Commentary: The Need to Unify Embedded Software and Hardware Design

Grant Martin, Chief Scientist, Tensilica and Co-Chair, DAC 2005 Technical Program Committee -- EDN, 3/9/2005

Embedded software and hardware can no longer be segregated in different sectors of the design world.  They need to come together.  To make this possible a more unified design flow is needed.  This new approach may come from the traditional hardware EDA tools vendors, the embedded software tools vendors, new system level design tools companies, or quite possibly the intellectual property (IP) companies.

EDA has traditionally focused only on the hardware part of the design process, and the needs of hardware designers, so it naturally has weak links to the embedded software world.  Similarly, embedded software tools and vendors have tended to deal only with the purer software side of the business.  The forays into more ‘system’ design tools from both EDA and embedded software tool vendors have failed in the past, partly because they took a parochial view (either hardware- or software-centric) of who was a system designer and the scope of their concerns.

With the advent of system-on-chip and many more deeply embedded electronic products, the traditional barriers between EDA and embedded software need to come down.  System architects need to understand the impacts on embedded software of their hardware choices.  Hardware designers find that deeply embedded systems rely on “SW-aware” hardware design choices.  Embedded software designers who are dealing with SOC platforms with multiple processors, on-chip buses and hardware accelerators, all of which interact directly with their software, must have a much deeper knowledge of the underlying hardware platform to develop reliable and both timing and functionally correct real time software - and the old abstraction approaches are breaking down.

But linking these two industries demands whole new ways of thinking and new ways of developing products and organizing development teams.  Given the natural resistance to change, integrating these separate design segments won’t happen overnight.  It will take time, effort, and more than a little bit of strong advocacy and management to make it happen.

More important than the detailed bridges that need to be built, as an industry we need a few visionary bridge-builders.  To date, I don't see enough agreement on the new design ideology required to realize this vision of embedded system design.  But there are hopeful signs emerging of some true vision.

One hopeful industry development is the emergence of much more complex target architectures - often called “Multi-Processor System-on-Chip (MPSoC)”.  Here the emphasis is on finding the right types of programmable processors - often heterogeneous, tailored to different parts of the application processing - and embedding them in a complex on-chip communications and memory architecture.  Successful design of MPSoC also requires finding the right programming models for tasks on processors and inter-processor communication.

With the rise of MPSoC, functions which used to be implemented in hardware only are now being looked at again, with the possibility of implementing them using application-specific configurable and extensible processors.  This means hardware and software people in the embedded product domain must share their expertise to explore the expanded design space offered by these new implementation technologies.  Exploring this design space is best done at the system and software levels.

At the Design Automation Conference, we are increasing our efforts to make sure the conference and exhibit and related activities appeal to the EDA, hardware design, embedded systems and embedded software design communities.  This year’s technical program will feature some very interesting embedded sessions. 

It will ultimately rest with customers - designers willing to try new approaches - to validate and grow those new solutions which provide real design value.  A good walk around the DAC 2005 exhibit floor could uncover a new approach that will be the foundation for the next generation of design.