AN2367 Datasheet PDF - STMicroelectronics

Introduction
   This application note describes an optimized memory copy for TCP/IP. It can also be used for any memory copy. This routine is written in ARM assembler, and can be called by any C program which conforms to the ARM procedure call standard.
   Using this optimized routine will accelerate the memory copy up to 20 times as compared to a standard implementation. This optimized buffer alignment routine has been integrated in the HAL Ethernet driver distributed by ST.
   This program has been fully tested and run inside a TCP/IP stack. However, STMicroelectronics cannot guarantee that this driver is flawless in all applications.