ARM Community: ARM in Servers: A Different Path - ARM Community

Ian Ferguson reflects on his time during the PowerPC era and the future of ARM in the server market.

Yes, I have been a little delinquent in writing another blog. The idea for this blog was triggered after participating last month on a panel associated with “Extreme Low Energy Servers” at an Emerging Technology Summit in San Francisco. Included on the panel was Glenn Keels, who works on Project Moonshot at HP and Karl Freund, the VP of Marketing at Calxeda. I started to reflect on the number of times in the past twenty years that the processor technology I represented was going against an incumbent in a specific market segment…And of a number of times the competitor has been x86 technology. Those of you with as much gray hair as me may be able to work out which two companies I was with for those difficult and ultimately unsuccessful campaigns from my LinkedIn profile. Extra points for the names of those products!

My memory turned to my times around PowerPC^TM in the early ‘90s. As I look at the opportunity ahead for ARM and the Connected Community associated with us in the server market, I feel we can learn from our past as we plan the future. Of course, the push of the PowerPC alliance in the PC domain was ultimately unsuccessful. There were several “nearly” moments including the backing off (the day prior I might add) of a massive product release from a tier 1 company just prior to a major industry event. Some other great stories that will, at some point, make a great few chapters in a book, including why I believe that although the specific market foray was unsuccessful, Motorola Semiconductor gained technology benefits from the alliance with IBM which paid out a massive return in different ways…but I digress…

The original value proposition that PowerPC offered was a 2:1 performance/price ratio over the incumbent. It would, we argued, be possible to deliver twice as much performance for the same price as an incumbent platform. Hindsight is, of course, 20:20. One of the pieces Motorola missed was that at the system level it was only the processor that changed. Because the remainder of the components needed to complete the PC on the motherboard remained unchanged from an x86 platform, the perf/price ratio was very challenging to meet. Higher levels of device integration would have been key. Motorola viewed the 2:1 perf/price advantage based on what was available on the market at that time as opposed to what coming next. So in reality, a greater benefit was needed to displace an incumbent. While getting an operating system running on a new processor architecture was relatively straightforward, the porting of legacy apps is extremely challenging. In the mid 90s, there was little in the way of open source and this is one of the major barriers that (for some areas of the server market) make its pursuit easier today than it has been in prior decades. Finally, a new player needs to bring something new to market that is of high enough importance to cause customers to consider making a switch. In the 90s, the key market care-about was performance. And ultimately, this was a vector that the PowerPC alliance struggled to successfully execute on. The disruptor now is power as datacenters start to be considered as “energy constrained systems”. The energy “bucket” available is certainly larger that a phone, but fundamentally, power is a capped resource inside which a company has to run its business. This new vector plays to ARM’s 21 year heritage of developing energy sipping technology. Couple this with the drive toward highly integrated “server-on-a-chip” components and we feel we are on a path to offer the market an exciting set of new option from which to select.

As I wrote last year, I am tremendously excited about the road ahead of us. There are challenges ahead for sure. But unlike previous crusades, the market is crying out for solutions on a vector that the company I work for is categorically best-in-class at delivering. Servers powered by 32-bit ARM technology will ship for revenue for the first time this year. “So what about 64-bit?” I hear some of you ask… Well, yes, one of the activities my team is working on this year is readying the ecosystem, particularly software, for 64-bit (AArch64 is the official term) ARM technology. Building an ecosystem takes time. And with ARM now being in a position to share models and open source deliverables, we are in a position to start the hard work. ARM has set the expectation that the first 64-bit devices will start to appear in 2014. Two semiconductor companies, Applied Micro and NVIDIA, have been announced as architecture licensees. This means they are designing their own CPU processor cores that are software compatible with the ARMv8 instruction set architecture (which includes AArch64 and the ARM32-bit ISA). In our earnings announcement on January 31st, ARM provided a glimpse of our roadmap, identifying the first two 64-bit processor cores that will be available to be licensed by companies from ARM. More details of the functionality of these cores will be announced by the year end.

At the moment, much of the recent US sports chatter has been about “Linsanity”: A basketball player called Jeremy Lin has come from nowhere and has had an incredible (albeit a relatively small) run of games for the New York Knicks. Media coverage is everywhere. Unfortunately, I am somewhat concerned that he will quickly become less effective as teams start to work out how to combat the new star. Time will tell of course. In a way, I want to liken this to ARM’s pursuit of the server market. The difference being that we are not looking at this as a near-term boom/bust engagement. This is a market program that will take years to roll out. We are committed to the pursuit of this course which we expect will yield more silicon, more OEM platforms and more software all with the key beneficiary being the end user in terms of choice.

Ian Ferguson, Director of Server Systems and Ecosystem, ARM, has spent years fighting from the corner of the underdog. Most of those scars are healing nicely. Ian is particularly passionate about taking ARM technology into new types of applications that do not exist or are at the very formative stages. Consequently, he is driving ARM’s server program with a view to reinvent the way the server function is implemented in networks as opposed to simply replacing incumbent platforms.