Editor's note: Recently, John Koon , the technical editor of " Semiconductor Engineering " magazine , published an article analyzing the reasons why the automotive market chooses Chiplets and the challenges currently faced in this field. He said, " Chiplet has regained attention in the automotive market, and the evolving electrification ( electrification ) and fierce competition have forced many companies to accelerate design and production plans. " China Exportsemi Net will present you with more detailed content below.
Electrification has ignited enthusiasm among some of the largest and best-known automakers, which are struggling to stay competitive in the face of very short market windows and changing demand. Unlike in the past, when automakers typically employed five- to seven-year design cycles, the latest technology in today's cars is likely to be considered obsolete within a few years. If they can't keep up, there will be a whole new crop of startups building cheap cars that can be updated or have features changed as quickly as software updates.
But software has limitations in speed, security and reliability, and custom hardware is what many automakers are now working toward. This is where Chiplets come in, and the focus now is on how to build enough interoperability in a large ecosystem to make it a plug-and-play marketplace. Key factors in achieving automotive chiplet interoperability include standardization, interconnect technology, communication protocols, power and thermal management, security, testing and ecosystem collaboration.
Similar to non-automotive applications at the board level, much design effort is focused on a die-to-die approach, which drives many novel design considerations and trade-offs. At the chip level, due to the increase in design performance requirements, the interconnections between various processors, chips, memories and I/Os have become increasingly complex, triggering a series of standards activities. Different interconnect and interface types have been proposed to serve different purposes, and emerging chiplet technologies for specialized functions (processors, memory, I/O, etc.) are changing the approach to chip design.
“Automotive OEMs recognize that to control their own destiny, they must control their SoC,” said David Fritz, vice president of virtual and hybrid systems for EDA at Siemens. “However, they don’t understand that EDA has been around since college in 1982. How far to go. Also, they think they need to get into the latest process node, where a mask set would cost $100 million. They also can't afford it because the talent pool is pretty small. Together, OEMs recognized that to control their own destiny, they needed a technology developed by others but that could be combined as needed to have a uniquely differentiated product that they believed would, at least for some, be A future-proof model year. Then the only one that fits the bill is the Chiplet.”
Chiplets can be optimized for specific functions, which can help automakers meet reliability, safety and security requirements using technology that has been proven across multiple vehicle designs. Additionally, they can reduce time to market and ultimately reduce the cost of different features and functionality.
Demand for chips has been growing over the past decade. According to Allied Market Research, global automotive chip demand will grow from US$49.8 billion in 2021 to US$121.3 billion in 2031. This growth will attract more automotive chip innovation and investment, and Chiplet is expected to be a big beneficiary.
But it will take some time for the chiplet market to mature , and it may be launched in stages. Initially, suppliers will offer proprietary molds in different styles. The partners will then work together to provide chiplets to support each other, as is already happening with some vendors. The final stage will be universally interoperable chiplets, supported by UCIe or other interconnection schemes.
Getting to the final stage will be the most difficult and will require significant changes. To ensure interoperability, enough parts of the automotive ecosystem and supply chain must come together, including hardware and software developers, foundries, OSATs, and material and equipment suppliers.
Momentum is building
On the bright side, not all of this is starting from scratch. Onboard chip-to-chip interfaces have always been used at the board level, modules and subsystems and will continue to be done so. Various chip and IP providers (including Cadence, Diode, Microchip, NXP, Renesas, Rambus, Infineon, Arm, and Synopsys) offer off-the-shelf interface chips or IP to create interface chips.
The Universal Chiplet Interconnect Express (UCIe) Alliance is the driving force behind the chip-to-chip open interconnect standard. The organization released the latest UCIe 1.1 specification in August 2023. Board members include Alibaba, AMD, Arm, ASE, Google Cloud, Intel, Meta, Microsoft, NVIDIA, Qualcomm, Samsung, etc. Industry partners showed widespread support. AIB and Bunch of Wires (BoW) have also been proposed. In addition, Arm has just released its own Chiplet system architecture, as well as updated AMBA specifications to standardize the Chiplet protocol.
“Chiplets have emerged, driven by necessity,” said Arif Khan, senior product marketing group director, Design IP, Cadence. “Growing processor and SoC sizes are reaching graticule limits and diseconomies of scale. Process technology advancements have brought The incremental benefits are lower than the rise in design costs per transistor and advances in packaging technology (2.5D/3D) and inter-chip interface standardization (e.g. UCIe) will facilitate chiplet development.”
Almost all chiplets in use today are developed in-house by large chip manufacturers such as Intel, AMD, and Marvell because they have tight control over the characteristics and behavior of these chiplets. But work is underway at all levels to open the market to more players. When this happens, smaller companies can begin to capitalize on what high-profile trailblazers have achieved to date and innovate around these developments.
"Many of us believe that the dream of having a ready-made, interoperable portfolio of chiplets may take years to become a reality," said Guillaume Boillet, senior director of strategic marketing at Arteris, adding that interoperability will come from the partner community. , they are addressing the risk of incomplete specifications.
This also increases the appeal of FPGAs and eFPGAs, which can provide a level of customization and updates to field hardware. "Chiplets are real," said Geoff Tate, CEO of Flex Logix . "Currently, it is more economical for a company to make two or more chiplets than for one company to make close-to-reticle-size chips with almost no yield. Chiplet standardization seems far away. Even UCIe does not have a fixed standard yet. Not everyone All agree on UCIe, die testing and who's responsible for issues when the integrated package doesn't work, etc. We do have some customers using or evaluating interfaces for eFPGAs that have changing standards such as UCIe that they can implement on their chips now. Use eFPGA to adapt to future standards changes."
There are other efforts to support chiplets, albeit for a few different reasons—notably the rising costs of device scaling and the need to integrate more functionality into chips that are constrained by reticles at the most advanced nodes. But these efforts also pave the way for chiplets in the automotive space, and there is strong industry support to make it happen. For example, the new Heterogeneous Integration Roadmap (HIR), sponsored by SEMI, ASME and three IEEE societies, looks at a variety of microelectronics design, materials and packaging issues to develop a roadmap for the semiconductor industry. Their current focus includes 2.5D, 3D-IC, wafer-level packaging, integrated photonics, MEMS and sensors, and system-in-package (SiP), aerospace, automotive, and more.
At the recently held 2023 Heterogeneous Integration Global Summit, representatives from AMD, Applied Materials, ASE, Lam Group, MediaTek, Micron, Onto Innovation, TSMC and other companies showed their strong support for Chiplets. Another organization supporting chiplets is the Chiplet Design Exchange (CDX) working group, which is part of the Open Domain Specific Architecture (ODSA) and the Open Compute Project Foundation (OCP). The Chiplet Design Exchange (CDX) charter focuses on various features of chiplets and chiplet integration, including electrical, mechanical and thermal design exchange standards for 2.5D stacking and 3D integrated circuits (3D-ICs). Its representatives include Ansys, Applied Materials, Arm, Ayar Labs, Broadcom, Cadence, Intel, Macom, Marvell, Microsemi, NXP, Siemens EDA, Synopsys, etc.
“Automotive company requirements for the functionality of each chiplet are still in a state of flux,” Siemens’ Fritz noted. "You have these problems at one extreme and you have these problems at the other extreme. That's the sweet spot. That's what's needed. These are the types of companies that can do this kind of work and then you can bring them together. So this mutual The operational thing isn't that big of a deal. The OEM might overcomplicate things by saying, 'I have to deal with all possibilities.' Alternatively, they might say, 'It's just like high-speed PCIe. If I want to communicate from one person to another, I already know how to do it. I have the driver running my operating system. This will solve a lot of problems and I believe this is the end result. ”
You can also click to view the original English text:
https://semiengineering.com/why-chiplets-are-so-critical-in-automotive/
