The NXP onboard flash runs at speeds up to 72 MHz – just as fast as the ARM7 core itself!  This means that the flash is “0 wait state” – the ARM7 never has to idle while waiting for memory to be fetched out of the onboard flash, assuming the next instruction is sequential.  Additionally, there are different techniques for cases where the instructions or data aren’t sequential that can accelerate the performance of the onboard flash memory.  Specifically, the NXP Memory Accelerator Module (MAM) can mitigate these negative efforts of branch instructions and data accesses. NXP chose to use MAM rather than a more traditional cache for several reasons, but primarily because embedded applications can demand “deterministic” response.  In other words, it needs to be predictable as to how fast the result will be provided to the user.  Cache performance, on the other hand, can be “lumpy” and uneven in its benefits.