please answer quickly is possible

1. Consider a fully-connected network with 120 inputs, two hidden layers with 256 units each, and 10 output classes. Assume that all weights and activations are stored as 8-bit values. A. Find the total parameter storage required in bytes. B. How many MACs (multiply-and-accumulate) are required to run one inference with this model? C. How much temporary storage (SRAM) is required to run this model? Remember you'll need to store the outputs from one layer and the outputs from the next layer at the same time. The consecutive pair of layers with the largest combined requirements sets the amount of SRAM required