Intel may be struggling to get 10nm processors out the door, but rival chip maker AMD is already shipping 12nm processors… and AMD CEO Lisa Su says the company is almost ready to make the move to 7nm.

AMD’s next-gen “Zen 2” architecture is manufactured on a 7nm processor, and Su says the company plans  to start sending Zen 2 samples to customers later this year, with mass production scheduled to begin in 2019.

Su’s remarks came during the company’s latest earnings call, and it sounds like the plan is to continue shipping 12nm chips this year and next year, while starting to mix in next-gen 7nm chips in 2019.

It also sounds like the first 7nm parts will likely be designed for servers and sold under AMD’s EPYC brand, rather than consumer-oriented chips sold under the Ryzen name.

But that doesn’t rule out the possibility that we could see 3rd-gen, Ryzen 3000 series chips later in 2019. A leaked product roadmap that made the rounds earlier this year suggests that AMD’s 3rd-gen Ryzen processors will be code named “Castle Peak,” and that they could also hit the streets in 2019.

In related news, AMD is also planning to sample 7nm GPU chips this year, based on the company’s Radeon Vega graphics technology.

Support Liliputing

Liliputing's primary sources of revenue are advertising and affiliate links (if you click the "Shop" button at the top of the page and buy something on Amazon, for example, we'll get a small commission).

But there are several ways you can support the site directly even if you're using an ad blocker* and hate online shopping.

Contribute to our Patreon campaign


Contribute via PayPal

* If you are using an ad blocker like uBlock Origin and seeing a pop-up message at the bottom of the screen, we have a guide that may help you disable it.

Subscribe to Liliputing via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 9,543 other subscribers

20 replies on “AMD: 7nm Zen 2 processors coming in 2019”

  1. Also AMD’s roadmap looks like they’re taking over and adopting Intel’s old Tick-Tock cycle. Although in this case, it’s more of a tick-tick-tock.

      1. No it’s more like a tick-tock-tock-tock cycle. The tick is for die shrinks and the tock is for architecture optimizations or in the past new architectures. Since Intel hasn’t done a die shrink for awhile, they haven’t had a tick lately and have only been continuing to optimize and enhance the Skylake architecture through different iterations.

        1. The more you post, the more I skip over (without reading at all) comments with your user name.

          1. Lol you actually read all the comments in the first place? I can’t imagine how you’d handle larger sites with more comments, especially political ones.

  2. I’d really like to know what’s going on with Intel behind the scenes, why they’re struggling with 10nm chips. If AMD beats Intel to true 7nm chips, that’ll be a huge blow to Intel. Qualcomm and Samsung have been shipping pseudo 10nm chips for some time as well (depending on your definition of 10nm). Intel’s 10nm chips better be worth the wait…

    1. From what I read recently, Intel’s been trying to pack more transistors into the same amount of space than they’d typically get from a process shrink. Typically, they could double the amount of transistors, but they’re going for even more – like 2.7x. And at least for the time being, it seems they’re trying to get a lubed Cadillac to fit into a dog house.

      I think Intel would have been better off in the long run in going with the more traditional gain and then refining the process with a future iteration – like Skylake to KabyLake to KabyLake-R/CoffeeLake.

      It’s a bit reminiscent of the 90nm process and 200mm to 300mm wafer transition around 2004. Intel had a bear of time getting all that – plus the Netburst core under control. It gave AMD a great opportunity. I think AMD’s on the cusp of another opportunity.

      And a healthy and competitive AMD is great for all of us.

      1. The Skylake architecture has been around for awhile now that I think about it. All subsequent architectures have just been additional optimizations and impovements of Skylake. Even after the die shrink, what Skylake started will still be around for sometime. Usually a generation after a die shrink, Intel introduces a brand new architecture, but they haven’t so far either.

        Anyways, it’s great that AMD is making a comeback, hopefully that’ll push Intel to get their act together. I’d like to see VIA make a comeback too. They are the 3rd player in the x86 business that few people know about. But at one point a long time ago, VIA was doing a lot of cool things with small form factor and low power designs even before Intel introduced Atom.

  3. So buyers of server parts are concerned about power consumption? I guess that’s good, but I usually think of power consumption being more important on mobile devices.

    1. Well yes power consumption is important on mobile devices, but when someone operates large server farms, that consumes a lot of power and will generate a massive power bill. If you can reduce power consumption of all your servers even by 25%, you’d be saving thousands every month. That adds up pretty quickly so reducing power consumption everywhere is always a plus.

        1. Yes but performance is also important and ARM server chips have horrible performance per watt ratios compared to x86 chips. You’re not going to be able to run an efficient farm of compute and virtualization servers with ARM chips anytime soon. Intel’s got that locked down with their Xeon chips which have amazing performance while keeping power consumption at acceptable levels.

          1. Lol, ARM actually has far superior performance per watt ratios. You always compare a single best ARM chip with single best Intel chip……that is irrelevant. Power of ARM comes from numbers…high amounts of cores strung together to match the power of a Xeon chip at half the price and half the power usage.

            It’s like comparing a giant that can pick up 2000 pounds vs 10 puny humans who can pick up 200 pounds each. But the 10 humans still eat less than half of what one giant eats. ARM and RISC computing is simply extremely efficient, most of world’s servers are going to be running on ARM eventually.

          2. The single best Intel chip isn’t the best because it’s a single core CPU. It’s the best because it’s a multi core CPU so your analogy there isn’t accurate. An Intel CPU would be like 10 giants that can pick up 2000 pounds each vs 10 puny humans who can pick up 200 pounds each. Not a single one of the top supercomputers or HPC server farms are powered by ARM. For very good reason too. Not every server is a web server these days anymore. High performance compute and data science is becoming a growing field and every bit of performance matters.

            Sure you can say 10 top of the line ARM CPU’s can match a single top of the line Intel CPU. And MAYBE they can perform more efficiently in that configuration. But don’t you see the problem with scaling out in that case? Going back to the big data use case, a lot of powerful servers are needed. 10 ARM CPU’s take up way more server space than a single Intel Xeon Phi for example. What happens when you have a massive compute load that requires 1000 Intel Xeon CPU’s? With your comparison, that would require 10,000 ARM CPU’s. That’d take up way more space and require way more servers.

  4. This is definitely bad optics for Intel. Which increases the odds that they take this threat seriously, that we get better competition / pricing… love it.

    Keep up the good work, AMD!

  5. It will be a huge win for AMD if they can get 7 nm chips out in volume within their stated timetable. I am not convinced. Getting to 7 nm is incredibly difficult. Doing it in volume even more so. I wish them luck.

    1. AMD is already sampling Zen 2 in high volumes so they are well underway.

      Paywalled content at SemiAccurate

Comments are closed.