DAC 2022: Democratization of chip design
At 59 of this weeke Design Automation Conference (DAC 2022) in San Francisco, a panel discussion focused on the question: “Is the democratization of chip design already underway?” Here, one of the panelists, Vic Kulkarni, Director of Strategy at Si2, provides a summary of the discussion and his own additional notes.
Before diving into its notes, to set the scene, the panel discussed how far we’ve come in terms of the democratization of silicon. This is based on the fact that silicon has become ubiquitous, is no longer limited to powerful servers, and silicon content has found its way into many systems. Custom silicon is no longer uncommon and is often a necessary step to enable differentiation. The panel explored whether silicon design and manufacturing is still only accessible to a few, or what could be done to remove the significant barrier to entry.
Here are the panel questions (asked by me) and notes from Vic Kulkarni.
Nitin Dahad: What do we mean by democratization of chip design?
Vic Kulkarni: Plato, the Greek philosopher, wrote in the early 1500s that equality without rules brings power-seeking individuals or institutions motivated by personal gain. They can become highly corruptible, which can eventually lead to tyranny. In this context, democratizationwith membership makes more sense to the chip design world, where there is a spirit of community, collaboration, interaction and member contribution, and above all, data security, to easily design and produce chips.
Today, the bar for creating innovative custom silicon designs has been raised to over $50 million per chip, in addition to building a software stack to make it a success. This has resulted in an elite class of IP, SoC, and silicon-to-system system designers.
We need to lower the barrier to entry from a financial and technology investment perspective for the broader ecosystem:
- Provide access to IP building blocks (soft and hard)
- EDA tools for academic researchers, students
- Innovative startups focused on new applications ranging from 5G to 6G, hyperscalers, AI, edge computing, automotive 3D sensors.
In this context, democratization with strong guardrails for the protection of intellectual property and the safety of all members. A proven model is Si2’s OpenAccess, an example of democratization with the membership of over 50 companies, including contributors, developers, ranging from foundries, IP, EDA and chipmakers. This now contains over a million lines of code.
Nitin Dahad: Is open source a key element of this democratization?
Vic Kulkarni: Open source is one of the key elements of democratization. It’s a catalyst because it eliminates the cost of purchasing commercial EDA tools. The work of Professor Kahng and contributors to the OpenRoad project is a prime example. Github, Kafka and RedHat are good examples. The FPGA has an infrastructure that enables democratization. RISC-V is another important example.
However, we can think of other options that EDA companies could offer with well-defined business models for democratization.
- A startup version, perhaps in SaaS
- Tiered pricing based on problem complexity, number of users
- Minimum fees for universities, students, researchers.
We also need to bring order out of chaos through the collaborative membership model. Technical contributions from the community, access control, security and traceability of downloads, can use a blockchain-based approach. A prime example of this is the emergence of a well-defined app store ecosystem with secure processed data APIs created and supported by the FAANG community.
The value proposition of such APIs is quite clear:
- Activates an ecosystem of developers
- Spur innovative apps developed on software platforms
- The APIs are well documented and provide secure, consistent and flexible interfaces for users.
The result of such democratization has seen an impressive growth of applications using platform-specific APIs. This resulted in an explosive growth FAANG companies and their platforms. Of course, the knowledge gap between having a nascent idea and converting it into working C or Python code for application workloads and translating it through the tape-out process is a big challenge.
Nitin Dahad: Are traditional EDA vendor business models helping or hindering democratization?
Vic Kulkarni: Traditional EDA vendor models are a barrier to democratization. However, EDA companies evolve to adapt to the needs of new entrants. Three major players, Synopsys, Cadence, and Ansys CDNS have announced priceable SaaS-like models for parts of their product portfolio, and it’s safe to assume there’s more to come.
The recent push towards cloud-based business models certainly allows new players in chip design to access a wide range of EDA streams without having to invest large sums upfront in compute tools and infrastructure. .. Silicon Catalyst is a great channel to enable startups with their in-kind partnership program.
Nitin Dahad: What about skills, what should be done to make it easier for non-specialists to exploit the democratization of silicon?
Vic Kulkarni: Before we get to democratization, there is already a huge talent gap, due to the recent growth in solid-state system designs using custom silicon for application-specific vertical systems. As mentioned, we now only have the elite class of IP, SoC and Silicon for system designers. Enabling the necessary workforce requires public-private partnerships – such as government funding of more than $450 million per year for Purdue University in the United States to massively increase the number of graduate students in multiple disciplines.
We also find that non-traditional education mechanisms can also come into play. Today, there are many online resources available. However, if this level of training is necessary, then democratization will not occur. We need better tools that can transfer ideas to silicon in a fully automated way – for example, by “compiling” from a high-level behavioral language, without requiring knowledge of layout, IP design, in manufacturing process, etc.
Focus more on off-the-shelf solutions, whether focused on a vertical market (e.g. automotive, AI/HPC, memory) or horizontal technology (power efficiency, multi-die systems, thermal). Another area that can make life easier for non-specialists is AI/ML-based automation.
Nitin Dahad: What is the most important thing for you in terms of where we are in this process, and the one thing you would like the audience to walk away from and think about?
Vic Kulkarni: I think the next wave of innovation will come from a system-centric mindset for silicon-to-system innovation with ecosystem collaboration. Dr. Mallik Tatipamula, CTO, Ericsson Silicon Valley, recently shared an important vision of a multidimensional ecosystem mindset to make the transition from 5G to 6G an achievable goal within the next 5 years.
What I’d like the public to think about are things like:
- How to overcome this barrier of talent shortage? What are some creative ways this can be done?
- What will be a more practical approach? For example, silicon or programmable systems. FPGA is a form, where it has already been manufactured.
- In the Edge IoT space, many are producing kits that allow the system to be customized to meet their needs. That’s a more likely middle ground.