How Semiconductor IP Companies Offer ASIC-Like Design Services

Between your home, car, doctor, and workplace—from your online posts to your purchases—data is grist for the mill to make modern life easier. In fact, the demand for data is insatiable. With IoT, analytics, machine learning, and more, data is only getting bigger and more complex. The engines driving it all are semiconductors. To meet new technical requirements, delivering more data at faster speeds and lower power and latency, fundamental SoC design paradigms must change. Moore’s law is sunsetting as we approach the reticle ceiling of manufacturing: It’s simply impossible to fit more transistors onto ever smaller chips. This means that the kinds of performance gains must come through a new era of semiconductor innovation.

Today’s innovations are delivering SoC designs that are larger and more complex than ever, bringing with them inherent integration challenges. This means that standard semiconductor IP blocks used for non-differentiating aspects of your chip design are less plug and play than they used to be. This new friction in the chip design process can cost you resources, time, and money. But what if standalone IP blocks came with a comprehensive IP solution within key applications, easing the integration path from die to board and beyond? What if your semiconductor IP vendor offered more solutions like an ASIC vendor does?

5 Steps to Meet Your Application-Specific Needs

Synopsys is the only IP company that has the scale and the infrastructure to address your needs in this way. We use a five-step approach:

• Maturity and portfolio: We have breadth in our high-quality IP portfolio. For example, an HPC SoC might use five or six different IP blocks: Ethernet, PCIe, memory interfaces, die-to-die, security, and so on. We have a consistent user experience for all of them, including support, integration, usability, and more, along with strong relationships with foundries to run test chips where we characterize the test chip and run it for you.
• PPA&L: We have multiple generations of architectures optimized for a wide variety of technical requirements that vary by application, whether it’s power, performance, area, or latency.
• Complete Solution: We don’t simply provide the IP blocks, for instance, a PHY or controller. We can provide the complete, pre-verified solution for the entire subsystem, so that the Interface IP is plug-and-play optimized for PPA&L.
• Integration and support: We give you the benefit of our experience building and testing systems, providing the recipe for good design in the form of collateral and support, so that your integration works right from the get-go. The collateral can include routing feasibility study, packages substrate guidelines, signal and power integrity models, and crosstalk analysis.
• Ecosystem: Our longstanding relationships include an entire ecosystem of partners, such as foundries, test equipment vendors, system vendors, and other experienced players in a wide variety of industries, from hyperscalers to the edge. We take a leading role in many standards bodies, often helping to define standards such as Ethernet, PCIe, UCIe, MIPI, USB and more, so that you can rest assured our IP will comply with key specifications and help you achieve successful interoperability. We work closely with major foundries, which gives us an understanding of process geometries and how those geometries are progressing.

Turnkey Semiconductor IP Solutions for Complex, Advanced SoC Designs

Much like development for an ASIC vendor, the Synopsys IP development process involves a great deal of due diligence: we understand design challenges in a deep way because we replicate large SoC designs in real-world contexts—with full integration exercises—to provide complete solutions to ease your process for advanced designs and enable your success.

We make sure we can place and route our own PHY, PCS, or MAC IP, and close timing at high speeds. We take our IP from the die, to the package substrate, and then to the board to make sure high-speed package escapes work on a minimal number of layers. We mockup a large customer SoC with hundreds of lanes together, and then we go deeper into the SoC to understand the cross talk and the IR drops.

We learn from our own experience and then we transfer that knowledge through a complete set of guidelines. Gone are the days of simply selecting IP and then receiving a .gds file of the function to your FTP site, leaving you on your own to do the hard integration work. Our guidelines tell you how to put the IP blocks together, integrate them into your package, and build your board. We make your design life easier—to get it right sooner—sparing you from an otherwise highly iterative design process.

While we cannot replicate the situation of every customer individually, we learn from our early adopters, using their unique experiences and challenges. As we work with them to ease the design process, we simultaneously build our library, leveraging these experiences and cross-pollinating them into recipes for your success.