The aim of this paper is to study the performance portability of OpenACC and SYCL implementations of sparse matrix-vector product for CSR and BSR storage formats on Intel CPU and NVIDIA GPU platforms. Using the reformulated performance portability metric we show how it changes for various sparse matrices and which implementation and format achieves better performance portability. Numerical experiments show that on CPU for CSR and BSR and on GPU for BSR, OpenACC is better for smaller matrices. On GPU the SYCL implementation for CSR allows to achieve better performance portability almost in all cases.