The x86 processor has been a dominant technology paradigm since the 1980s and has been the world’s most widely used computing architecture for most of that period. Today brings me something I never expected to see: AMD and Intel work together About the future of x86. Rival chipmakers have formed a collaborative x86 ecosystem advisory group to improve interoperability, smooth integration, and generally simplify life for developers, ISVs, OS manufacturers, and OEMs.
Why now? Each processor architecture speaks a different “language” that determines software compatibility. Intel and AMD processors “speak” x86, while processor designs such as the Apple M-series, AWS Graviton, and Qualcomm Snapdragon “speak” Arm. For decades, x86 has dominated both data center computing and PCs. Arm, on the other hand, currently dominates smartphones and IoT. But Arm has recently been making inroads into the x86 space, as seen by its growing popularity in hyperscale data centers and its new role in Microsoft’s Copilot+ PC initiative.
Arm has a process for working with its ecosystem on instruction set updates, connecting intellectual property using a consistent set of interfaces and IP. On the other hand, x86 is architecturally led by Intel, and collaboration with AMD is primarily achieved between x86 ISVs. Indeed, to date, there hasn’t been a coordinated approach to putting AMD and Intel on the same page as customers would like.
Yesterday I did something I never would have thought of: I joined a conference call with two CEOs, Lisa Su of AMD and Pat Gelsinger of Intel, to discuss this initiative. I had many individual conversations with them and was impressed by their level of agreement. Ms. Sue even commented that she found it odd that Mr. Gelsinger answered my questions over and over again “in exactly the same way” that he would have answered them. And this isn’t just a joint initiative between AMD and Intel. It’s backed by the tech giant’s CEO. Read about what I learned during that conversation.
Full disclosure: AMD and Intel are customers of my company, Moor Insights & Strategy, as are most of our major customers and competitors.
Market pressures and technology evolution affecting X86
AI is just the tip of the iceberg when it comes to the changes that AMD and Intel must face today. In addition to computers of all kinds, from smartphones to laptops to data center servers, chip-supported intelligence is being built into industrial and residential IoT devices, smart wearables, and more automotive subsystems than ever before. and the list goes on. At the same time, the chips themselves have become much more complex, including custom chiplet designs and elaborate 3D packaging that would have been unimaginable just a few years ago. (I say this as someone who has worked at high levels in the semiconductor industry for decades.)
In this regard, the new advisory group will further deepen the appeal of x86 across a variety of platforms, making it clearer and simpler for software developers and OEMs to take advantage of x86’s unique reach and compatibility. We look for ways to allow us to continue developing our products.
Speaking of which, Gelsinger told me a story yesterday that I had never heard before. Back in the late 1980s, he led the team that developed the 486 processor at Intel. “Decades ago, when I started designing the 486, a marketing manager came into my office and wrote ‘compatibility’ at the top of the whiteboard. You know, this old thing. There’s just a value proposition where you can plug it in and it still works, right?” That’s been a major part of x86’s appeal for 40 years.
It’s also worth remembering that all of these companies, including Intel and AMD, have collaborated in the past on industry standards for technologies we now take for granted, such as PCI/PCIe and USB. To further integrate x86, this group is inviting interested hardware professionals with first-hand experience with how the architecture is implemented across a variety of settings, including cloud, data center, client, network edge, and embedded devices. Seek input from hardware and software providers.
How can AMD and Intel coexist with cooperation and competition?
Mr. Su was very frank on this point. “Frankly, in the past, some customers may have wanted more integration because it made it easier to integrate in the software stack.” In that context, working with Intel on the x86 standard She believes doing so is a “huge benefit for customers and the ecosystem.” He went on to explain that if the two companies can “align these architectural directions,” they can continue to innovate and differentiate in implementation.
Gelsinger cited the security model as an example of how the two companies are “not benefitting our ecosystem” by pursuing different approaches. “I think we could have done a better job in the past and upstreamed shared operating system level support more aggressively,” he told me. He added that Linus Torvalds, in particular, has expressed particular interest in this area as an area where the two companies could work together better.
What both companies have in common is that they are both havens for world-class engineers. Sue told me that she and Gelsinger sponsored high-level discussions among the company’s engineers. “What we found is, first of all, there is a deep respect between both teams,” she said. “Yes, we’ve been competitive for X number of years, and we think we have two of the best processor teams in the industry. And we know what problems we’re going to solve going forward.”
If the advisory group operates as planned, x86 customers and partners will be included in architecture discussions early. “This allows x86 to innovate much faster,” Su said. “And yes, obviously Pat will continue to develop his products. We will continue to build ours. There will be competition, but there will also be an expansion of x86 capabilities.”
Both CEOs also agree that there is a significant market opportunity for x86 going forward. Gelsinger was careful to point out that it’s not a zero-sum game between x86 and Arm. Still, he said, “There are billions of Arm cores in the world. [and] To some degree, I think all of this is in the target market for x86 availability going forward. ” This was the answer to my question about potential new ways to scale x86, considering that x86 is already ingrained in large-scale edge environments for data centers, storage, PC, and industrial . Further expansion of x86 is likely to occur first in automotive, mobile, and embedded devices.
What X86 ecosystem partners can gain from this initiative
The x86 Ecosystem Advisory Group’s roster of ecosystem partners includes Broadcom, Dell, Google, HP, HPE, Lenovo, Meta, Microsoft, Oracle, Red Hat, Epic Games’ Tim Sweeney and the Linux Foundation’s Torvalds. I am proud of the success of this effort, which includes many of the most important CEOs in the technology industry, including Microsoft’s Satya Nadella, Broadcom’s Hock Tan, Oracle’s Safra Catz, and Lenovo’s Yang-Ching Yang. I was shocked by the amount of supportive words from people who wanted it. , Michael Dell and others.
“x86 has been the foundation of modern computing for more than 40 years, and we want to ensure it continues to evolve and benefit everyone for years to come,” said Nadella. said. He specifically cited the role the group can play in “shaping future x86 architecture capabilities” and promoting “software consistency and standard interfaces.”
In his comments, Mr Tan looked to the future. “We are at a crossroads in the history of computing, and the x86 architecture decisions we make today will impact programming models, frameworks, and systems for decades to come,” he said. He noted that Broadcom will be able to offer a perspective on both silicon development and x86 virtualization thanks to the VMware acquisition.
Regarding the founding members, Mr. Hsu said: You know, what we’re trying to do with this is continue to be able to address a wider range of applications faster and more easily. ”
Gelsinger added, “The goal is to grow the x86 ecosystem and create front-line energy. That’s why we’re shaping it. That’s why the advisors we’ve assembled for this… For him, it’s about “taking the best software ecosystem in the world that’s ever existed and positioning it firmly for the future.”
What success means to the X86 Ecosystem Advisory Group
For this effort to be successful, several things need to happen. First, Intel and AMD must live up to the goodwill expressed by Gelsinger and Su during their conversation. One of the things I love about engineers is that they have a common spirit of getting things done. Both AMD and Intel have a lot of talented engineers who are bringing x86 to life. Therefore, I am confident that if both companies respond to input from their partners and foster true collaboration, we can achieve great things together.
When it comes to interacting with partners, groups need a well-understood decision-making process that everyone agrees with. This means that quick and well-vetted decisions are made on behalf of the whole group, not just some members. Given the nature of this group, I would like to see it move a little more quickly and decisively than some industry standards bodies. In any case, all member companies must ensure that their views are fully taken into account. My hope is that this group’s meetings will be full of candor and its published standards full of transparency, as no one among its members seems to be cowering.
One specific point: success also depends on being able to address aspects of the ecosystem beyond x86 instructions. I’m especially thinking about built-in accelerators like SSE, AVX, AMX, and SHA. When I asked CEOs about this particular issue, Gelsinger had this to say: “We think AVX, AMX, security enhancements, all of those things will be on the table, because that’s the kind of feedback we want right now.” What AMD and Intel will put into the product , each individually makes its own decisions about how to compete in the market, but about everything related to the ecosystem. [and] I bring both application-level and OS-level ISVs to the table for discussion. That is the premise for a group of very powerful companies to join the advisory group. ”
At that time, Ms. Sue told me that she would answer my question “exactly the same way.” She added, “What we’re trying to do is make this easy for our customers in the ecosystem… There are a lot of other places where we can differentiate and we’re going to differentiate. And that market is huge, right? We’re in a large, growing market, so let’s make it easier for the industry to embrace x86 and innovate.”
That’s music to my ears and is definitely the main reason behind the group members’ glowing quotes referenced above. Only time will tell how well AMD and Intel execute on their blueprint for success in this endeavor. Until then, hats off to Lisa Sue and Pat Gelsinger for coming together like this. If executed well, I’m confident that x86 will further improve its standing.