Smartphones and PCs

There was recent big news that shipments of smartphones passed PCs. So while I don’t believe the PC is dying, I do believe there is a message here. The volumes were interesting too. Smart phone volumes topped 100m units for the first time, while PC units were a still respectable 92m units.

So is that the death of the PC? It is true PC shipment growth has slowed in terms of the past, but it still shows respectable growth. IDC stated shipments were 346.2m units in 2010, a 13% growth factor. Overall growth from 2010-2015 is expected to grow 6%, as per IDG . That growth curve yields roughly 450m units in 2015 for the PC market. Any more growth, and 500m units is possible. A market that is 500m units is not a small market. PCs are a valuable tool for many businesses and many people. And it looks very likely to stay that way. What is more possible in the PC market is the pressure on desktop sales as laptops (and net books) become powerful enough for many “knowledge workers”. That puts pressure on ASPs for the silicon manufacturers and on design teams, but doesn’t change the fact that there is a bit market out there.

What we are seeing, then,in my opinion isn’t the death of the PC. It’s the emergence of something new.

Cell phones aren’t new. And smart phones aren’t particularly new. But the level of experience Apple brought to the market with the iPhone, and the integration of services in iOS and the SDK provided to developers are new, as is the idea of a central shopping place for customers in the app stores. And Google’s response with Android created a market stratification similar to what existed in the PC market years ago. Apple took their usual path and created a controlled, garden and closed market. Google took the Microsoft path with their uncontrolled and somewhat chaotic open market. The space Google carved out in the mobile market is the space that used to be owned by Microsoft. It took Microsoft 3 years to respond, and Windows Phone 7 was a worthy response, but it remains to be seen if there can be 2 or more players in either the closed or open markets.

That feels new to me. And the level of experience has energized the smart phone market. It used to be that smartphones were 10% or so of the market. In 2008, in Q2 smart phone sales were 32.2m units, with total sales of 139m units. In 2009, they were 17%, roughly 170m units of smart phones out of roughly 1b total units. Thats 50% growth. Last year, they were roughly 20%, about 350m units so smartphone sales almost doubled last year. Streaming and location services are offering personalized services like never before. With streaming services like Pandora and, you can get a personalized music stream on your device, whenever you want. With location-based services you can get turn-by-turn navigation for car travel and unique services like applications that use your GPS location and geo-tagged businesses to tell you what bars, restaurants, cultural, or entertainment venues are near you when you walk around. These types of services are new things that were not offered so easily to the masses in the past, and the level of convenience they provide is part of that new experience. And those aren’t the only ones, just two quick examples.

These examples speak to part of the integration of services in the smart phone operating systems, as well as how well they are exposed to developers in the device SDKs. And the idea of an “app store” where the platform owner provides a built-in ecosystem to reach the consumer is also very new. So in a extremely rapid manner, developer ecosystems have sprung up. Having started Windows programming in 1987 with Windows 1.03 and having lived through how long it took both the Mac and Windows ecosystems to become well-formed – this is unusually fast.

So I am sure some of you out there are asking yourselves “very interesting Phil but what does this have to do with gaming and graphics samples from Intel?”.

Now, maybe not so much. But over time, quite a bit. Intel is investing heavily in Atom to have a part that is able to address that market. And over time, Atom will be refined to hit the mark. It is already starting to reach where it needs to be, and the design wins that are starting to appear are an existence proof of that.

And over time you can expect to see that focus be reflected in the gaming and graphics samples. First, we need to take care of home base, the PC. And this first year you can expect that to be our target. Over time, though, it should not be unexpected to see this team “dip their toes” in the Atom/device space and as that space becomes an increasingly important market to Intel that the gaming and graphics samples would adjust to include that space as a target.

How fast and how much, it is too early to talk to. But over time adding this space should not be a surprise.

I’d be interested in hearing in the comments what sort of samples developers feel would be helpful in that space. While this isn’t a commitment to implement what we hear in any dialog, it is always good to listen to the customer. And don’t be afraid of radical suggestions. For instance, would standardizing on OpenGL-ES even on the desktop be useful, since that is the standard for devices? OpenGL-ES 1 or 2? How important do developers feel WebGL samples be in 3 years? These are the sorts of discussions I would love to have.

For more complete information about compiler optimizations, see our Optimization Notice.


anonymous's picture

Yeah, I think that, while PC is slowing down a bit, it won't be replaced any time soon. Even then, smart phones are bringing out new and innovative technologies now like remote deposit capture, and I'm sure that helps them out a lot. Gthomaz also makes really good points as to why this happens; PC's take much longer to become outdated and so are replaced less often. Thanks for the post!

gthomaz's picture

Hi Phil!
Great post! I agree with you.. the market is changing and we need to be smart. We need think about..

But, the fact that the shipments of smarthphone are greather than PC's is not a big surprise.
First: People don't upgrade your's PCs so frequently as buy a new mobile/smartphone.
Second: In general , one family has 1 mobile (or two) for each person. But, in another hand, a ''standard family' (with 4 people) don't have 1 PC for each person. So.. for me it's quite logical..
I don't know around the world, but here in Brazil smartphone market is growing a lot (and the prices going cheap, thanks god!). Pc's market still the same...

philip-taylor (Intel)'s picture


would a sample around MLAA, as in here

be useful?


steve-hughes (Intel)'s picture

Hi Phil

On PC you can get revenue from game apps in 2 different ways. One way is to write a middle of the road app which spans a wide range of platform specs, the other is to build a cutting edge monster that makes everyone wonder how they made the top end platform sing like that. Both are equally valid as marketing techniques.

I suspect we will soon see the same behaviour in the Smartphone ecosystem. Some developers will build apps that span the whole gamut of devices while others will target specific systems with “blow your mind” apps. Yes, I agree that a good cross platform dev tool chain is great for the first kind, but we should not neglect the second type, who will be pushing the envelope by writing the technological equivalent of “Crysis” on Smartphone in the coming years. These apps will push the ecosystem to greater heights.

Developers need access to platform specifics and relatively low level code paths if phone games are going to shine.

Great Blog BTW.


philip-taylor (Intel)'s picture

that is roughly what XNA Game Studio gives you on Windows 7, XBox 360, and Windows Phone 7, if you can live with the subset of features it provides.

the integration of touch and accelerometer like I mentioned isnt there yet, but its a decent start. it does have an interesting idea wrt the content pipeline too.

how important is the content pipeline?

It seems to me that the goal of one code path might be extremely hard, and that the goal of one art asset database and a content pipeline that targets, or retargets, different form factors might be merely hard.

how interesting is an example of how to do that, in an SDK?

numerator747's picture

An API that allowed you to target PC and mobile would be a good thing. Debug on the PC, deploy on mobile.

anonymous's picture

I do not think the number of sales reflect the use of PCs. I think it reflects how often the people are upgrading their PCs. We are stuck with transistors for the foreseeable future. This is why Intel and the other chip manufactures are going to multi-core processors as the enhancements to transistor processor speeds are maxed out. People really are not seeing the reason to upgrade. Only when the computer starts to break down do they consider upgrading.

The phone technology is still advancing. This gives the consumer a reason for upgrading. This is why you are seeing large numbers of sales coupled with the fact that most phone purchases are subsidized by the phone companies.

I do not know if Intel or any of the other chip manufactures are investing R&D dollars into a replacement to the transistors. I believe this is where we need to go to see great sales realized again.

anonymous's picture

Hi Philip - great post! I found your posts through and am looking forward to reading more from you.

PCs are dead, PCs are dead, PCs are dead. If it's repeated enough, it must be true, right? Let's hope not, because it sure is being repeated a lot. I think desire for PCs may be on a permanently downward trend for that considerable segment of the population with minimal computing needs, but there will still be plenty of us left who want to do more than just browse the web. Hopefully.

Still, viable cross platform APIs would be so welcome... I see slow but steady progress there. The app I work on would really benefit from running on phones and (especially) tablets, *in addition to* PCs. A lot of people are very interested in that. But until the cross platform / fragmentation situation gets better, it seems Windows remains the only platform that's viable to support for this app. It's too bad because the hardware in modern smartphones and tablets seems capable enough, especially compared to Intel graphics on Windows. It's just a software problem.

Aside from cross-platform, performance and feature set are also very important. Intel graphics have typically been around 5-10 years behind the competition in these areas. Even if I made an effort to get my customers to avoid them, plenty would still have them - especially on laptops, and of course you can't upgrade graphics in a laptop. So they're the minimum standard that I now have to code for and test on. I say 'now' because for over a decade I had to avoid this problem (among others) by relying exclusively on software rendering (via D3DRM, remember that?). Finally I've had the opportunity to rewrite this app and make a return to hardware rendering. But with shaders now. Such luxury!

So I've tested my new app on some relatively recent Intel graphics hardware such as the X4500 and 3150, and the results weren't too bad. They're definitely usable and the performance isn't bad on lower resolutions and simple workloads. But there's still no support for anti-aliasing or non-power-of-two textures, and the texture filtering could be better. I'm looking forward to trying the graphics in Sandy Bridge as it sounds like some big improvements have been made. I was thrilled to find confirmation that anti-aliasing is finally supported, but I still don't know if the performance is good enough to be usable for an app with modest demands such as mine.

Some kind of a sample to show how to get the most performance out of anti-aliasing on Intel graphics might be helpful. Not sure until I get some hardware to play with, though. Preferably a Windows tablet with capacitive touchscreen, good viewing angle, multi-touch capable out of the box, Sandy Bridge, usable anti-aliasing, respectable battery life, weight under 5lbs, and pricetag under $700 or so. Is it asking too much? :) (Pity Z670 will still be 3150, but hey.. 3W, eh?)

philip-taylor (Intel)'s picture


Eventually standards that help with what I call the "spanning the device continuum" strategy will appear.

OpenGL-ES is there already for 3D graphics. OpenAL exists for audio, but isnt as widely available as OpenGL, yet.

A cross-platform Input library is an obvious next step, as in something that maps key input to keyboards and virtual keyboards, pointing input to mice/trackpads/touchpads ( and gestures across all of them )and spatial input to joysticks/accelerometer input so developers dont have to understand all the details on every platform would be nice.

SDL does some of this today for graphics/audio/input, but then there is still the underlying platform services.

One good thing is it means there is plenty of interesting work. The converse problem would be worse.

gaston-hillar's picture

Philip, nice post. I agree with your comments: It’s the emergence of something new. In fact, the PC won't die because developers need a PC to create their software. So far, it's impossible to code with a tablet.
However, all these new platforms with so many different operating systems are a nightmare for developers. We're always working with alpha versions, beta versions and slow emulators that make it very difficult to be productive. When you finish learning one platform, a new one appears. Are we going to be able to survive this fragmentation?



Add a Comment

Have a technical question? Visit our forums. Have site or software product issues? Contact support.