An Instruction Scheduling to Compensate Simple Bypassing Topologies

摘要

Bypassing path enables the result of an execute stage to be available in the execute stage of functional units processing subsequent step of pipelining as soon as the execute stage gets finished. However, complexity of bypassing path becomes subject to rapid expansion if number of functional units increases. Applying the method of Repeated Instruction Allocation, by allotting same operation on several functional units, we could have the same effect that virtual bypassing paths were provided.