The first thing you want to optimize is the lwip_standard_checksum() routine from src/core/inet.c. You can override this standard function with the #define LWIP_CHKSUM your_checksum_routine().

There are C examples given in inet.c or you might want to craft an assembly function for this. RFC1071 is a good introduction to this subject.

Other significant improvements can be made by supplying assembly or inline replacements for htons() and htonl() if you're using a little-endian architecture. #define lwip_htons(x) your_htons() #define lwip_htonl(x) your_htonl() If you #define them to htons() and htonl(), you should #define LWIP_DONT_PROVIDE_BYTEORDER_FUNCTIONS to prevent lwIP from defining htonx / ntohx compatibility macros.

Check your network interface driver if it reads at a higher speed than the maximum wire-speed. If the hardware isn't serviced frequently and fast enough buffer overflows are likely to occur.

E.g. when using the cs8900 driver, call cs8900if_service(ethif) as frequently as possible. When using an RTOS let the cs8900 interrupt wake a high priority task that services your driver using a binary semaphore or event flag. Some drivers might allow additional tuning to match your application and network.

For a production release it is recommended to set LWIP_STATS to 0. Note that speed performance isn't influenced much by simply setting high values to the memory options.