Sparse Matrix-vector Multiplication on GPGPU Clusters: A New Storage Format and a Scalable Implementation

Sparse matrix-vector multiplication (spMVM) is the dominant operation in many sparse solvers. We investigate performance properties of spMVM with matrices of various sparsity patterns on the nVidia "Fermi" class of GPGPUs. A new "padded jagged diagonals storage" (pJDS) format is...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2012 26th IEEE International Parallel and Distributed Processing Symposium Workshops s. 1696 - 1702
Hlavní autoři: Kreutzer, M., Hager, G., Wellein, G., Fehske, H., Basermann, A., Bishop, A. R.
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.05.2012
Témata:
Bandwidth
Computational modeling
CUDA
Error correction codes
GPGPU
Instruction sets
Kernel
Sparse matrices
Vectors
ISBN:1467309745, 9781467309745
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Sparse matrix-vector multiplication (spMVM) is the dominant operation in many sparse solvers. We investigate performance properties of spMVM with matrices of various sparsity patterns on the nVidia "Fermi" class of GPGPUs. A new "padded jagged diagonals storage" (pJDS) format is proposed which may substantially reduce the memory overhead intrinsic to the widespread ELLPACK-R scheme while making no assumptions about the matrix structure. In our test scenarios the pJDS format cuts the overall spMVM memory footprint on the GPGPU by up to 70%, and achieves 91% to 130% of the ELLPACK-R performance. Using a suitable performance model we identify performance bottlenecks on the node level that invalidate some types of matrix structures for efficient multi-GPGPU parallelization. For appropriate sparsity patterns we extend previous work on distributed-memory parallel spMVM to demonstrate a scalable hybrid MPI-GPGPU code, achieving efficient overlap of communication and computation.
ISBN:1467309745
9781467309745
DOI:10.1109/IPDPSW.2012.211