With over nine billion processing cores shipped in 2012, ARM is without any doubt, the largest core logic designer on the planet Earth. Yet the company is often attacked by its competitors that they are behind the curve in terms of manufacturing process, which should lead to company being less than competitive in the high performance field.
That FUD (Fear, Uncertainty, Doubt) falls apart when we are discussing about deployments of millions of cores on systems such as Intelligent Traffic Grids (for example, soon-to-be-built Intelligent Highway in Netherlands). Still, ARM needs to compete on the high end as well, and here is where today's announcement fits perfectly. The company announced that it has completed the first tape-out using the 16nm FinFET Technology at Taiwan Semiconductor Manufacturing Company (TSMC). The Cortex IP selected for the task is the upcoming 64-bit ARMv8 architecture, dubbed Cortex-A57.
ARM Cortex-A57 is the first processor using the 64-bit ARM instruction set (ARMv8)
The technology is now available to all of its customers, which is something Samsung, Qualcomm and Nvidia will all take note. However, given that both Nvidia and Qualcomm are very busy with their own custom architectures, and Samsung is a part of Common Platform manufacturing alliance - the biggest winners in this round might come from mainland China. HiSilicon (a division of Huawei), MediaTek, Rockchip all stand to gain big form getting access to 16nm FinFET technology and the following technologies, which are all on offer:
Alongside the press release, we received the following self-congratulatory quotes from the company representatives.
- ARM Artisan physical IP libraries
- TSMC Memory Macros
- TSMC Open Innovation Platform
Tom Cronk, Executive VP, General Manager, Processor Division, ARM Holdings:
"This first ARM Cortex-A57 processor implementation paves the way for our mutual customers to leverage the performance and power efficiency of 16nm FinFET technology. The joint effort of ARM, TSMC, and TSMC's OIP design ecosystem partners demonstrates the strong commitment to provide industry-leading technology for customer designs to benefit from our latest 64-bit ARMv8 architecture, big.LITTLE™ processing and ARM POP™ IP across a wide variety of market segments."
Dr. Cliff Hou, Vice President of R&D, TSMC:
"Our multi-year, multi-node collaboration with ARM continues to deliver advanced technologies to enable market-leading SoCs across mobile, server, and enterprise infrastructure applications. This achievement demonstrates that the next-generation ARMv8 processor is FinFET-ready for TSMC's advanced technology."
All in all, the scene is getting heated. Intel plans on coming strong with their 22nm and 14nm SoC, facing continuously moving target and going head to head against ARM using the same graphics IP (Intel mobile hardware uses Imagination Technologies PowerVR GPUs), Common Platform camp is pushing for 20nm and 14XM, while TSMC is now firmly planted with its 16nm process. Furthermore, given that Intel’s 22nm FinFET process was targeted by SIA as a 26nm, the question is who is going to start the paper wars and claim that 14XM is not 14nm, Intel’s 14nm is 16nm, TSMC’s 16nm is just 16nm...
Yes, we are heading into a silicon manufacturing battle royale.