Nucleus NET contains various areas that can be tweaked to help improve throughput. Nucleus NET will send data at the priority of the calling task and will receive data at the priority defined by EV_PRIORITY and TM_PRIORITY in net_cfg.h. Increasing these priorities should provide the most improvement on systems with a large number of tasks. It is important to keep EV_PRIORITY and TM_PRIORITY at the same level relative to each other. It is also important that no task using NET API's be set at a higher priority at than the afore mentioned defines.
Chapter 16 - Optimizing Nucleus NET of the Nucleus NET Reference Manual discusses other areas that can be tweaked to improve performance. Two of the optimizations mentioned in this chapter, The NU_Block_Copy, and Checksum optimizations can be found in TechNotes MG44758 and MG44810 respectively.
Please note that the TCP_Resource locking semaphore is a commonly misunderstood design feature of NET that is thought to cause bottlenecks. The locking semaphore is crucial to protecting internal NET data structures to allow multiple users. Every effort has been made to ensure the semaphore is held only during global data structure access to protect the integrity of global variables and ensure the integrity of the stack. This design feature should provide little if any performance degradation and should not be tampered with in any way.