講題：Introduction to Parallel Computation on a GPU Supercomputer

時間：9/26(二)下午 13：30

地點：綜合三館837室

摘要：High efficient parallel computation on GPUs is one of the challenging topics of high performance computing over the last few years. Many methods which have risen to popularity in recent times are not well suited to GPU computation. Parallel computing method has evolved from MIMD-type (multiple instruction multiple data) to SIMD-type (single instruction, multiple data) vectorized processing on a huge vector computer to processing (shared or distributed memory) recently. The high efficient computing method requests a new design philosophy to rewrite the existing algorithm to the new one with SIMD-type instruction. The speech consists of the two major parts such as the hardware design of GPU architecture and the programming model of CUDA language. We will show several demonstrations including computational fluid dynamics, finance engineering problem on a GPU supercomputer.

＜＜歡迎老師、同學踴躍參與！！＞＞