Last month's move by AMD to license the initial 64-bit ARM cores for its server use could signal a brave attempt towards a new direction, as well as finally giving up any plans to competitively fight Intel on the X86 performance front. Aside of the still relatively healthy GPU business, what is the impact on AMD futures? As a follow-up to yesterday's history lesson, we look at three possible scenarios that could happen for AMD.
So, at least looking at the CPU business, AMD is stuck in a very unenviable position right now. The X86 competitive position vs Intel is likely the worst since the Athlon days a decade ago across all market segments except ultra low power, with no near end in sight, while their ARM entry will bring multiples of that competition on both business and technology fronts. On top of it, taking ARM now may make the pundits ask whether AMD is really committed to stay in the X86 at all. The company has been forcibly bleeding, unfortunately, some of their best engineering talent, yet acquiring resources at high price – Seamicro etc – which may not be such a fit after all, especially if the CPU problems aren't solved.
Nevertheless, let's take a calm look at the possible scenarios from now on:
1 – 'Gung Ho' approach reversing the brain-drain: the management would go all-out and somehow raise the money to create three parallel beefed-up top CPU core design teams: one for high end X86 for the Excavator and beyond; another for the ultra low power X86 'Jaguar' follow ons for tablets and such; and finally, proper ARM design team for their own custom core. All these, of course, also assume investment in 'uncore' and chipset support teams to support the each core team fully, for complete up to date system platform offerings.
This would be the optimal choice to give the market and the user community renewed confidence in AMD as a CPU vendor, but will need to overcome two major troubles. One is the financial one, which only Arabs can seemingly solve at this stage, since the US government wouldn't let the Chinese buy AMD – and they're not interested anymore, anyway. Of course, the implication of the resulting Abu Dhabi controlled strengthened 'Arab Micro Devices' vs very Israeli friendly Intel, where Intel Israel saved the day a decade ago with the Core design, and whose rumoured next CEO in line is whispered to be Dadi Perlmutter right from there, would give interesting twists to any future major Middle East conflict, and those come on regularly, as we all know.
The other trouble is the management attitude. AMD would have to bring back someone like Jerry Sanders or alike to lift the spirits. Rory could maybe do a part of the 'spirit lifting' job with his speeches, but he's not a semiconductor guy, let's be honest. At the very least, he'd need decision making assistance by some proper semicon industry experts, possibly outside the current AMD team, but who believe in the company. Managing and supporting three CPU teams plus a GPU team, including the APU Fusion parts in more than just client market, would be a tough job, but it may be the only way AMD re-gains the technical excellence needed to be taken seriously in the market. Naturally, this would also mean strengthening that GPU capability as well.
2 – 'Graceful Shrinking' of a managed sort, let's say: The first step could be – as our friend Charlie said at Semiaccurate – ceasing the high end X86 core development post today's Piledriver, thus leaving the company with just low end X86 and ARM. If the management feels that the high end battle is gone – since Steamroller is another small interim step, and Excavator is two years away, unlikely to beat Intel unless revolutionary improvements in place – this may make sense: just focus on the still promising X86 ultramobile, tablet and smartphone market for the next few years, while in parallel somehow beefing up the ARM design capability to match the incumbents in that space.
Over time, if the Windows fades away as, rumours say, its creator Bill Gates is now seemingly focusing on reducing the world population by deadly vaccines and chemtrail vapours (not to mention hoarding the natural seeds for the rich in the Arctic vaults while the rest of populace get poisoned by GMO), the X86 will become even less important, and the low end X86 team can go away as well, or be converted into the second ARM team – so a possible future AMD would have two ARM teams, one for low end mobile; another for ARM servers. And that's where that Seamicro thing may finally become really useful beyond servers – how about making clusters of smartphones, each with two low power Seamicro interconnects outside, to link to other phones directly at the CPU level, bypassing the vendor, OS, security and other limitations?
3 – "Gradual Collapse", also of a managed sort, like Digital or Compaq in the past, for those who remember. Killing off the high end would again be the first step, including this time attempting to sell off the related IP as well. This would be followed possibly by an announcement to, say, sell off most or all of the ATI GPU division – few years ago, only Intel or Nvidia would have been the prospective buyers, but now half a dozen bigwigs would be queuing up for this jewel, including Samsung, Qualcomm and, yes, the Chinese – the Beijing brethren still miss the high end GPU in their portfolio as of now. AMD management and shareholders need quick cash, and this is one sure way to go about it.
Mubadala sends a message to AMD employees
This would be followed by turning off – or likely selling off – the low end X86 division. VIA could maybe step in here by asking its rich parent Formosa Plastics for some cash to consolidate its low power X86 focus and become truly competitive, and finally give Taiwan a noteworthy X86 CPU platform capability. Heck, they could take the high end stuff at the same time, and, as the vendor is still comparatively small, I don't think Intel would have problems allowing them to keep the AMD X86 license extensions, especially to avoid any monopoly related questions from US FTC.
Finally, a slow or underperforming release of the first ARM product – or even no release at all – would be the 'icing on the cake', leading to the dismissal of the rest of the team, if they can't be sold off to another ARM vendor. Seamicro interconnect would then likely be sold off to someone else, likely at minor fraction of what AMD paid for. And, that's about it – AMD would be wrapped up and history by then.
Why not an AMD Alpha?
Actually, there is a very unlikely – but actually very plausible – scenario as well. Remember AMD has/had quite a few of the top Alpha CPU designers, and what was the world's best CPU architecture, untimely murdered by the corporate merger shenanigans, is now revived very well in China, with 16-corers and beyond?
Well, the solution is simple. Give up X86 altogether, and make AMD an all-RISC company with two platforms: renewed Alpha (in partnership with Samsung, once the Alpha 2nd source, and maybe the Chinese) at the high end, using the Hypertransport and Seamicro for various interconnect options, and ARM at the low end, both server and mobile. I am very confident AMD Alpha team could create cores, and uncores, that would not just beat anything Intel has, but also put IBM Power 7+ and its successors, currently the worlds fastest CPUs, to shame. It would ensure long profitable hgh-end future for AMD too.
After all, it's Linux open source time, and the dependence on an instruction set to get commercial software vendors' ports, is lesser than ever. How about this?
Personally, I hope to see the Scenario 1 happen, for the better of whole computer industry. Over the past 30 years, personal computing has moved from 'space age' respected industry to local hoodlum (Ah Beng in Singapore & Malaysia) car repair shop level of appreciation, no thanks to the rapid decrease of both technology innovation and true competition, coupled with consumerism dominationg the whole industry now. The decline or death of AMD would bring us further to 'one CPU, one OS, one everything', sea of remote controlled Androids (humanoids, not just phone OS), controlled by malicious One World government, by technology monopoly, among others – sounds familiar?
Part 1 of "ARM – Possible Beginning of the End for AMD?" can be found here: http://vrzone.com/articles/arm–possible-beginning-of-the-end-for-amd–part-1/18054.html