GlobalFoundries announces 14nm validation with AMD Zen silicon

By Joel Hruska 

For months, there’s been speculation around the future of AMD’s Zen CPU, as well as questions regarding the state of GlobalFoundries 14nm process technology. While the foundry initially canceled its own 14nm-XM process in favor of licensing Samsung’s IP for the 14nm node, later reports claimed that the foundry was still having trouble with Apple’s A9 SoC. As a foundry, GlobalFoundries doesn’t normally make major announcements on behalf of its customers, but today is an exception.

According to GF, it has demonstrated “silicon success on the first AMD products using GlobalFoundries most advanced 14nm FinFET process technology… AMD has taped out multiple products using GlobalFoundries’ 14nm Low Power Plus (14LPP) process technology and is currently conducting validation work on 14LPP production samples. Today’s announcement represents another significant milestone towards reaching full production readiness of GlobalFoundries’ 14LPP process technology, which will reach high-volume production in 2016. The 14LPP platform taps the benefits of three-dimensional, fully-depleted FinFET transistors to enable customers like AMD to deliver more processing power in a smaller footprint for applications that demand the ultimate in performance.”

“FinFET technology is expected to play a critical foundational role across multiple AMD product lines, starting in 2016,” said Mark Papermaster, senior vice president and chief technology officer at AMD. “Globalfoundries has worked tirelessly to reach this key milestone on its 14LPP process. We look forward to Globalfoundries’ continued progress towards full production readiness and expect to leverage the advanced 14LPP process technology across a broad set of our CPU, APU, and GPU products.”

AMD’s Zen is expected in late 2016, early 2017.

14LP is the second generation of 14nm technology available at GlobalFoundries; the company demonstrated first-generation 14nm LPE back in January. AMD’s upcoming CPU architecture, Zen, is currently expected to debut at the end of 2016 or early 2017, with Zen-based APUs following at an undisclosed later date.

High-powered CPUs on a low-power process?

Last month we had an opportunity to sit down and talk with GlobalFoundries about the particulars of its 14nm technology and its technical partnership with Samsung. One question we had was whether or not Samsung’s 14nm technology would be a good match for AMD’s x86 CPUs. Samsung, after all, has no experience in the high-end, high-power CPU market, and its 14nm process node explicitly references low power designs.

Historically, CPU architectures that emphasized low-power have had limited clock frequency headroom and scaled relatively poorly to higher TDPs. This is one reason why ARM deployed technologies like big.Little, and many of the changes that AMD made to its Kaveri and Carrizo line of processors were meant to cut power consumption at the expense of maximum clock frequency. AMD has said that it intends to scale Zen across its entire product stack, which raises the question of whether or not a low-power process can keep up.

According to GlobalFoundries, it has the option to customize its process technologies to the needs of individual customers. While AMD has previously stated that it will not pay for any custom silicon, our understanding is that this applies to major technology shifts. AMD, in other words, won’t pay GlobalFoundries or TSMC to build a custom silicon line that suits a particular product. When it designed Bulldozer, AMD agreed to pay GF to build a 32nm PD-SOI (Partially depleted silicon-on-insulator). That doesn’t mean the two companies can’t work together to adopt less drastic types of optimization.