answer the following based on the figure below: How many forwarding paths exist in the pipeline? To reduce the branch penalty, now assume the branch decision point( target and directions computation) is moved into the ID stage. In the presence of a control flow dependency between two subsequent operations where the first one is a branch and the second one is an arithmetic operation how many NOPs ( bubbles) need to be inserted between the 2 instructions?

To reduce the branch penalty, now assume the branch decision point (target and directions computation) is moved into the ID stage. How many additional forwarding paths need to be added, to eliminate as many stalls as possible for branches depending on prior instructions?

Consider the 8 -stage in-order pipeline shown in the Figure below. The pipeline is identical to the 5-stage pipeline covered in class except for splitting the MEM stage into four MEM stages (to improve its timing). Only MEM3 and EX is a producer of data, memory addresses are still computed in EX. There exist forwarding paths from all sources to all consumers. Branch decision and branch targets are computed in the EX stage. The register file allows writing and reading of a register in a single cycle. Assume the 2-source. 1 -destination operands-as in the RISC-V ISA. The pipeline does not support branch prediction. Answer the questions below