massive intel lay off

[sarepairman2]

https://hardware.slashdot.org/story/16/04/19/2210226/intel-confirms-major-layoff-12000-worldwide-11-percent-of-workforce

what do you think?

[dannyf]

It doesn't surprise me the least bit. For quite a few years, I have been telling friends what I saw in China: the IC design houses, the fab facilities, the massive funding of anything technology there, etc.

If I were running a chip maker (Intel, Qualcomm, TI, linear, etc.), I would be ***extremely*** concerned about my future 10 - 20 years down the road.

[sarepairman2]

eh, think about all the govt/american industrial usage.

no one is ever gonna trust anything built in china

[EEVblog]

Wow, that's big.

[Stonent]

Contributing factor #1... Microsoft.

The processor requirement for Windows Vista was 1GHz. The processor requirement for Windows 7 was 1GHz, 8, 8.1 and 10 as well.

Microsoft has not upped the specs on their processor requirements in 10 years.
Windows 10 is a free upgrade for 3 operating systems. So now you have the most modern OS running on your computer and it didn't cost you a dime.

If you wipe a 6 year old laptop and install Windows 10 on it cleanly, it runs decently.

Factor #2. Intel overproduces.

There is massive performance overlaps in their processors. A 5 year old i5-2520M almost identical performance to a I5-4310M.
Some I5s are slower than previous generation I3 processors.
Some I5s are faster than some I7 processors.
Some Pentiums are faster than I3 processors.

It seems like they keep increasing the top end of their processors without knocking out mid and low range when they should.

They refuse to have a performance spot that isn't covered between top and bottom ranges.  They need to set a performance floor every year and drop processors under that performance, or eliminate the massive mid range.

Factor #3 ARM

I shouldn't have to explain this one.. It's pretty self explanatory to anyone who has a mobile phone.

Factor #3.5 Atom

It never has really become what it should have been, Intel's mobile phone processor.

[ECEdesign]

Essentially they are running into the end of the physics that allowed simple shrinking of the transistors.  There are exciting new technologies like III-V Nitride semiconductors but they are slower at this point.  There needs to be more research in this area which I believe is mostly going on at the university level and not much in industry.  I would like to specialize in semiconductors, the quantum mechanics is really interesting but with such an unpredictable job market in semiconductors maybe I should go into something more in demand like signal processing or something like that.

[NiHaoMike]

4 years ago, I built a new PC to use as my main one, right at the 5 year "major upgrade" mark I have traditionally followed. Now, I don't see any good reason to replace the CPU/motherboard in one year. I did, however, upgrade the GPU and monitor 1.5 years ago. Not sure when I would upgrade the GPU again but I plan on keeping the 4K monitor for 10 years or so.

[Stonent]

So is it time to seriously look at GaAs and InGaAs again?

You had me at "transistor switching speeds up to 250GHz"

[EEVblog]

IBM is still doing as big or bigger but they have tuned the process and can do it out of the public view. They've now passed 4 years of declining revenues.
I hope the OP doesn't mind the thread drifting to tech layoffs in general.
What I don't understand is that if IBM with WATSON is trying to capture new market and revenue why they don't ask WATSON what they should do to turn their failing fortunes around. It can't come up with a worse idea than their current strategy.
Personally I think IBM has already had its KODAK moment. Given KODAK's long history and IBM share that much, it faded away relatively quickly as did Polaroid.

IBM is all but dead as we knew it, I don't think anything can really save it.

[botcrusher]

x86 has lasted a long time. Are we starting to see it reach the end? x86 destroyed several other major technological alternatives to it. After all, the AIM alliance is but a forgotten name.

The only other core i can think of having lasted so long is ARM.

Then again, in the quest to get smaller, why don't we increase the cpu die size. It's not like motherboards couldn't be made more compact to accomodate for an increase in die size. Above that, what ever happened to 3D gates, and gate stacking?

[jeremy]

Then again, in the quest to get smaller, why don't we increase the cpu die size. It's not like motherboards couldn't be made more compact to accomodate for an increase in die size. Above that, what ever happened to 3D gates, and gate stacking?

There are tons of problems with this: propagation delay, the fact that cost is very highly a function of die size due to the spatial distribution of faults, parallel programming is hard, etc

[uncle_bob]

x86 has lasted a long time. Are we starting to see it reach the end? x86 destroyed several other major technological alternatives to it. After all, the AIM alliance is but a forgotten name.

The only other core i can think of having lasted so long is ARM.

Then again, in the quest to get smaller, why don't we increase the cpu die size. It's not like motherboards couldn't be made more compact to accomodate for an increase in die size. Above that, what ever happened to 3D gates, and gate stacking?

Hi

A lot depends on how many people you believe will buy $4,000 to $16,000 CPU's. If the answer is dimensioned in the "many thousands per day" you *might* have an interesting product. If the volume is in the dozens per day range, you won't even pay the mask costs for the first run. Certainly there is no money left over to re-hire a bunch of people to design, test, manufacture, and market the beast. Consider that it's not just the CPU, you also need all the support chips as well. That $4,000 CPU will go into a computer costing way over $20K by the time you buy it in a store. How many people will buy one of those?  I'd bet not many.

The real issue is that the world figured out a tablet or smart phone will do all the things they need to do. It won't do all the things *I* need to do, but I'm down in that "dozen per day" group.

Bob

[tom66]

Then again, in the quest to get smaller, why don't we increase the cpu die size. It's not like motherboards couldn't be made more compact to accomodate for an increase in die size. Above that, what ever happened to 3D gates, and gate stacking?

Multicore processors are an evolution of this idea. It is easier to make a multicore processor where each processor need only communicate with a single shared bus, clocked at a much lower rate, rather than one larger, more powerful single core CPU.

But most programs don't parallel too well.  Video encoding and decoding are one of the most parallel programming tasks most computers will encounter, whereas something like web browsing and games are typically limited in how many cores can be used simultaneously. So a 64-core CPU wouldn't necessarily be much good.

[CatalinaWOW]

The answer is just what Intel said it is.  The desktop market is dying.  Why?  Because the big markets no longer need any more out of them than they have been able to provide for a decade or more.  And because the various companies that serve that market went to the cloud as a way to convert the market to a subscription model. 

While this geek community, and a few others (physicists, atmosphere modelers and so on) could benefit from bigger and faster processors we constitute a trivially small market that can't pay back the investment to develop those processors.  Word processing, facebook, even payroll and inventory management are doing just fine with current processors.

So, it was a nice ride while it lasted, but until the next mass application takes off it is going to be tough sledding.  Self driving cars maybe.  Personal genetic level health monitors and editors maybe.  More probably something none of us has imagined yet.

[ECEdesign]

Personal genetic level health monitors and editors maybe

Protein mapping and genome stuff takes a huge amount of processor power.  The biophysicists are having fun with CRISPR and getting some cool results but this is more at the university research level.  Joe at home has no need for this kind of computer...

[Tomorokoshi]

I hoped that somehow Kodak would have been able to leverage their large-scale precision coating technology to some kind of solar panel production.

There probably isn't too much Intel can bring to solar that someone else isn't already doing.

At this point I don't need a faster processor, but it would be nice to have one that used less power.

[Brumby]

Joe at home has no need for this kind of computer...

For now.

[Kilrah]

4 years ago, I built a new PC to use as my main one, right at the 5 year "major upgrade" mark I have traditionally followed. Now, I don't see any good reason to replace the CPU/motherboard in one year.

+1. I changed my 2007 PC in 2010 because it was just too slow to run a game, CPU pegged at 100% - and intended to have the new one run 4 years instead of 3 so I went one notch up the CPU category. Well... it still runs perfectly fine and there is no application that can't run because it's too slow.

I only built a new machine in 2014 because I had to travel to another country for a project and didn't want to carry the old one with me, and while I went top of the line it was deceivingly not that much faster. Sure there's a difference, but nothing through the roof, maybe a 2x difference in 4 years - while before it used to be at least 3x in 3 years, which incidentally was my "rule" i.e. don't upgrade unless I get at least 3 times better.
The old PC is still at the old location and I use it perfectly fine when I go back there a few times a year.

So yeah, performance improvements have tapered off drastically. Too bad because I would certainly buy something significantly better, the faster those 42 megapixel RAWs process the better - but it feels like satisfying the 3x rule will probably take another 10 years or so at this rate.

[AndyC_772]

I can't remember the last time I used a PC whose CPU was 'too slow'. Every one I've bought in at least the last five years has been 'fast enough' for everything I want to do with it, and that includes games as well as work.

So, no need to upgrade, and no new CPU sales.

Add to that the well documented reasons not to want Windows 10, and the fact that a new PC will come with it, means I won't be upgrading voluntarily any time soon. I'll keep using my Windows 7 machines a few years longer instead.

Intel: if you want to sell more chips, kick Microsoft up the arse and tell them to take the telemetry crap out of W10. Tell me when you've done it, and I'll consider a new PC.

[Kjelt]

You have to keep on re-inventing your business and transform at the right time in order to survive.
Intel missed out on the huge mobile market, suffering from the collapsing pc market, they should have bought Arm when they could have, now it is too late.
So downsizing and getting the bussiness healthy again is a good move, probably the only option left.
Is it sad? I don't know, if I look at some of these obesitas companies with layers of management and bureaucracy completely obstructing fast acting and supporting its customers, it's pretty disgusting.
Most companies that do recover such a large scale reorganization tend to be better and more profitable companies than they were before.

[coppice]

Intel missed out on the huge mobile market, suffering from the collapsing pc market, they should have bought Arm when they could have, now it is too late.

You might have missed this, but Intel had a big ARM business, an ARM architecture licence, and developed several of their own ARM cores. They sold this business to Marvell. They even developed a cellular platform in the 90s, in cooperation with ADI. ADI still sells the Blackfin, which resulting from this collaboration. Try looking at the Blackfin's architecture and instruction set, and you'll see an ARM that was stretched into a DSP. Intel's problem is not that it didn't see the importance of mobile, or how ARM could be a good solution for mobile. Their problems lie in their execution.

[Kjelt]

You might have missed this

Perhaps through my not native english but you miss my point: Intel should not have bought an ARM license, they should have bought ARM, the entire company 

[Deni]

Joe at home has no need for this kind of computer...

Joe at home most likely do not need computer (as we refer to it) at all - what he needs is a device that allow him to browse the net, send e-mails, watch video etc. Of course, in the past the only device capable of doing this all was traditional PC. Nowdays, you have smart TV's, Android tablets/phones etc. Very small percentage of people actually use "computer" for it's "original" purpose.

I guess Ken Olson was right (1977: There is no reason for any individual to have a computer in his home) in that sense. And that explains PC sales decline.

[coppice]

You might have missed this

Perhaps through my not native english but you miss my point: Intel should not have bought an ARM license, they should have bought ARM, the entire company 

If anyone buys ARM, ARM is dead. The whole appeal of ARM stems from its openness to all manufacturers. What Intel did with the Xscale was the practical approach - innovate around the ARM instruction set, while not blocking everyone else out completely. They just did a very bad job of this.

[Kjelt]

If anyone buys ARM, ARM is dead.

Ok, lets say that is true and this would have happened at the turn of the century, Intel purchased Arm and Arm was dead.
Then what? What brand of microcontrollers would all those billions of mobile devices be running at