AMD, the processor and server vendor on the path of diversifying its product portfolio to rely less on x86 products – a market it has been unable to compete with Intel in – said it expects to start sampling its first 64-bit ARM server processors by the end of this quarter.

There is an ongoing race among chipmakers that license the ARM architecture to bring 64-bit chips to market since 32-bit ARM failed to gain any serious traction with server vendors and customers. Other companies in the race include AppliedMicro, Broadcom and Cavium, to name a few.

All these players are also competing with Intel, whose answer to ARM has been a 64-bit Avoton System-on-Chip based on its Atom architecture. This competition will not be easy.

The Atom 2000 part’s SpecInt rate was 106 at 20 watts during a benchmark test, according to an Intel spokesman. AMD’s upcoming eight-core part’s SpecInt rate was about 80, running at 25 watts. SpecInt is a benchmark specification for CPU integer processing.

In December, the 64-bit ARMs race saw its first casualty. Calxeda, a Texas startup that was expected to be among the first to bring a chip to market, shut down, having failed to raise enough funding to keep going.

AMD has been extremely vociferous about its ARM ambitions and Andrew Feldman, general manager of the company’s server business unit, has been bullish on the architecture’s future in the data center market.

During his keynote at this week’s Open Compute Summit in San Jose, California, Feldman said ARM would command a quarter of the entire data center market by 2019. The architecture’s success will be due to its low power requirements and relative ease of customization.

“Custom ARM CPUs [will be] the norm for mega data center,” he said. “And AMD, we will be the leaders in ARM CPUs for servers.”

The summit was organized by the Open Compute Foundation, which oversees an open source hardware design community formed by Facebook in 2011. In addition to announcing the deadline for availability of its 64-bit ARM samples, AMD announced that it would contribute a new server design based on its ARM chips to the open source project.

The company’s 64-bit ARM CPU is the Opteron A1100 Series, codenamed “Seattle.” It supports four- and eight-core ARM Cortex A57 processors, up to 4MB of shared L2 and 8MB of shared L3 cache.

Along with sampling the processor, AMD said it would provide a development platform, including an evaluation board and a software suite. The software tools include a standard UEFI boot and a Red Hat Linux distribution.

Providing a sample platform and tools for developers is crucial to success of the architecture, since it is new and ARM and its licensees will have to work hard to convince developers to try the new architecture after the industry has been standardized on x86.

ARM, the UK company that licenses its core architecture to companies like AMD, is also actively trying to make the product more attractive for developers. This week at the Summit the company announced a standard for ARM SoCs to bring some uniformity to the ecosystem so that developers do not have to rewrite code for every vendor’s product.

rLakshmi Mandyam, director of server systems and ecosystems at ARM, said the standard was developed together with software companies like Canonical, Citrix, Red Hat, Microsoft and Suse, hardware vendors HP and Dell, as well as chipmakers such as AMD, Applied Micro and Cavium.

The work focused on “areas where having a standard simplifies life from a software perspective,” she said.