The PC's past and Intel's future

The leading purveyor of PC chips doesn't really think a lot about the desktop anymore.

Is the desktop PC on the road to oblivion? Well, let's put it this way: it's hardly an Intel priority anymore.

Yeah, desktops will still be around in 2016, but it's not something Intel -- which makes most PC processors -- thinks about a lot.

Survival in the age of the big-screen smartphone and tablet is what Intel thinks about.

A recent 75-page study from Goldman Sachs titled "Clash of the Titans" puts it, rather delicately, this way: "We believe the ongoing share shift in consumer computing toward smartphones and tablets and away from traditional PCs will be negative for Intel."

Of course, Intel has no intention of fulfilling analysts' dire prophecies. So, it has a newfound laserlike focus on mobile for its "client," aka PC, business.

And Intel's mobile future can pretty much be reduced to two code names: "Broadwell" and "Bay Trail."

Let's look at Broadwell first.

"With Broadwell there won't be a desktop update. Broadwell is focused on mobile," said a source whose company sells PCs in the U.S. and who gets briefed by Intel on future processor road maps.

That source calls Broadwell -- due in 2014 -- a "half tick," referring to Intel's Tick-Tock model for microarchitectural changes.

The half that's been left out is, basically, the desktop. Of course, a PC vendor could go out and buy a Broadwell circuit board and stick it in a desktop, but Broadwell hasn't been conceived with that in mind.

Click to expand...

It doesn't sound half bad, what do you guys think?

Read more at the source.

I think its a smart move. Probably 98% of consumers don't use any more computing power than is found in today's smartphones. They browse the web, play a few flash games, play some media files, maybe apply a preconfigured filter to a picture they took with their mobile device (I know very few people who still use a dedicated camera). For the consumer, portability and ease of use of top priority. With mobile chips powerful enough to do all of their necessary tasks, you can afford to get a few dedicated systems that will be easy to use: think mobile phone, tablet, media server, media players. With a half dozen independent (limited configuration) systems, software can be designed to automatically integrate into the home network...think Xbox Media Server but without the ability to render 3D games...easy and perfectly sufficient for the average consumer.

Now for the large corporation, the big contracts Intel is after, all they want is processing capability per $$$. While I don't foresee low power chips taking over the server market anytime soon, it may be worth trying to decrease the space and power consumption component of the $$$. If Intel can put out a chip with similar specs to today's server parts at half the wattage and space, they have just won a lot of corporate customers. Obviously they will have to carefully balance their processing power:Watt ratio as you start taking hits to power as you go down in watts, but I'm confidant (as I'm sure they are) that they can do that to yield a profit. They will still be producing full power chips as there will always be a need for maximum performance where shear performance wins out over efficiency and $$$ every time. Sometimes you just have to throw power at it to get the job done.

Now for us power consumer users. Sorry, but we are not a big enough market segment for Intel to really care about. And as we all have noticed, you don't need a state-of-the-art CPU to drive even the most demanding games. Sure they are happy to take their server tech and let it trickle down through the consumer line and since the majority of people still are based on desktop configurations, it makes sense for them to put some R&D $$$ into other consumer techs. Maybe a re branded server chip clocked a little higher. However, I do not see this mentality continuing much more than a few more generations. I foresee this middle market segment (between consumer chips and professional servers) disappearing from Intel's product line. In reality gamers can probably get away with future consumer chips (maybe a dual CPU system) and consumer power users (3D rendering, Photoshop, VMs, Data analysis) will invest in professional components as they are doing professional tasks anyways.

As always, us power consumer users, with more motivation than sense, will figure out a way to get the most out of the current tech.

For Intel, I think its a smart move.

Any realistic discussion of Intel's future needs to include Microsoft as well (IMHO.) To date they have 'grown up' together in the desktop market. AMD provided just enough competition to keep Intel moving ahead and on the S/W side there was similar ineffective competition (OS/2, Apple, Linux...) to keep Microsoft from getting too slothful. I know the full details of the situation are a lot more nuanced than this generalization but overall I think this is an accurate picture.

Microsoft is also moving towards a mobile platform with Win8, perhaps even to the detriment of the desktop market. However Microsoft is also moving toward the ARM architecture. If that proves to be more successful due to the head start that ARM provides WRT power usage vs. processing horsepower, the Wintel marriage could wind up on the rocks. This could even move into the server space where power consumption costs $$$ and perhaps even into the desktop market as portable apps become more popular.

On the flip side I see no reason that Intel could not use Android for their Intel based portables, but I have heard no mention of that.

Lastly, I am unaware of any reason that Intel could not license the ARM architecture and apply their own expertise to optimize it for their needs. The biggest obstacle is likely to be the Not Invented Here syndrome as Intel's engineers produce amazing products and I'm sure it would rankle them to have to adopt someone else' design.

And as far as our concerns go (e.g. having the most powerful computers and pushing them further up the horsepower slope) I think the processor is becoming less important. Two areas where computational power makes a difference are crunching and gaming. Both of these are dominated by GPU processing at the moment. Supercomputers which used to be huge highly powered processors are now huge arrays of PC class processors. However the Cray Titan (as of October the most powerful supercomputer) uses 18,688 16 core Opterons and 18,688 Nvidia Tesla GPUs. (I looked that up expecting to find a continuation of the move toward more smaller CPUs but found instead another field that is moving towards GPUs.)

Any way we look at it, the situation we've seen for decades - more powerful processors running more sophisticated software on the desktop - is likely to change radically. Left for another day is how the Internet and the Cloud are figuring into this as well.