IBM officials want to remind businesses that Intel’s x86 processors aren’t the only server chips on the market.
Company engineers were at the Hot Chips 2016 show in Cupertino, Calif., this week to talk about Big Blue’s upcoming Power9 processors, which are scheduled to hit the market next year and promise a broad range of capabilities that will make the architecture attractive for an array of workloads, from hyperscale data centers to machine learning and artificial intelligence (AI).
They reiterated many of the key assets of Power9, from its new architecture that will offer up to 24 processing cores and its ability to run with a wide array of accelerators like GPUs, field-programmable gate arrays (FPGAs) and application-specific integrated circuits (ASICs) for faster performance to embracing such technologies as Nvidia’s NVLink 2.0 and PCI Express 4.0.
It’s a cornerstone of IBM’s larger effort to carve into Intel’s dominant share of the server chip market, where more than 90 percent of all systems run on x86 chips from Intel. Big Blue is one of a number of chip players—including Advanced Micro Devices and its upcoming Zen chips and ARM with its push into the high-performance computing (HPC) space and supercomputers—that see changes in workload and processing demands in data centers as an opportunity to grab some share away from Intel.
It reportedly was the message at Hot Chips, which came a week after IBM officials were in San Francisco outside of where Intel was hosting its annual Intel Developer Forum (IDF) to talk about Power9 and the OpenPower Foundation.
“We want people to know there is an alternative to x86 chips and that alternative can bring a lot of performance with it,” Dylan Boday, IBM Power engineer, told eWEEK last week standing outside of the Moscone Center, home to IDF. “At the end of the day, most people want choice, but they also want to see advantages to that choice.”
IBM traditionally had developed Power chips to run only in its Power servers. However, the company three years ago—with such partners as Nvidia and Google—launched the OpenPower Foundation, enabling third parties to license the architecture to create their own Power-based systems. It was part of a larger effort to embrace open technologies—such as Linux, OpenStack and the Open Compute Project (OCP)—for its Power architecture.
The work is paying off, according to IBM officials. At the first OpenPower Summit last year, the group had about 130 members. That has since grown to more than 200. At the same time, there are more than 2,300 applications that run on Linux on Power, they said.
In addition, there are projects underway with both Power8 and Power9. Google and Rackspace in April announced a project called “Zaius,” in which they are working on a dual-socket OpenPower-based server design powered by Power9 and that will include OpenCAPI and Nvidia’s NVLink connectivity, DDR4 memory and the CAPI acceleration technology. The companies hope to contribute the design to the Facebook-led OCP. IBM also is working with Nvidia and original design manufacturer (ODM) Wistron to build a next-generation Power8 system for HPC environments.
IBM Takes Aim at Intel With Upcoming Power9 Chips
IBM also is getting testimonials from customers like Kinetica, which offers an in-memory database technology powered by GPUs. CEO Amit Vij told eWEEK that his company is adopting the Power architecture and testing Nvidia’s NVLink to accelerate the performance of its products. Kinetica is seeing a three- to four-times performance increase using Power chips with NVLink as well as significant gains in other areas.
Industry analysts have said that enterprises and service providers are looking for an alternative to Intel to help drive innovation in the market, protect them against supply chain problems and serve as leverage in price negotiations with Intel. In this case, the Power9 chip could prove to be an asset, according to Patrick Moorhead, principal analyst with Moor Insights and Strategy.
“If someone is seriously looking for an alternative to Intel, then IBM is in a good position,” Moorhead told eWEEK, adding that if a business is just looking to gain leverage in price negotiations, then IBM “is no different than anyone else.”
That said, he likes what IBM is doing with the architecture, including having multiple ports in Power9 for different accelerator technologies, which could further enable manufacturers and system makers to more easily optimize the chips for particular workloads.
“What they have is a kind of Swiss army knife of accelerator ports,” Moorhead said.
The analyst said he expects to see the adoption of non-IBM Power systems grow in China, but that for the architecture to really take off, a major end user like Google will have to decide that it will move a percentage of its workloads onto the chips.
According to a roadmap laid out at the OpenPower Summit in April, IBM next year will release Power9 SO, a 14-nanometer design with 24 cores (twice that of Power8) and aimed at scale-out environments. Later in 2017, the Power9 SU with an enhanced microarchitecture will come, followed by a third version. Between 2018 and 2020, 10nm and 7nm “partner chips” using the Power8 and 9 architectures may be available. Power10 will come after 2020, sporting a new microarchitecture, officials said.