DocumentCode :
3740630
Title :
Structural Agnostic SpMV: Adapting CSR-Adaptive for Irregular Matrices
Author :
Mayank Daga;Joseph L. Greathouse
Author_Institution :
AMD Res., Adv. Micro Devices, Inc., USA
fYear :
2015
Firstpage :
64
Lastpage :
74
Abstract :
Sparse matrix vector multiplication (SpMV) is an important linear algebra primitive. Recent research has focused on improving the performance of SpMV on GPUs when using compressed sparse row (CSR), the most frequently used matrix storage format on CPUs. Efficient CSR-based SpMV obviates the need for other GPU-specific storage formats, thereby saving runtime and storage overheads. However, existing CSR-based SpMV algorithms on GPUs perform poorly on irregular sparse matrices, limiting their usefulness. We propose a novel approach for SpMV on GPUs which works well for both regular and irregular matrices while keeping the CSR format intact. We start with CSR-Adaptive, which dynamically chooses between two SpMV algorithms depending on the length of each row. We then add a series of performance improvements, such as a more efficient reduction technique. Finally, we add a third algorithm which uses multiple parallel execution units when operating on irregular matrices with very long rows. Our implementation dynamically assigns the best algorithm to sets of rows in order to ensure that the GPU is efficiently utilized. We effectively double the performance of CSR-Adaptive, which had previously demonstrated better performance than algorithms that use other storage formats. In addition, our implementation is 36% faster than CSR5, the current state of the art for SpMV on GPUs.
Keywords :
"Sparse matrices","Heuristic algorithms","Instruction sets","Graphics processing units","Libraries","Arrays"
Publisher :
ieee
Conference_Titel :
High Performance Computing (HiPC), 2015 IEEE 22nd International Conference on
Type :
conf
DOI :
10.1109/HiPC.2015.55
Filename :
7397620
Link To Document :
بازگشت