Platforms and marketing: What a concept
Platforms and marketing: What a concept
By Gary Smith, EE Times
September 12, 2002 (4:51 p.m. EST)
URL: http://www.eetimes.com/story/OEG20020911S0062
Platform-based design can be traced back to reference designs during the era of vacuum tubes. The idea was to show engineers how a vendor's products, in this case tubes, could be used to solve engineering problems.
Reference designs have been with us ever since. They gain popularity whenever there is a major breakthrough in technology. During the 1970s, there was a minor shift in the use of reference designs. Engineers were starting to use these designs in production, at least for the first few systems. The reference designs became a fast way of getting a product to market. If you were out first, you could grab market share and, after the design became more familiar to your engineers, you could start exploring some of your own architectures using the reference design as a guide.
As the microprocessor became part of these designs, engineers found that they could leave the reference design as is and gain competitive advantage using software rather than hardware. This was the start of the embedded board market, and the point where reference designs became platform-based designs.
In essence, a platform is a frozen architecture. Once the architecture is frozen, you may standardize the interfaces and give the engineers some choice of building blocks. The choices range from none (where you make your modifications using software) to multiple choices on multiple blocks. Supposedly, all of the interfaces have been verified for all of the various block combinations.
Obviously, the choices are very limited in the beginning, and then grow, giving the platform builder the time to verify more blocks. That is until the architecture becomes obsolete, which is the main problem with platform-based design. That's why the word "flexible" is often attached to platform-based design.
Unfortunately, flexibility is easier said than done. The be st example is the PC platform. Keep in mind that once you change the architecture, you tend to disrupt your interfaces. That calls for a major effort in verification and in setting up the new standards required to keep the platform viable. The dollar cost of Intel's engineering effort to maintain architectural compatibility a six-syllable word for disrupting as few interfaces as possible has been considerable.
Unfortunately, the hype is in full bloom. To cover up the fact that platform-based design is not the answer to all design questions, marketing has broadened the definition of platform-based design until it has become a fairly meaningless buzzword.
One clever design technique that is starting to be called platform-based design is the derivative design. The difference in a derivative design is that the initial system design was "architectured" so that it could be modified fairly rapidly, using a field programmable gate array (FPGA) or software. It is now possible to do der ivative designs on a system on chip (SoC), with a good architecture and a strict hierarchical design methodology, by simply replacing cores as you would ICs in a PC board design. Therefore, derivative design has become a powerful tool for the power user community, a far cry from the platform-based design of the mainstream community.
In looking at a system design, you need to understand the three styles of design. The SoC design style is used when your competitive advantage is in silicon design. The embedded design style is where your competitive advantage is in software development. The component-based design style is used where the markets are unable to support expensive design costs or where the real competitive advantage lies outside the electronic design itself.
The embedded design style is the obvious place for platform-based design. It can also be used in mainstream SoC design, but keep in mind that the higher up in the design flow, the greater the impact on the completed design. All major silicon design breakthroughs happen at the architectural level. If your architecture is frozen, as it is in platform-based design, your ability to use silicon as a competitive advantage is severely limited.
The bottom line
Unfortunately platforms cost an arm and a leg to design, and, especially, verify. They also take 18 months, or as long as three years, to develop and fully populate with the necessary models. With the short life expectancy of most designs the time you have to recoup your investment is severely limited. ROI is a big issue. That means the numbers of financially viable platform-based designs are fairly low. Platform-based design may be a good fit for the embedded software design style, but it is only a temporary solution for SoC designers.
Gary Smith is chief analyst, EDA, at Gartner Dataquest.
Related Articles
New Articles
- Quantum Readiness Considerations for Suppliers and Manufacturers
- A Rad Hard ASIC Design Approach: Triple Modular Redundancy (TMR)
- Early Interactive Short Isolation for Faster SoC Verification
- The Ideal Crypto Coprocessor with Root of Trust to Support Customer Complete Full Chip Evaluation: PUFcc gained SESIP and PSA Certified™ Level 3 RoT Component Certification
- Advanced Packaging and Chiplets Can Be for Everyone
Most Popular
- System Verilog Macro: A Powerful Feature for Design Verification Projects
- System Verilog Assertions Simplified
- Smart Tracking of SoC Verification Progress Using Synopsys' Hierarchical Verification Plan (HVP)
- Dynamic Memory Allocation and Fragmentation in C and C++
- Synthesis Methodology & Netlist Qualification
E-mail This Article | Printer-Friendly Page |