Performance and Issues in Pipelining
Throughput: Throughput of a pipeline can be defined as the number of results that have been getting per unit time. It can be denoted as:
T = n / [m + (n-1)]. c = E / c
Throughput signifies the computing power of the pipeline.
Maximum speed up, throughput and efficiency are the perfect cases but these are not achieved in the useful cases, as the speedup is limited due to the many factors:
- Data dependency between successive tasks:There might be dependencies among the instructions of two tasks used in the pipeline. For example, one instruction cannot be going ahead until the previous instruction returns the results, as both are interdependent. One more instance of data dependency will be when that both instructions try to modify the similar data object. These are known as data hazards.
- Resource Constraints: When resources are not available at the time of implementation then delays are caused in pipelining. For example, if single common memory is used for both instructions and data and there is require to read/write and fetch the instruction at the similar time then only one can be carried out and the other has to stay. One more example is of limited resource like implementation unit, which may be busy at the required time.
- Branch Instructions and Interrupts in the program:A program is not a directly flow of sequential instructions. There may be branch instructions that modify the normal flow of program, which delays the pipelining implementation and affects the performance. Likewise, there are interrupts that postpones the implementation of next instruction until the interrupt has been serviced. Interrupt and the Branches have harmful effects on the pipelining.