Trouble May be Ahead for Intel Medfield's Fight Against ARM
2012 is generally seen as a do-or-die year for Intel as far as its smartphone and tablet play is concerned.
Next year, Intel will have to show that it can deliver on promises of a competitive tablet and smartphone chip platform to be able to compete with ARM vendors - especially since ARM vendors will be encroaching on Intel territory with the release of Windows 8 for ARM.
A Digitimes article suggests that Intel's 32nm Medfield platform will have a tough time to gain traction as Intel missed the chance to establish close relations with first-tier smartphone makers. Its existing ties with notebook makers may not be enough as most of them are just recovering from failed smartphone and tablet products and may be reluctant to risk more of their money in a market that is dominated by Apple and a few Android handset makers. The fact that Intel said that it has Android 4.0 ICS running on Medfield may not help much and the article suggests that Samsung's commitment to Medfield may simply be a strategy of covering its bases - just in case Intel succeeds.
Intel's market entry is largely speculation at this point and will strongly depend on the capability of Medfield in performance and especially power-saving disciplines. Even more important will be Intel's ability to establish credibility in a market it has not much to show for. Its pitch of manufacturing power may not work so well as ARM vendors have not had troubles supplying cutting edge processors so far and Medfield is actually a process generation behind its desktop/mobile chips, which will be shipping in 22 nm when Medfield becomes available as a 32 nm platform - a circumstance that will cost the smartphone/tablet chip some pizazz.
We have not seen Intel, when it played the role of an underdog, failing more frequently than succeeding and there is clearly doubt that it can compete with all ARM vendors at the same time. However, we also know that Intel has substantial design and manufacturing talent at its core and that it is at its best when it is under tremendous pressure. It may be too early to discount Intel's opportunity in the smartphone and tablet market.
As a side note: anyone has links to benchmarks of Atom vs. ARM? I know that Atom is useless, but I want to see how it compares to ARM... not that you can directly compare the two since these are two different platforms, but still - something?
Windows 9 comes out with an Intel double die shrink between 18nm - down to 14nm
No ARM manufacture joint effort can help here: Intel wins in five years, in ALL fronts!
But you of course buy what is good now, right? Well...it's not Intel right now, is it?
1, they can't deliver at a price point to displace ARM in the smartphone / tablet market.
This is some what likely in my opinion, given that Intel is one company VS several, as ARM licenses its technology out to others while Intel is the sole developer and manufacturer of said technology.
2. They deliver at a competitive price point, and deliver matched or improved performance. This could be potentially disruptive to the current "ecosystem" of mobile devices, since joining mobile devices with current desktop PC architectures simplifies development to some degree, and would potentially change the battle as far as desktop / mobile unification. Especially seeing as how Windows 8 is being designed as a universal OS, but basically being an entirely separate fork from its x86 twin. That starts the similar issue apple faced when transitioning from the powerpc arch to x86. For years they had to build 2 versions of everything, as did developers. Not an elegant solution.
However it remains to be seen, since things do tend to gain a momentum of their own at some point, and we either very VERY close to being beyond the point of any return for an architecture switch in the mobile market, or already past it
My .02
I am curious to see how ARM's power requirements and cost scale as it approaches competitiveness with Atom class workloads. That may be the best indicator or whether this will be a prolonged conflict and will will prevail.
Second, let us assume for a moment that they fail bigtime in the mobile sector. All they would have to do is sign onto an ARM use agreement and start mass producing. They would kill off every other ARM producer in the market, and then they could kill ARM and force everyone back onto x86. I love Intel, but at the same time I would not put this past them.
Either way, they will win in the end.
Lastly, the article mentions large phone companies saying "Apple and a few Android handset makers" When I think they meant to say "HTC, a few Android handset makers, and Apple" if they had meant to be accurate about the market today.
X86 architecture is over 30 years old, it has a very poor memory model and really should be replaced. x86 has been kept breathing by Microsoft refusing to develop on other architectures for very sensible commercial reasons, and only the rise of Android and its collection of Apps has suddenly spurred Microsoft to realise there is an alternative. ARM has its limitations and is hardly new and cutting edge but it is FAR in front of x86. Intel recognised the issues when they bought/developed x-scale based on ARM 5 but lacked the insight to turn it into the product it could have been. I bet some executives are really pleased that they sold the x-scale PXA line to Marvell in 2006 just as the mobile revolution took off... Intel still holds an ARM licence, dont be surprised when Medfield fails and Intel launch there own ARM 9 design!
There's one thing that ARM can't do that x86 can do.
Compatibility. Even if MS adds compatibility to ARM, what about other software vendors? It cost a pretty penny to design a program to run on a completely new instruction architecture.
I don't care if there's a GPU that can run around an OC'ed quad Crossfire of Radeon 6970s or SLI of GTX 580s, consumes only 1 watt under load, cost only $10, and can ONLY RUN GLIDE API. No modern game uses Glide API anymore.
Are you crazy? A windows tablet can do EVERTHING a droid or ipad can do. You can browse the web and play angry birds, and it won't feel any different. It is right that Photoshop, MS Office, or Visual studio is not as nice on a tablet, but it's better than not being able to use that software at all.
This is one example that clearly shows that just because Intel throws a ton of money at something that they aren't going to automatically succeed as a result.
True, it's not like Intel has sold more GPUs then other companies or anything... o wait they have, yea their graphics suck but they still outsell all others
I think compatibility is becoming less and less important as the mobile realm is drawing out its own software market distinct from the desktop and building its own APIs around its particular advantages. There is tons of software out there for smartphones and tablets and they fill the current needs of the market. If the needs change, the developers will respond because they will step in to fill those voids and earn the bucks as quickly as they can. You don't have to wait for some software vendor to port its application to a mobile platform. If they don't do it, and there's a consumer demand for that kind of software, someone else will write it and snag the proceeds.
So yes, you can't run modern GTX 580-type games on a tablet... but that's not a problem because nobody buys a tablet to play high-end games developed for DirectX or OpenGL: the mobile gaming market has its own set of tools and APIs that fill the current needs for gaming development. Just like nobody buys a tablet to do high-end video and audio editing. It's the wrong tool for the job. Likewise, you can't just port something like MS Office from x86 to ARM. You have to adjust the interface to take advantage of the features that mobile platforms provide, e.g., touch and gestures. A straight port of MS Office to ARM would be a disaster. The office app that offers a mobile-friendly interface is the one that's going to win the day. IOW, desktop apps belong on desktops and mobile apps deserve their own development methodology. This is why porting and compatibility are becoming less important to developing new from the ground up. Why would I want something as huge and bloated and old as MS Office on my tablet? (Other than the fact that I'm already locked into a vendor-specific proprietary document format. But people are becoming less tolerant of this restriction of freedom, since they want their information to be accessible on a variety of different platforms and devices.) Give me something fresh and new and something that's specifically designed for a mobile interface.
As a developer, if you think there is some kind of software out there that millions of mobile consumers would love to have, but doesn't yet exist... please let me know and if it's feasible I'd be happy to pursue it myself and split some of the proceeds with you.
"We have not seen Intel, when it played the role of an underdog, failing more frequently than succeeding and there is clearly doubt that it can compete with all ARM vendors at the same time. However, we also know that Intel has substantial design and manufacturing talent at its core and that it is at its best when it is under tremendous pressure. It may be too early to discount Intel's opportunity in the smartphone and tablet market."
What a load of rubbish. What history has shown us is that Intel can't do anything well without a 3 year head start on the competition and a full blown monopoly that they can control with their patents, i.e. integrated graphics, discrete graphics, phones, mainframes, HPC accelerators.