Fuse lui instruction and addi instruction as a single macro-operation in the decode stage of Rocket Core CPU. The rest of stages see them as a signle macro-operation. That means number of instructions to be executed is reduced and performance is increased.
How to use:
- Set up Chipyard and simulate the default example.
- Run command
make clean - Relace the original RocketCore.scala file in Chipyard with this one.
- Run command
make - Run benchmarks.