What is the shortcoming of using MIPS or FLOPS as a measure

Solution

MIPS - million instructions per second. This metric simply counts how many instructions a processor can execute each second.

MIPS does not reflect how much work is done per instruction. For example, RISC processors have simple instructions, and even though they can execute instructions faster than CISC processors, they do not necessarily do more work.

Different instructions can take a different amount of time to execute. For example, add instructions execute faster on most processors than branch instructions. Most MIPS ratings are based on the fastest executing instruction and should be considered as Peak-MIPS, while MIPS rating under \"normal\" use will be lower.

It is problematic to compare different processor architectures as a processor with a single very fast instruction and other slow instructions will have a higher MIPS rating than processors with all fast instructions, although the latter will execute most programs faster.

The actual rate of processing depends significantly of the internal circuitry of the processor. It is possible for a processor with a higher clock rate to process Instructions slower than one with a lower rate.

Early processors did not have a floating point unit and had used integer unit to perform floating point operations which was very slow. Because of that, FLOPS were much lower than MIPS. Current processors usually have multiple floating point units to efficiently pipeline floating point operations. FLOPS are a decent measure of a processor\'s speed, but do not reflect the computers performance as it does not account for I/O or memory performance.