SiFive CEO: How RISC-V will shape the future of computing

Recently, Patrick Little, chairman, president and CEO of SiFive, talked about how RISC-V is shaping the future of computing, how SiFive is gaining momentum in applications from embedded to data centers, and how the company is leading the way as AI becomes "ubiquitous."

Here are the details of the article:

I want to zoom out a little bit first, because we're so focused on what's going on every day, and talk about how the larger gears in the industry are turning and gaining more and more inertia and momentum over time. One of them is that architectures and interfaces, whether it's hardware or software interfaces, are moving to open standards, and there's almost no disagreement. The only debate is the speed of that progress - how long it will take. And closed architectures that rely on vendor lock-in actually limit innovation because they are driven by only one company and are limited by the resources, capabilities, and talent that company can afford. We believe that over time, those remaining proprietary standards will gradually disappear. In all the customers I've talked to in the last four and a half years, no one disagrees with this, it's just a matter of time.

The market should not be limited to just a few key players because that doesn't open up innovation and allow all the best ideas to emerge. Obviously, coming from RISC-V and SiFive, we believe that open standards will not only bring more choice to the ecosystem, but also more stability. If you've been in this industry long enough, you've seen systems like the Alpha architecture or the MIPS architecture come and go. When you bet everything on one architecture from one company, you're betting the next few decades on whether that company can continue to move forward. If they can't, then your investment in that architecture will be gone. This has happened many times over the past few decades.

So, in the long run, community-driven collaboration has advantages, where we can leverage everyone's input, best ideas, and expertise. One company is good at software, another is good at tools, some are focused on IP, but together they form a very powerful, robust, and more importantly, long-lasting ecosystem. We feel it every day - in the future, all architectures and interfaces, whether it's software or hardware, will move towards open standards. In ten years we will look back and say, "Remember when this was proprietary? One company controlled the entire space."

Another very important paradigm shift, which I think is overlooked by many people but not by most customers, is that hardware will enable the vision of software, not limit it. Today, it's the other way around, and most software developers have to look at what the hardware can do and then work within the framework of the hardware. If the hardware is provided by a proprietary company, they can only provide limited functions and systems, which limits the innovation that software developers can achieve. Whenever I meet with software developers, this is almost the only topic they want to talk about: how to break free from the limitations of hardware and innovate freely? Why can't the hardware keep up? Why can't the hardware step back and make way for developers? I hear this complaint from our large customers every week.

So, developer innovation cannot be limited to the capabilities of the hardware, especially when hardware often lags behind and limits innovation. A lot of application developers and software developers have great ideas, but it takes too long to implement them in hardware. We want to change that with flexible, customizable hardware. So one of the fundamental philosophies of our company, and the core philosophies of RISC-V, is that you need customization and flexibility in hardware. If someone says, "I have a great idea that I want to implement in software," you shouldn't tell them to wait three years for a new version, but you should say, "Let's look at the infrastructure and see if there are microarchitecture extensions to support your idea." I think that's the key to unlocking the potential of our industry.

This is the core of RISC-V - its purpose is to free developers from the constraints of proprietary computing. Everyone has a RISC-V architecture license, and they can start designing at any time, depending on their creativity. You just have to follow the architectural standards, but you can be as creative as you want on top of it.

RISC-V has three very important advantages that I think many people still don't really understand and appreciate today. These advantages will drive it to a very important position in the next few decades and are very consistent with customer needs. The first advantage is the community-driven collaboration I just mentioned - "No one person is better than all of us." This is the core concept of RISC-V. Everyone participates together, puts their respective strengths, weaknesses, capabilities and investments on the table, and then integrates them, everyone can benefit, and this is a level playing field.

The second point that is usually overlooked by most people, but it's the first thing customers want to talk about, is that standard programming interfaces enable an interoperable ecosystem. With RISC-V, you have a standard programming interface, and the other end of the interface could be my core, his core, one of my cores, or a mix of all of them. You're not dependent on one company - it could be 10 companies, 100 companies can come and go, and the ecosystem can continue to thrive. Many customers do their own designs, and their designs are sitting side by side with some of our cores, which is actually the norm. So they are combining their capabilities with things that they don't want to do or can't do yet. We are in a state where customers are mixing and matching our capabilities with their capabilities, and even our competitors. That's the idea behind RISC-V, and what's really powerful is that when you think about new processors or new capabilities in the future, it can all run behind a standard programming interface. So no matter what happens to one company, the interface remains the same, and you don't have to give up the investment you've made. As long as the standard continues to evolve, it will continue to lead the industry forward.

This is to me, standard programming interfaces are key at this stage, but it is equally important that they remain key to success as we move into vector calculations, matrix multiplication, and artificial intelligence (AI). The reason is that we already have a large company using a proprietary standard, but no developer is willing to migrate their work to some new proprietary standard. Open standards allow developers to write code knowing that they can use hundreds or even thousands of different choices on the other side of the interface.

The third obvious advantage is the elimination of dependence on a single supplier. You may think that a company is strong and it will always be strong, but history tells us that this is never true. Companies rise and fall frequently, and if you bet everything on a company, no matter which company it is, you will regret it over time. Therefore, eliminating dependence on a single supplier and allowing companies to come and go as the wind is the right way, and it is also the reason why the entire industry or industries are moving to open standards.

Finally, RISC-V hardware can be customized to the specific needs of the customer. You might ask, "Isn't any architecture customizable?" The answer is: not exactly. Only those who have the architecture license can really change the underlying product itself. In the RISC-V community, everyone can influence the direction of the architecture, and everyone is a RISC-V licensee, which allows each licensed company to grab the architecture and customize it to improve the performance, reduce power consumption or reduce the area of ​​the product. From the smallest micro customer to the large infrastructure company, the RISC-V architecture is flexible enough.