This document reports the results of EE2 test 2.8, where the MIP matrix sizes are increased for the selected block sizes. When the proposed method is used, an L-shaped causal template is used to generate the prediction block. In test 2.8a, the prediction block generated by matrix multiplication has the same size as the current block, while in test 2.8b, for blocks sizes of 32x16, 16x32 and 32x32, a prediction block of 16x16 is generated by matrix multiplication and then up sampled to the size of the current block.