As performance demands and system complexity increase, MCU/DSP processors eventually will not be able to keep up. With an FPGA, there are at least three ways to scale processing performance, and these can be implemented in any manner or combination. The first method is to use either a high-performance external processor, or one to multiple embedded processors in the FPGA. The second method is to add custom instructions in line with the processor code to accelerate specific processor instructions, such as floating point, add, multiply, subtract, and divide instructions. Finally, the third method is to accelerate data transformation with application-specific hardware like DSP blocks or video/graphics processors. Performance scaling using one or more embedded processors includes the asymmetrical multicore and the I/O companion chip processing options, as shown on this slide. In many cases when using an FPGA, the user can integrate the processing functions of the external host processor into an embedded processor on the FPGA; processor core examples for Intel® FPGA's currently include the Intel® Nios® II, Arm® Cortex®-M1, and Freescale Coldfire V1 processors. A single embedded processor may run out of performance as design complexity increases. In this case, it is possible to move to a multiprocessor, or asymmetrical processor design. Asymmetric multiprocessing means that multifunction products can have a dedicated processor for each main function, such as the main processor, image processor, communications processor, display interface processor, and so on. Smart phones are designed this way. Developers used to build systems like this with multiple MCU/DSP/graphics devices on the board. Now all the digital logic can be integrated on programmable logic, using the FPGA as the SoC. If it is necessary to keep the external host processor and the user still wants to take advantage of FPGA's benefits, the FPGA concept described above can be applied, using the FPGA as the coprocessor or I/O companion chip to the primary processor. Using the Intel® approach, the designer can combine the existing host processor and the FPGA on the same board and integrate it all into one FPGA as another option. The external host processor is used for system processing, but such a fixed function processor frequently lacks the key interfaces or functions for industrial control systems. This is where the FPGA can fill the feature gap. The processor can be offloaded by moving processing tasks inside the FPGA, including smaller processors like the Nios® II processor to partition the tasks. Also, all processor functions can be integrated in the FPGA; this is the SoC option that was mentioned earlier in the application examples.