A number of us received AMD’s roadmap update late last week ahead of company's official announcement regarding its server portfolio. Towards the end of the deck is slide 33 (see below) which provides with a very clear indication of where AMD wants to go in the long run.
Entitled “Why ARM will win in the long run?”, the company claims that ARM’s strategy is the one which will win, ultimately. Smaller, Lower cost and higher volume CPUs have always won, it says.
Perhaps more importantly is the claim that OEMs, ODMs and largest customers want ARM to win, presumably against Intel (not sure whether Imagination Technologies or MIPS is in the picture).
The first two players have already seen the kind of innovation that the ARM ecosystem is capable of. After all, the overwhelming number of tablets and smartphones are built around ARM-based solutions.
When we talked to Andrew Feldman, the former CEO of SeaMicro who now heads AMD’s Data Center Server Solutions Team, he confirmed that the 28nm Seattle will be sampling in the first half of next year and will be available in the second half of 2014.
Faster processors and smaller geometries are not what will make processors faster in the long run. “We need a different mindset” he added.
He also told us that “Warsaw”, AMD’s follow up in the 2P/4P server market, will cater for a declining but still large market with institutional customers that will take longer to migrate to an ARM ecosystem. Andrew made it clear though that AMD is not giving up on the “2+ socket” market, at least not yet.
Seattle, AMD’s first 64-bit ARM server, will provide with up to 4X the performance of the Opteron X-Series (for 16 Core SoCs with support for 64GB DRAM). Expect the A57 to debut at 2GHz and to include a flurry of IP AMD gathered from Seamicro and others.
AMD is currently worth just under $3 billion, which would make it an attractive target for any company, like Oracle or IBM, who would like to acquire a “proven server processor supplier”.