What Will Change In Design For 2015?

This year more than 26 people provided predictions for 2015. Most of these came from the EDA industry, so the results may be rather biased. However, ecosystems are coming closer together in many parts of the semiconductor food chain, meaning that the EDA companies often can see what is happening in dependent industries and in the system design houses. Thus their predictions may have already resulted in their investment in tool development.

The predictions are segregated into four areas: Markets, Design, Semiconductors, and Tools and Flows. In this segment predictions related to design are explored.

“In electronics, years rarely offer an earth-shaking technological moment,” cautions Chi-Ping Hsu, senior vice president, chief strategy officer for EDA and chief of staff to the CEO at Cadence. “They are more likely to represent the fruits of long labor in various ways, but they can be profoundly significant.” He asks us to consider that 40 years ago, in 1975, the first personal computer (Altair 8800), first digital camera, integrated optical circuits, flatbed scanners, text-to-speech synthesis, and the first liquid-crystal display were among the highlights. “That was a watershed year,” says Hsu, and he expects 2015 to be another such watershed year.

More companies will migrate to SoC
“There is a clear trend that the system on chip (SoC) with one processor is moving to multiple processors with cache coherency,” points out Adnan Hamid, chief executive officer of Breker. “Coherency will be required beyond the CPU clusters, as GPUs, DMA engines, application processors, and I/O devices will all have to be coherent to some extent. Three-level caches will dominate, and all on-chip buses, interconnects, and on-chip networks will support cache coherency. Multiprocessor solutions will be so plentiful than even some relatively simple (IoT) chips will include them by default.”

This is a trend also seen by Ken Karnofsky, senior strategist for signal processing applications at Mathworks. “Traditional semiconductor customers are moving away from isolated DSPs and more toward SoCs, resulting in the chips becoming less single-function and more multi-function with a breadth of capabilities.”

Hsu predicts more specialized processors: “One of the key shifts will be from away from a central processing unit (CPU)-centric approach to architectures that are centered on streams of data.”

While processors are important, SoCs need to communicate with the rest of the world. Hsu reminds us that “verified, standards-based IP has become a ‘must-have’ for system design enablement companies to thrive. Design IP lets the specialists encapsulate the complexity of implementation.”

All of those pieces have to be connected together. “Standards-based bus architectures will become more common,” predicts Bernard Murphy, chief technology officer at Atrenta. “For big designs this is old news, but it will become increasingly common in designs that used to be based on more ad-hoc communication or proprietary bus architectures. Expect to see switch architectures for smaller designs and subsystems and NoC architectures for the top-level of large designs.”

Marco Casale-Rossi, product marketing manager within the design group of Synopsys, notes that the top three challenges are all related to complexity. “First, silicon integration capacity will reach 1 trillion transistors by the end of this decade and almost 10 trillion by the end of the next, to say nothing about software. Second, designs are both analog and digital, high performance and low power, posing simultaneously challenges that until very recently only could be addressed separately. And third, the breadth of active technology nodes is expanding, meaning that more than 10 process technology nodes may contribute critical ICs to a single electronic product.”

One natural progression is that hardware/software co-design will become more important than in the past. “We see the need for heightened awareness of interdisciplinary skills,” says Karnofsky. “Engineers need to have some understanding of the parts that effect the piece they are designing. More and more system architects will be required to understand the impact of target hardware on their design choices, and design engineers will need to rely on higher-level modeling and workflow automation to cope with complexity.”

Team structure and work flows may also change. “We cannot simply add more engineers to a team, especially verification engineers,” says Randy Smith, vice president of marketing at Sonics. “Instead, we need to work more intelligently. We can adapt agile development methodologies to meet IC design challenges. By doing so, the old waterfall design method can give way to a quicker, more agile process. I predict 2015 will be the year IC and SoC design become agile to meet the time to market requirements, particularly markets like Internet of Things.”

This could lead to a significant change in focus. “Functional integration at the system level will become more important than transistor scaling as die size becomes increasingly irrelevant in a growing variety of applications,” explains Paul Pickle, president and chief operating officer of Microsemi. “At the same time, approaches once considered routine will be re-examined in the quest to add more system capabilities.” Pickle goes on to explain how the introduction of FPGA capabilities can reduce cost and power and increase performance. This was an idea that Semiconductor Engineering explored in the article “An Architectural Choice Overdue For Change.”

New markets for FPGAs
Several people see the FPGA market changing. “FPGAs will be even bigger in 2015,” predicts David Kelf, vice president of marketing for OneSpin Solutions. “The 14nm devices will significantly move the volume/cost tradeoff curve, not to mention the power barrier, to a point where these devices will become widespread in high-volume applications such as automotive, communications, and multimedia.”

“In the IoT market, security is an increasing concern for remote, increasingly compact and power-conscious advanced systems that offer an ideal gateway for malevolent hackers,” says Pickle. “There is a critical need for FPGA-based security solutions that prevent tampering, counterfeiting and the installation of malicious code. We also see FPGAs increasingly replacing ASSP/ASICs for a variety of applications.”

One of the key developments that both Kelf and Pickle see is improved software toolsets that help manage the design flow. “Many FPGAs will be designed with a C-based flow, possibly using standards such as OpenCL, without any translation to an HDL,” predicts Kelf. “Such devices could propel the use of C-based design into the mainstream.”

Memory
The expectation is that many designs will stick at 28nm. “We already see die sizes just below 4 mm² in 90nm and it is clear that both 55nm and 28nm will be key next-generation technology process nodes,” points out Ron Lowman, strategic marketing manager for IoT within Synopsys.

This may bring about increased focus on some aspects of a typical chip. “2015 will be a great year to be in the memory business,” says Charlie Cheng chief executive officer for Kilopass . “There will be a lot of pressure to improve PPA at this node. Embedded memory bit-cells have all reached their limits—embedded flash cannot scale below 40nm; SRAM power and speed have hit a ceiling; and new technologies such as (ReRAM, MRAM) show a lot of potential but are far from reality. Given those opportunities, there will be renewed investment in the development of new bit-cells. 2015 will a year for innovation in embedded memories, with great opportunities for breakthrough technologies that significantly improve on PPA.”

The interfaces to off-chip memory may also see some action. “The adoption rate of LPDDR4 will be happening faster than most of us would have anticipated,” says Bob Smith, senior vice president of marketing and business development for Uniquify. “This is driven where the need for speed is insatiable, but it also needs to be balanced with low power. Expect to see lots of new designs incorporate LPDDR4 solutions in 2015 with early production coming on line as well. The volumes generated by the mobile markets should make the DRAM vendors very happy.”

Analog
Many of the emerging markets also require integration of analog components, predictably sensor interfaces and baseband RF circuitry. “These analog interfaces will have to work with multiple sensor types and provide the RF for multiple standards with differing power requirements,” says Kelf. “This would enable standard platforms to be created that would provide the necessary volume/cost tradeoffs.”

But do these components all need to be on a single die? “Packaging advances, such as system in package (SiP) solutions allow a technology-agnostic combination of analog, digital, mixed-signal and RF blocks,” says Pickle. “This provides a mix of superior performance at lower cost and with higher levels of functional integration than would be possible with die-level integration.”

Impact on the industry
What does all of this mean for some of the largest companies in the industry? “M&A in semiconductors has moved from acquisition of smaller companies, to more big billion dollar companies consolidating,” says Taqi Mohiuddin, senior director of marketing for Evans Analytical Group. “That leaves less space for smaller companies under $1 billion making it alone. Expect larger acquisitions to continue in 2015.”

Design companies may be migrating to adjacent areas that expect to see rapid growth. “There’s an entire new class of OEM-type companies from other industries adding IoT capabilities to products and lowering the cost,” according to Karnofsky.

The structure of the companies is also changing. “There is a new breed of system-design company that has taken direct control over its semiconductor and subsystem destinies,” explains Hsu. “These systems companies are reaping business (pricing, availability), technical (broader scope of optimization), and strategic (IP protection, secrecy) benefits by taking more control over their system-design destiny.”

[Last year, Semiconductor Engineering reviewed the 2014 predictions to see how close to the mark they came. You can see those in part one and part two of the retrospective. We will do the same with their predictions this year.]