Sparse linear algebra operations are very important across many areas of science. Therefore, efficiently finding ways to perform these matrix computations is essential. The rapid development over the past few years in heterogeneous computing with or including accelerators has made finding new ways to process these operations important and challenging. However, the primary focus of this goal is through sparse matrix vector (SpMV) operations or sparse-sparse matrix operations (SpGEMM). Little attention is paid to sparse-dense matrix multiplication (SpMM), which is also used in important areas. In this paper, we propose to conduct a performance analysis of sparse matrix formats, specifically COO, CSR, ELLPACK, and BCSR. We aim to study these formats in SpMM operations in various conditions and provide a corresponding performance analysis. To facilitate this, we first propose a set of benchmarks for this evaluation. We then conduct a study of the formats themselves to provide an understanding of their behavior in various conditions.