Software quality is directly proportional to simulation speed.
This is obvious, even intuitive, for engineers.
Faster simulations mean more tests can be run, which in turn means more bugs can be found, which results in higher quality.
Reduced schedules can be a side benefit of speed.
While this is obvious, why is it so important right now?
One example is server SoCs, where software/systems test suites can include hundreds of tests, each consisting of hundreds of billions of instructions.
If the virtual platform performance is 100 MIPS, this test suite could take over one week to run.
If the performance is five times faster, running the test suite takes 1 day; ten times faster and it runs overnight.
This simulation speed is especially interesting with the new generation of ARMv8 based server SoCs.
It is also interesting in areas such as image recognition, where hardware accelerators sit next to the CPUs on the SoC.
In this paper we discuss virtual platform simulation performance, including how to take advantage of the multiple cores on the
host PC for increased simulation speed. Examples of virtual platforms with 1) ARMv8 multicore processors and
2) hardware accelerators will be shown to illustrate how simulation speed can be accelerated.