RVS instrumentation has incredibly low execution time and memory overheads on embedded targets. By offering flexible instrumentation approaches, RVS tools make best use of the capabilities and limitations of each system.
Sometimes, instrumentation can be configured using only a single processor instruction (or even none with some target architectures). This means you can fit more tests into each build of your code, reducing the time needed to upload new tests and download data and increasing the efficiency of your test project.
For more information on the factors that govern instrumentation overheads using a fun example, see our post on the overhead of structural coverage analysis on Doom.