Strassen's Matrix Multiplication on GPUs

We provide efficient single-precision and integer GPU implementations of Strassen's algorithm as well as of Winograd's variant. On an NVIDIA C1060 GPU, a speedup of 32% (35%) is obtained for Strassen's 4-level implementation and 33% (36%) for Winograd's variant relative to the sg...

Full description

Saved in:
Bibliographic Details
Published in2011 IEEE 17th International Conference on Parallel and Distributed Systems pp. 157 - 164
Main Authors Junjie Li, Ranka, S., Sahni, S.
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2011
Subjects
accuracy
Complexity theory
CUDA
GPU
Graphics processing unit
Kernel
Matrix decomposition
matrix multiplication
Memory management
Strassen's algorithm
Vectors
Winograd's variant
Online AccessGet full text
ISBN1457718758
9781457718755
ISSN1521-9097
DOI10.1109/ICPADS.2011.130

Cover

More Information
Summary:We provide efficient single-precision and integer GPU implementations of Strassen's algorithm as well as of Winograd's variant. On an NVIDIA C1060 GPU, a speedup of 32% (35%) is obtained for Strassen's 4-level implementation and 33% (36%) for Winograd's variant relative to the sgemm (integer version of sgemm) code in CUBLAS 3.0 when multiplying 16384×16384 matrices. The maximum numerical error for the single-precision implementations is about 2 orders of magnitude higher than those for sgemm when n = 16384 and is zero for the integer implementations.
ISBN:1457718758
9781457718755
ISSN:1521-9097
DOI:10.1109/ICPADS.2011.130