SUMMARYThe CUDA model for graphics processing units (GPUs) presents the programmer with a plethora of different programming options. These includes different memory types, different memory access methods and different data types. Identifying which options to use and when is a non-trivial exercise. This paper explores the effect of these different options on the performance of a routine that evaluates sparse matrixvector products (SpMV) across three different generations of NVIDIA GPU hardware. A process for analysing performance and selecting the subset of implementations that perform best is proposed. The potential for mapping sparse matrix attributes to optimal CUDA SpMV implementations is discussed.

Abstract. NVIDIA have released a new platform (CUDA) for general purpose computing on their graphical processing units (GPU). This paper evaluates use of this platform for statistical machine learning applications. The transfer rates to and from the GPU are measured, as is the performance of matrix vector operations on the GPU. An implementation of a sparse matrix vector product on the GPU is outlined and evaluated. Performance comparisons are made with the host processor.

The CUDA model for GPUs presents the programmer with a plethora of different programming options. These includes different memory types, different memory access methods, and different data types. Identifying which options to use and when is a non-trivial exercise. This paper explores the effect of these different options on the performance of a routine that evaluates sparse matrix vector products. A process for analysing performance and selecting the subset of implementations that perform best is proposed. The potential for mapping sparse matrix attributes to optimal CUDA sparse matrix vector product implementation is discussed.

No abstract