根据查询速度从R中存储矩阵的最佳数据库设计

I have hundreds of matrices that need to be used in R and most of them are around 45000x350 each. What I'd like to do is find an optimal database software choice and schema to store the data in and be able to call subsets of the matrices from the database. This needs to be done with the object of extracting the data be as fast as possible.I have hundreds of matrices that need to be use