MuL is designed for flexibility. Towards the south it can support
multiple south-bound api. Whereas towards the north, it interfaces
to applications via a ML-API and RESTful NB-API. Applications using
ML-API has the option of being hosted inline the same process
address space as MuL core to execute in-context and gain maximum
performance.With the SDKs that come with MuL, the app developer or
the app itself need not know how and where they are running.
The multi-threaded core is optimized using subtle cache-aware
programming to get the max mileage especially from Intel CPUs.