Accurate cross-architecture performance modeling for sparse matrix-vector multiplication (SpMV) on GPUs

摘要

This paper presents an integrated analytical and profile-based cross-architecture performance modeling toolrnto specifically provide inter-architecture performance prediction for Sparse Matrix-Vector Multiplicationrn(SpMV) on NVIDIA GPU architectures. To design and construct the tool, we investigate the interarchitecturernrelative performance for multiple SpMV kernels. For a sparse matrix, based on its SpMVrnkernel performance measured on a reference architecture, our cross-architecture performance modeling toolrncan accurately predict its SpMV kernel performance on a target architecture. The prediction results canrneffectively assist researchers in making choice of an appropriate architecture that best fits their needs from arnwide range of available computing architectures. We evaluate our tool with 14 widely-used sparse matricesrnon four GPU architectures: NVIDIA Tesla C2050, Tesla M2090, Tesla K20m, and GeForce GTX 295. Inrnour experiments, Tesla C2050 works as the reference architecture, the other three are used as the targetrnarchitectures. For Tesla M2090, the average performance differences between the predicted and measuredrnSpMV kernel execution times for CSR, ELL, COO, and HYB SpMV kernels are 3:1%, 5:1%, 1:6%, andrn5:6%, respectively. For Tesla K20m, they are 6:9%, 5:9%, 4:0%, and 6:6% on the average, respectively. ForrnGeForce GTX 295, they are 5:9%, 5:8%, 3:8%, and 5:9% on the average, respectively.