Intel profit sinks 27% on dreadful PC sales

[SiliconWars]

Being x86 doesn't give much of an intrinsic disadvantage for either in the markets you're specifying, not vs its competitors anyway.

I admitted that the dependence is loosening, but that doesn't mean that an x86 processor can't be successful in tablets and phones and various other platforms. Anyone who thinks otherwise must be in denial since existing hardware already contradicts this.

But like I alluded to, it's not just about performance. How much do you think the "intel tax" is worth in the current market? Nothing basically. Outside of the tiny enthusiast market, who actually buys a tablet that isn't an iPad or Nexus?

Intel's prices are too high, even if the technology is superior. ARM told them long ago that this was the case. Their business model needs changing in order to compete.

AMD are just hoping to make survival cash on what they can get. x86 is utterly irrelevant in this space.

 

[podspi]

Microsoft has a role in this slump besides the fact that most mainstream consumers are content with the capabilities of a tablet. Had Windows 8 been made properly, the sales of PCs would've at least stayed the same.

I like Windows 8 a lot. The main problem it has is that it doesn't offer much for desktops. I feel for tablets/convertibles it is much, much better.

 

[Soulkeeper]

I don't buy into the whole "post pc era" nonsense
I'm more inclined to support the "pc plus era" as, I believe, the acer CEO recently coined.

 

[georgec84]

PCs aren't going away. They simply don't need to be bought or upgraded at the rate as they did maybe 4-5 years ago to keep pace. Today's CPUs and GPUs seem to have better "longetivity".

I mean, a few generations ago, the performance leaps were incredible. I may be wrong and I'm going off rudimentary impressions, but certainly shrinking down to < 22 nm means less performance gain and more attention to power consumption. And that's what it should be. But because of the reduced performance leaps, there simply isn't as much incentive for people, both casual and enthusiast, to upgrade.

 

[2is]

What about price or power draw or a combination of them both? What does x86 offer to the consumer now apart from the legacy, which is rapidly diminishing?

Most business as well as residential computers (desktops and laptops alike) don't run on iOS or Android operating systems with ARM processors. There's still a huge market for x86. A shrinking one, but a huge one nonetheless, and this will be the case for sometime to come.

 

[Maximilian]

Windows 8 being a total balls up across all platforms dosent help things.

 

[shady28]

We all know performance jumps slowed, but few realize how much.

The link below uses SPEC benchmarks to graph CPU performance increases (single threaded) over the last 17 years.

The late 90s we were seeing +50-60% jumps in performance per year.

Starting around 2001 / 2002, you can see the curve become less steep, and since 2005 the avg has been about 21%.

I want to say multi core / multi threading nullifies a lot of this, but in the real world not many applications make extensive use of multi threading. Even when they do, rarely is the load evenly balanced. Meaning, you don't see the full benefit of multi core CPUs. In fact, the single biggest thing that was cool about Athlon x2 and c2d that I recall hearing from users was that their virus scan wasn't as intrusive.

Plus, we've had multi core since 2005/2006 in consumer desktops.

Another way of looking at this :

2005-2008 was like 1995-1996
2008-2011 was like 1996-1997

Until something re-revolutionizes this pattern, like quantum computers or something equally revolutionary, the PC market will be quite boring (and IMO, it will slow further until then).

I also believe the tablet and phone SoC markets are now mature.

Look for boredom there too in late 2013 and beyond (meaning, sales may continue to decline).

The die shrink game is one of diminishing returns once you go below 45nm, and pretty much everyone is well below that now.

http://preshing.com/20120208/a-look-back-at-single-threaded-cpu-performance

 

[dma0991]

I like Windows 8 a lot. The main problem it has is that it doesn't offer much for desktops. I feel for tablets/convertibles it is much, much better.

The core elements of W8 is still a major improvement over W7, like faster boot up times. The major letdown for W8 is the poor implementation of the Modern UI which is a mixed bag of reaction of good or bad. I find it is acceptable but needs a lot more refinement.

The focus should be desktop first, Modern UI second. Would've preferred to see the Modern UI overlay over the desktop with a semi transparent background while leaving enough space for the taskbar with a Start menu button to never lose focus. A Charms bar that disappears is a PITA as you would have to invoke it every time.

Overall, W8 feels as though you need more steps to achieve the same task. Till Microsoft corrects that, W8 will just be my virtual machine.

 

[grimpr]

We all know performance jumps slowed, but few realize how much.

The link below uses SPEC benchmarks to graph CPU performance increases (single threaded) over the last 17 years.

The late 90s we were seeing +50-60% jumps in performance per year.

Starting around 2001 / 2002, you can see the curve become less steep, and since 2005 the avg has been about 21%.

I want to say multi core / multi threading nullifies a lot of this, but in the real world not many applications make extensive use of multi threading. Even when they do, rarely is the load evenly balanced. Meaning, you don't see the full benefit of multi core CPUs. In fact, the single biggest thing that was cool about Athlon x2 and c2d that I recall hearing from users was that their virus scan wasn't as intrusive.

Plus, we've had multi core since 2005/2006 in consumer desktops.

Another way of looking at this :

2005-2008 was like 1995-1996
2008-2011 was like 1996-1997

Until something re-revolutionizes this pattern, like quantum computers or something equally revolutionary, the PC market will be quite boring (and IMO, it will slow further until then).

I also believe the tablet and phone SoC markets are now mature.

Look for boredom there too in late 2013 and beyond (meaning, sales may continue to decline).

The die shrink game is one of diminishing returns once you go below 45nm, and pretty much everyone is well below that now.

http://preshing.com/20120208/a-look-back-at-single-threaded-cpu-performance

Great information, thanks.

 

[2is]

We all know performance jumps slowed, but few realize how much.

The link below uses SPEC benchmarks to graph CPU performance increases (single threaded) over the last 17 years.

The late 90s we were seeing +50-60% jumps in performance per year.

Starting around 2001 / 2002, you can see the curve become less steep, and since 2005 the avg has been about 21%.

I want to say multi core / multi threading nullifies a lot of this, but in the real world not many applications make extensive use of multi threading. Even when they do, rarely is the load evenly balanced. Meaning, you don't see the full benefit of multi core CPUs. In fact, the single biggest thing that was cool about Athlon x2 and c2d that I recall hearing from users was that their virus scan wasn't as intrusive.

Plus, we've had multi core since 2005/2006 in consumer desktops.

Another way of looking at this :

2005-2008 was like 1995-1996
2008-2011 was like 1996-1997

Until something re-revolutionizes this pattern, like quantum computers or something equally revolutionary, the PC market will be quite boring (and IMO, it will slow further until then).

I also believe the tablet and phone SoC markets are now mature.

Look for boredom there too in late 2013 and beyond (meaning, sales may continue to decline).

The die shrink game is one of diminishing returns once you go below 45nm, and pretty much everyone is well below that now.

http://preshing.com/20120208/a-look-back-at-single-threaded-cpu-performance

I think the slow advancement is directly related to necessity. There are no killer apps out there that require a lot of power. I'm sure if there were, we would have seen far greater performance jumps from generation to generation.

 

[cbn]

From the link in the post #32:

52% to 64% gain per year vs. 21% gain per year....that is a pretty massive difference when compounded over several years.

A few questions:

1. How much of this slowdown is due to interconnect delay?

2. Would CPU architects have more reasons to speed up single thread if I/O became faster? (eg, I am thinking of the all the converging memory hierarchy stuff like phase change and memristor replacing DRAM and NAND,etc)

 

[Munky]

Heard it all before... desktops are dead, tablets are the future, blah blah blah.

Funny how people keep spending money on various tablet accessories, apps and services that try to replicate what the PC could already do better and cheaper.

 

[IntelUser2000]

52% to 64% gain per year vs. 21% gain per year....that is a pretty massive difference when compounded over several years.

This is nearly all due to the power wall. I'm pretty sure if 1300W was the limit rather than 130W, we'd have seen similar scaling for well... few more years. But even that was bound to end soon. Single thread performance and die size/power use has a exponential relationship. Doubling single thread performance requires quadruple the die size and power use, if you can do it. So they opted for multiple cores instead. The timeline where you see a steep drop is when first multi core chips became widespread.

Also, now there's a minimum voltage floor that's hard to lower. Before you lowered power by lowering voltage, somewhat offsetting the power gain. Now you have almost nothing to fall back to in terms of voltage.

So that's why manufacturers decided to change the other aspects, like user experience, and form factors.

 

[Exophase]

But like I alluded to, it's not just about performance. How much do you think the "intel tax" is worth in the current market? Nothing basically. Outside of the tiny enthusiast market, who actually buys a tablet that isn't an iPad or Nexus?

Intel's prices are too high, even if the technology is superior. ARM told them long ago that this was the case. Their business model needs changing in order to compete.

AMD are just hoping to make survival cash on what they can get. x86 is utterly irrelevant in this space.

I thought we were talking about the viability of the x86, not Intel's business practices. You send the wrong message if you blame the instruction set.

This is nearly all due to the power wall.

Exactly. What some don't realize is that phones have already hit the same power wall, and while tablets may have a bit of wiggle room left they don't have a lot and will hit their wall soon. Once the wall is hit advancement slows down.

 

[frozentundra123456]

The core elements of W8 is still a major improvement over W7, like faster boot up times. The major letdown for W8 is the poor implementation of the Modern UI which is a mixed bag of reaction of good or bad. I find it is acceptable but needs a lot more refinement.

The focus should be desktop first, Modern UI second. Would've preferred to see the Modern UI overlay over the desktop with a semi transparent background while leaving enough space for the taskbar with a Start menu button to never lose focus. A Charms bar that disappears is a PITA as you would have to invoke it every time.

Overall, W8 feels as though you need more steps to achieve the same task. Till Microsoft corrects that, W8 will just be my virtual machine.

This more steps to do the same task seems to be a thing with MS these days. Take the ribbon in Excel. To do page setup for printing takes 3 clicks now instead of 2 previously. Not a big deal, but it is less intuitive as well. You have to click file, then print, then page setup. Logically wouldnt you set up the page *before* you print it?

 

[frozentundra123456]

This is nearly all due to the power wall. I'm pretty sure if 1300W was the limit rather than 130W, we'd have seen similar scaling for well... few more years. But even that was bound to end soon. Single thread performance and die size/power use has a exponential relationship. Doubling single thread performance requires quadruple the die size and power use, if you can do it. So they opted for multiple cores instead. The timeline where you see a steep drop is when first multi core chips became widespread.

Also, now there's a minimum voltage floor that's hard to lower. Before you lowered power by lowering voltage, somewhat offsetting the power gain. Now you have almost nothing to fall back to in terms of voltage.

So that's why manufacturers decided to change the other aspects, like user experience, and form factors.

Unfortunately, the slow down in improvements is a basic characteristic of almost every technology. The easy gains come first, while after a certain point additional gains become very difficult to obtain.

Another example is cars. Look at how much performance improved from the from the 1930 to the 1970s. The improvements for the next forty years, especially the last 10 or 15 have been much slower. Right now tablets and smart phones are in the early, easy to improve phase. We will see how long that can continue.

 

[pablo87]

Took a quick look at Intel's #'s:

- Revenue surprisingly stable YoY. Intel has pricing power esp. in the cloud.
- a good portion of the lower margins is depreciation increasing.
- depreciation is still running well below capex though YoY capex was down.
- Property plant and equipment is up $10.7B in 3 years, after hardly changing the previous 8 years.
- R&D expenses are up, but tax expense are down YoY from 27% of income to 23% on a global basis.
- from my perspective cash flow (income+depreciation less capex less dividends less stock buyback) was negative 514MM. When I look at cash flow before dividends and stock buybacks, its $1.6B. Based on close today, market cap, less cash of $5B net of debt, is 17x cash flow. Rich.

I understand they are investing in the future, and the top line to me is really good but the cash flow and the cash levels are not very good relative to market cap.

 

[2is]

Don't think cars is a good example. I've owned 4 cars '89 '98 '06 '13 (still own the last two) and the improvements between the first three is quite drastic in every way imaginable. The last one I intentionally purchased a base model econo car as a company car.

I think a better analogy is... Working out. Improvements come easily when you first start going to the gym but get exponentially harder as you improve. You have to work out harder, longer, more often, all while getting less results... It's why a lot of people either quit, or start taking... "vitamins"

 

[beginner99]

What does x86 offer to the consumer now apart from the legacy, which is rapidly diminishing?

Not having to wait several seconds for any half-complex website to load?

 

[meloz]

I think this is good news for consumers. With AMD not providing any real challenge, a drop in profits will be a good reminder for Intel to not jack up the prices too much. They need the sales volume, and for volumes they need to keep prices of their products within a sensible range.

 

[MisterMac]

ExoPhase - wait for the full benchmarks of Exynos 5.

Then we'll see the Year on Year benefits


As others have said - there's a wall in all of the SemiConductor industry.
Unified shaders or multicore\threaded alike-inventions are needed before we see major jumps.

Secondly - once Html5 arrives most people won't feel much difference from ARM\LP x86.
The massive overheads and double computations from flash in browsers these days is fukkin insane.

 

[itsmydamnation]

I think a better analogy is... Working out. Improvements come easily when you first start going to the gym but get exponentially harder as you improve. You have to work out harder, longer, more often, all while getting less results... It's why a lot of people either quit, or start taking... "vitamins"

Dont mean to be a pain in the arse, but if that's the case your doing it wrong , prilepans chart, HIIT, knowing how to train correctly, etc. The problem with the analogy is that people dont know there bodies very well. The engineers working on these scaling problems understand them extremely well yet still struggle.

 

[2is]

Dont mean to be a pain in the arse, but if that's the case your doing it wrong

That could certainly be the case.

 

[cytg111]

.. Dont mean to be a pain in the arse, but if that's the case your doing it wrong

- Arh.. thats a stretch, I'd say, on a whole, that analogy is more right than wrong. Certainly, if you're doing very stupid things in the gym, less can be more, but the elite of many sports will train multiple times a day just to stay at their current level. When starting out in the gym, going once a week for half an hour will still yield you results.

 

[krumme]

Took a quick look at Intel's #'s:

- Revenue surprisingly stable YoY. Intel has pricing power esp. in the cloud.
- a good portion of the lower margins is depreciation increasing.
- depreciation is still running well below capex though YoY capex was down.
- Property plant and equipment is up $10.7B in 3 years, after hardly changing the previous 8 years.
- R&D expenses are up, but tax expense are down YoY from 27% of income to 23% on a global basis.
- from my perspective cash flow (income+depreciation less capex less dividends less stock buyback) was negative 514MM. When I look at cash flow before dividends and stock buybacks, its $1.6B. Based on close today, market cap, less cash of $5B net of debt, is 17x cash flow. Rich.

I understand they are investing in the future, and the top line to me is really good but the cash flow and the cash levels are not very good relative to market cap.

Thanx. It will take forever for me to understand the numbers, so good to get at resume

I dont see anything threatening Intels cash cows and stars (cloud)?

They will continue to generate solid profit for many years on near monopoly markets, - unless someone by pure chance found the magic 100%+ button on Kaveri

So at least for the medium outlook this is a very, very healthy business.

What i dont understand is why the shareprice have dropped so much over that last years. Isnt the server segment enough on the more long perspective?

I mean how much profit is in the markets where Atom and low TDP IB/Haswell is headning, fighting kabini/temash and A7/little-big & NV. Its peanuts. I dont understand what they want there, why not expand and develop - perhaps even beeing innovative- the markets where they have monopoly power because they have superior competences and technology here?