by Anton Shilov
07/09/2013 | 11:20 PM
It is well known that chip designers like Advanced Micro Devices and Intel Corp. offer customized microprocessors to companies like Apple, Amazon, Facebook and Google in a bid to win large-scale contracts for their Opteron and Xeon products. However, the day, when web giants decide that the time has come for fully-custom, or at least semi-custom, chips for their scalable datacenters is not far.
Just several years ago three server giants – HP, Dell and IBM – generally commanded the market with industry-standard offerings that suited most businesses. Nowadays there are companies like Apple, Facebook, Google or Microsoft, who run massive cloud datacenters and require customized equipment. Some companies – Facebook and Google – even assemble their custom servers in-house. In a bid to solve specific problems, microprocessors inside servers for these large customers are specifically tailored for their needs. The second round of customization will not only involve some customization of microprocessors and I/O, but will likely include made-to-order chips with specially designed I/O and even processing engines, according to Andrew Feldman, general manager of SeaMicro and corporate vice president at AMD.
At present Mr. Feldman seems to be a strong believer in ARM architecture and its ability to be adapted for different needs. He foresees that many leading web-scale companies will need specially tailored processing cores that will deliver maximum amount of performance at minimum power requirements as well as very scalable input/output technologies that are built into chips.
“This vast change in the cost and time to market opens the door for large CPU consumers (mega data center players) to collaborate with ARM CPU vendors – say by paying part of the cost of development – in return for including custom IP that advantages the Mega Data center owners software/offering. There are conversations underway on this topic with nearly every major megadata center player in the world. The mega data center owners are building warehouse scale computers. It is not surprising that they have ideas on custom IP that would advantage their own software and would like that IP embedded into a CPU – ARM makes this feasible by bringing down the cost and development time of CPU creation,” said Mr. Feldman in an interview with GigaOM web-site.
The high-ranking AMD executive estimated that one could build an entirely custom chip using the ARM architecture in about 18 months for about $30 million. By contrast, it takes three or four-year time frame and $300 million to $400 million in development costs required to build an x86-based server chip based on a new micro-architecture.
Many datacenter giants already have a number of chip specialists and are looking at co-developing ARM-based chips that will take advantage of the greater levels of customization offered outside of the CPU so they can optimize for their own applications’ needs. At the same time, he does not reveal how will AMD and Intel behave in a situation when their largest customers essentially become their rivals in at least some ways.