Overview
MUBLAS-GEMV is an optimized implementation of GEMV for NVIDIA GPUs. This implementation automatically adjusts the thread-block size based on the theoretical performance model before launching kernel. This code will be integrated into ASPEN.K2 in the near future.
Download
| MUBLAS version 1.5.38 | source code | |
| MUBLAS version 1.5.31 | source code | |
| MUBLAS version 1.5.24 | source code | |
| MUBLAS version 1.5.14 | source code | 1772525 byte |
| MUBLAS version 1.4.28 | source code | 11676498 byte |
| MUBLAS-GEMV version 1.3.1 | source code | 487640 byte |
| MUBLAS-GEMV version 1.3 | source code | 29466 byte |