Myths about DSP versus ASIC have been around for a long time. And, we don’t talk much about ASIP’s. TI has sold application-specific instruction set processors (ASIP’s) for a long time; but we haven’t actually called them that nor have we articulated the benefits of these. Over time, the perception has been that DSP’s are more expensive and power hungry compared to ASIC devices. We should put to rest some misperceptions. First, an ASIC device generally supports only one application or standard. It usually has an embedded processor to control and manage various ASIC IP blocks. The ASIC device is optimized by reducing bit precision to match exactly the computational requirements, and there are hardware architectures (IP’s) developed to reduce clock frequencies, etc. with the result being small die area and low power. An ASIP, on the other hand, adds special instructions to the embedded processor to accelerate computations, such as an OFDM modem; and the result is a much smaller program memory footprint and reduced processor clock frequency, resulting in a combination of lower power and lower cost. We also add function-specific accelerators to our DSPs such as for video and audio compression and image processing. The cost-performance and power of these types of devices can intersect with the ASIC’s via proper instruction and accelerator selection. However, the ASIP retains the full flexibility of a software programmable device; while the ASIC is fixed function. In the case where standards are immature and evolving, the ASIP provides a means to accelerate deployment of new standards, one year ahead or more compared to ASIC solutions. The TI TMS320F28069 is an OFDM MCU that supports a variety of narrowband OFDM PLC standards such as IEEE P1901.2, G3, PRIME, and other smart grid applications.
