Dense and Sparse Matrix-Vector Multiplication on Maxwell GPUs with PyCUDA
| Autor: | José Antonio Ortega-Toro, Manuel Ujaldón, Francisco Nurudín Álvarez |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Memory hierarchy
Computer science 05 social sciences 050301 education Sparse matrix-vector multiplication Pascal (programming language) Computational science Matrix (mathematics) 0501 psychology and cognitive sciences SIMD 0503 education Texture memory computer 050104 developmental & child psychology Sparse matrix computer.programming_language |
| Zdroj: | Communications in Computer and Information Science ISBN: 9783319579719 CARLA |
| DOI: | 10.1007/978-3-319-57972-6_16 |
| Popis: | We present a study on Matrix-Vector Product operations in the Maxwell GPU generation through the PyCUDA python library. Through this lens, a broad analysis is performed over different memory management schemes. We identify the approaches that result in higher performance in current GPU generations when using dense matrices. The found guidelines are then applied to the implementation of the sparse matrix-vector product, covering structured (DIA) and unstructured (CSR) sparse matrix formats. Our experimental study on different datasets reveals that there is room for little improvement in the current state of the memory hierarchy, and that the expected Pascal GPU generation will get a major benefit from our techniques. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|