Clocks per Instructions Retired - CPI
Equation: CPU_CLK_UNHALTED.CORE / INST_RETIRED.ANY

Category: Basic Performance Tuning Ratios

Definition: High CPI indicates that instructions are requiring more cycles to execute than they should. In this case there may be opportunities to modify your code to improve the efficiency with which instructions are executed within the processor. The CPI can get as low as 0.25 cycles per instructions.When the front-end cannot provide instructions and micro-ops at a rate high enough to fill the Reservation Station it is likely that the execution engine is starved waiting for micro-ops to execute and the measured CPI is high.High number of L0 DTLB misses indicates that the data set that the workload uses spans a number of pages that is bigger than the L0 DTLB. The high number of misses is expected to impact workload performance only if the CPI is low - around 0.8. Otherwise it is likely that the L0 DTLB miss cycles are hidden by other latencies.