Slide 33 of 46
As mentioned earlier, the notion of agile computing represents an important long-term direction for silicon, one unlikely to receive R&D attention from industry, and yet of particular benefit to the military.
In a COTS silicon world, MIL-SOC requires a methodology that facilitates the introduction of new and diverse forms of IP, both in silicon and in related peripheral technologies (e.g. sensors, optical). Unlike commercial SOC, where for reasons outlined above it is likely the IP blocks will eventually be tied to specific silicon access providers (e.g. design houses), MIL-SOC will require the tools, methodologies, and standards needed to encapsulate, abstract and model, and integrate additional non-commercial forms of IP.
Once the capabilities are there to build such IP, MIL-SOC needs specific forms of IP built for it (e.g. high-performance mixed-signal RF or A/D in a low-voltage, digital COTS-MOS technology). Research is required to make this possible. Software abstractions (APIs) are also essential here.
Finally, without an efficient path to design verification, validation, and implementation from system specification, the military will not be able to develop reliable MIL-SOC systems in a timely and cost-effective way. We need design systems that will support design at high levels of abstraction (well above todays RTL levels), support verification and validation at that level, and provide an efficient (compileable) path to MIL-SOC.