalability on regular cartesian meshes (equal length in each dimension), its performance and scalability for flat cartesian meshes (much smaller length in one dimension) is still a concern. In this paper, the authors perform studies on a 16-rack (16384-node) Blue Gene/L system that demonstrates that a combination of the network topology and the communication pattern of P3DFFT can result in early network saturation and consequently performance loss.
- Format: PDF
- Size: 243.5 KB