Intel's 14nm Broadwell Delayed Because of Low Yield 96
judgecorp writes "Intel has put back the delivery of its 14nm Broadwell desktop chip by a quarter because of a manufacturing issue that leaves it with too high a density of defects. The problem has been fixed, says CEO Brian Krzanich, who says, 'This happens sometimes in development phases.'"
The good news is that it is just a defect density issue. A first round of tweaks failed to increase yield, but Intel seems to think a few more improvements to the 14nm process will result in acceptable yield.
14 nanometers should be enough for anyone. (Score:5, Funny)
14 nanometers should be enough for anyone.
Re: (Score:2)
Re: (Score:2)
There are in the number .3048, which is in the standard SI unit.
Re: (Score:3)
Nah, we're not quite at that point yet. I've seen estimates that we could see 1nm processes by 2030, but many people say anything below 5nm (expected circa 2020) isn't feasible. Either way, we're at about the manufacturing limit of the Newton and Thomson/Bohr/Rutherford universe. Atoms are between 0.3 to 3 Angstroms in size. That's 0.03nm to 0.3nm. If we want to go smaller than that, we have to construct our devices out of something other than atoms, and it's assuming that subatomic/quantum forces don'
Re: (Score:2)
or stop using electrons
Re: (Score:3)
Two minor points.
Electrons are Easy to use and last a long time.
Subatomic particles on the other hand are much hard to deal with. Also not many of them are actually smaller.
Re: (Score:2)
Photos are easy to use (kinda) and last a long time. Oh and they are definitely smaller.
Re: (Score:2)
Re: (Score:3)
If we want to go smaller than that, we have to construct our devices out of something other than atoms
You could shrink the atoms using muons, but overclockers will suffer a nasty surprise.
Re: (Score:2)
The thing to remember is that roughly about 1nm is the end of the line for the current process for chip making. It doesn't mark the end of circuits, it just means we need a different method. That could be with Silicene, nano tubes, or heck even quantum computers. This isn't the end of the progress of processors, it's just the end of Moore's Law. There might be even a 0.5nm or 0.1nm era, but there will be some serious diminishing returns for that (unless someone is really clever)
Re: (Score:3)
14 nanometers should be enough for anyone.
It's not a problem of 14 being enough, it's a problem of 14 being too much.
Comment removed (Score:5, Informative)
Re: (Score:3)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2, Informative)
Re: (Score:2)
For others like me who had to look this up, link [wikiquote.org].
You had to look this up? Now I feel old. Also, get off my lawn!
Re: (Score:3)
Re: (Score:2)
I guess it is less for cost savings reasons (during production) and more about power savings (during operation) and probably also for performance reasons.
Caveat: I'm not a hardware specialist, not to mention a chip specialist, but I assume (from physics perspective) that shorter distances will reduce power consumption for same amount of transistors. Also, if the total die-size required for e.g. 1b transistors goes down from x to x-y, you could keep the die size but increase the transistor numbers.
Smaller ch
Re:640nm ought to be enough for anyone.... (Score:4, Insightful)
The main reason die shrinks happen is usually at the behest of the manufacturer, and rarely at the behest of the customer (unless we are talking die sales, which we usually don't for Intel, and which is a whole different ball game). There are always downwards pricing pressure on the manufacturers (a tad less on Intel, I'd think, given their elite position amongst fabs) and in order to preserve their margins, they work in these shrink transitions w/ their customers. But make no mistake - for customers, those shrinks imply requalificaiton and a whole new product development cycle before they can go to market w/ those. They'd rather get their price cuts on the same die, except that the manufacturers typically won't give them that.
In the past, the reason to go for shrinks was improved clock speed, and more recently, it's power consumption. But power consumption alone doesn't drive such market trends, particularly given the expenses incurred - what really drives it is cost. But again, as I said, we're really past the point where shrinks would result in any significant cost savings.
Re: (Score:2)
Thanks, that was quite helpful in understanding the manufacturer's perspective as well.
Re: (Score:2)
Upto 50nm, it was fine, but is now in the region of diminishing returns. The cost savings that were always synonymous w/ shrinks are no longer there, since the process costs easily outweigh the cost savings per die, even assuming a 100% yield.
I wasn't aware of a 50nm node or shrink, but 65nm was the last "cheap" node (which corresponded to a 55nm shrink). The next popular node was 45nm didn't really take off before the shrink to 40nm. TSMC (one of the major foundaries) actually blew-off the 32nm and 22nm nodes completely and only productized the 28nm shrink and 20nm shrink.
Usually, the "shrink" is purely a cost/die = (cost/wafer)/(yield*die/wafer) issue. Since in a shrink, the cost/wafer is mostly constant but the die/wafer goes up, the cost/
Re: (Score:2)
Why can't we just improve for the sake of improvement?
Don't worry, Intel doesn't invest in the next generation of chip technology and say to themselves, "this is cool, but we're never make money on it."
Re: (Score:3)
Because a wafer of pure silicon has a high, fixed cost. The smaller you can make a chip made from that wafer, the more chips you can make and the lower the cost for each chip. Intel's recent 22nm chips have recently been around 200mm^2. If they were made on 640nm, they would be about 200*640/22=5800mm^2, or a square of about 7.6cm (3 inches) on a side.
Also, smaller wires use less electricity and generate less heat. You wouldn't want 640nm Haswell in a laptop.
Re: (Score:2)
No but I would love someone to build one just to compare the performance.
The biggest savings are in the capacitors (Score:2)
If you reduce all 3 dimensions of a wire by the same proportion, you'll get a wire with highter resistance, not lower. The energy savings from reducing the feature sizes come from reducing transistors and inter-wire capacitance. With a smaller capacitance, you need less charge to turn the transistors on or off, using less power and letting them switch faster.
That is, untill you let too much current leak through them. Make them too small, and you'll consume more power again.
Re: (Score:2)
The comment is only half the storry. The other half is related to yield. Take for example a 12 inch wafer and make chips 3 inches square. You get at most 4 chips. Smaller chips = more die.
Now the defect density part of the formula. Say you have 100 defects randomly spread on the wafer. Your yield of 3 inch chips would be zero for most of your manufacturing. Dice the wafer into 500 chips by having smaller die size and now your yield is 4 of 5 die or better is your yield as sometimes randomly scattered
Geat! Time to cover my short position in Intel... (Score:2)
Short term setback for Intel. They will get yield up eventually. I just hope it's before they run out of cash to run operations...
Re: (Score:1)
I just hope it's before they run out of cash to run operations...
Lolwut? Yeah, umm, that's not even remotely a concern.
Re: (Score:2)
Sorry, tongue was firmly in cheek on that one..
Re: (Score:1)
On the other hand, the guys at Altera have bet the bank on Intel, so they're likely praying that Xilinx's 16nm TSMC process gets delayed.
While Intel has utter dominance on their market, Altera is in catch-up mode...
Re: (Score:2)
TSMC's metal 1 pitch is 64nm in 20nm, and Intel's 22nm is 90nm.
14/16 is indeed expected to have ~64nm pitch, so it's not better than TSMC's 20, but it's a great leap for Intel.
"disabled": Not quite. When I'm running 491MHz internal, you can't just disable arbitrary logic on me. The slower parts may get away with disabling columns, but you can't change my timing without breaking my design.
Also, all the hard IP is not redundant, and there's more and more of it.
No real long-term impact (Score:2)
Since, in practice, they'd want to get rid of old stock before selling their shiny new product, this isn't really that much of a problem.
It's not like AMD is going to magically beat Haswell before Broadwell is released. It would be nice if they did, though...
Next steps: 7nm, 1nm (1000pm), 800pm, ... (Score:2)
... where does it end? I had to actually check what the atomic size of Silicon is (111pm), so there are only a few years left (maybe 10-20) to reach the atomic level. Then what? I'm really curios as I'm quite impressed how this development came - actually how quick...
Re:Next steps: 7nm, 1nm (1000pm), 800pm, ... (Score:5, Interesting)
Potentially it can keep going until the size of a transistor is just a few electrons across but as we get closer to that point quantum teleportation becomes more of an issue. This is cool video that explain some basic stuff about transistors and the end of Moore's Law.
https://www.youtube.com/watch?v=rtI5wRyHpTg [youtube.com]
Re: (Score:1)
Potentially it can keep going until the size of a transistor is just a few electrons across but as we get closer to that point quantum teleportation becomes more of an issue.
More often referred to as electron tunneling or quantum tunneling.
Re: (Score:1)
a few electrons across?
An electron is a point particle, it has no diameter. I think you mean a few *atoms* across. You cannot make a transistor out of a single atom.
Quantum teleportation?
I think you mean quantum tunneling.
sigh...what has happened to the Slashdot of Old, where the comments were insightful and informative. and written by people who actually know the subject they are talking about....
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:3, Funny)
Maybe we could build computers out of Planck planks. They're really small.
Re: (Score:2)
Re: (Score:2)
Too bad that all the unicorn horns have been hunted and sold to China for superstitious erectile dysfunction remedies.
Re: (Score:2)
I can't see how photonics could be made fast. It can lead to a huge bandwidth, but not much improvement on single-core speed from electronics.
That is, unless your photons have a very big wavelenght, and are poralized, because then you enter spintronics domain, and those have either a huge potential, or some very big problem that nobody discovered yet. The breakthroughs that will make spintronics real are mostly at the "how to assembly superconductors with extreme precision" area, and I think we can apply a
Re: (Score:2)
If you made a law preventing any transistors below 14nm, the architectural work would continue making things faster.
Architectural changes have contributed more to speedup than transistor size over time. They are not independent, since smaller transistors allows more integration and co-location, but from where I sit, there's plenty to be done in computer architecture to make them faster and plenty to do in software to stop blowing away so much performance on fripperies, bad drivers, bad memory management and
Wouldn't put too much in this (Score:2)
They released Haswell in June, they've barely had time to sell that so Q4 2013 to Q1 2014 is still ahead of their yearly tick-tock. They're not announcing any delay to Airmont which is their mobile 14nm chip and we all know one quarter to or from won't change much in the desktop/server market. In related news AMD posted their Q3 earnings today and their CPU sales are still down, their gross margin is down but on the bright side the console sales are finally coming in so overall they're making a profit this
Re: (Score:2)
So are Intel's. This is not a good time to sell x86 CPUs. AMD has a chance to reverse their trends, since they are just a small player, but they'll have to steal that market from Intel.
Not the real issue (Score:4, Interesting)
Intel has produced two new generations of processor that were WORSE than Sandybridge. Higher power use (under load) and far less over-clockability. The newer part were ONLY better (in desktop systems) if you intended to use their new instructions (vanishingly unlikely) or the integrated graphics (which would be pointless- people buy expensive Intel CPUs to partner them with expensive GPUs from AMD or Nvidia).
Intel, of course, were in the same position with the waves of Core2 parts, each of which essentially overlapped each other in performance generation on generation (although power consumption was much improved over the first generation of Core2 i7 parts).
Intel currently doesn't know exactly where to go in the near future, and is attempting to hedge its bets by trying various things. It is currently undercutting its own HYPER-expensive ULV mobile high-end parts with the new 4-core 'atom' Bay-trail chips that seek to go head-to-head with ARM. Because current high-end ARM is so good, Intel is forced to sell a very dangerously good chip (dangerous to Intel's profits, that is) into low and mid-end tablets, running Android or Windows8.1
However, even Intel's first decent 'Atom' part ever (after 5+ attempts) is beaten by Nvidia's somewhat lame Tegra 4, and Qualcomm's Snapdragon 800. It is exterminated by Apple's new ARM chip, soon to be seen in Apple's new iPad refresh.
Intel's 22nm process, and use of FinFETs, has been a total disaster so far. A process advantage, and custom designed chips, doesn't allow Intel to beat ARM parts coming from commodity foundries at TSMC and Samsung in 28nm. Sure, Intel can make its own chips smaller than those on the previous process, and theoretically get more parts per wafer, but the per wafer costs rocket, the yields drop (initially), and insanely expensive new plants have to be built to service the new process.
What does Intel get from spending all this new money on R+D? At this moment in computer history, almost nothing. The x86 is dying, and everyone BUT Intel builds ARM solutions. Every major player has a GPU (graphics) solution as good as Intel, and Intel isn't within a million miles of matching the AAA-gaming GPU designs from AMD and Nvidia (despite the fact that Intel has spent more money than every graphics company combined, across their combined periods of existence, to create its own GPU solutions).
Intel simply has no current use for its expensive 14nm process. It has built the factories, so it is engaged in a waiting game- waiting for mobile parts to roll off the 14nm production lines that have clear market advantages over its current mobile chips. It just isn't worth Intel's time launching another round of non-improved parts. The market has changed forever, and on-one wants to buy "this season's Intel" for the brand loyalty reasons previously apparent.
Intel fanboys want 6-core and 8-core parts, but Intel is extremely loathe to risk introducing better value into the desktop market. If Intel properly sold 6-core solutions, they would have to sell 6-core i5 parts, and these would beat-up their EXTREMELY profitable 4-core i7 parts. Intel is too in love with the status quo.
If Intel's bay-trail 4-core parts prove good enough for tablets and non-gaming laptops, and they will do having greater performance than the more than adequate mobile 2-core core2 parts used in the first decent cheap laptops years ago, where does most of Intel's mobile biz go from here? Bay-trail parts (unlike those years old mobile 2-core core1/core2 laptop chips) also do all the video decoding in hardware, allowing flawless playback of all current video content (and bay-trail is strong enough to do CPU enhanced decode of 4K video recorded in h264).
Bay-trail is the part Intel moved Heaven and Earth NOT to produce. Bay-trail is the final step on the race-to-the-bottom for x86 based computers that most non-AAA gamers will need. If the only real money Intel makes ends up from chips lie Bay-trail, Intel is done.
Think about this. In a few weeks, you
Re: (Score:2)
In a few weeks, you will be able to buy quite decent Android tablets for $150 using 4-core bay-trail. A little hacking, and you've got yourself a $150 dollar Windows8.1 tablet. A $150 tablet running PROPER unrestricted Windows.
Yeah, 'cause Windows is, you know, free and stuff.
At retail, you'd be paying about $100 for Windows alone.
Re: (Score:1)
A little hacking, and you've got yourself a $150 dollar Windows8.1 tablet.
Do you really think we won't be pirating the shit outta Windows when we do that?
Re:Not the real issue (Score:4, Informative)
Re:AMD (Score:4, Funny)
Might explain why I went with an AMD processor, and might get a video card from them soon.
'I like my chips like my women: hot and slow.'
Re: (Score:2)
At this point in time Intel on the desktop covers more of the first three aspects.
If Intel goes for a more mobile offering for a few generations, AMD will be back in consideration.
Pull an AMD (Score:3)
I have a theory that those T-edition chips Intel made that are just underclocked, hyper-efficient, ultra-low wattage editions of their recent chips are actually just ones that wouldn't run properly at the normal stock clock. I never heard a solid claim that they actually had different voltage regulation circuits or something like that. They just underclocked them and made them have a higher tendency to not click to a full multiplier level as often or for as long.
Re:Pull an AMD (Score:5, Interesting)
Except the cores are probably small compared to the L3 cache, so most failures will be in the cache, not the cores.
Back when I worked in the chip business, we designed them so some components could be disabled if they failed the manufacturing tests, but there were very few that we could actually sell that way. Either the fault would be in the components that couldn't easily be disabled, or there'd be multiple faults in too many places to make it viable.
I have wondered myself whether the low-power CPUs are just the bin that wouldn't work at normal power levels.
Re: (Score:2)
Problem with this logic is that power consumption is a factor. While the chips are underclocked, they are also undervolted by a proportional amount. Undervolting with underclocking was a rare pasttime by overclockers, some done as hobby and others done in a quest for performance/watt crown. The chips binned for the highest clocks and the chips that overclock the best run at higher clocks on the same voltage, or more efficient. The same chips can be undervolted and perform the highest clock at their lowe
Re: (Score:2)
Re: (Score:1)
Actually check out Haswell's die configuration [wccftech.com], the integrated graphics takes up about 2 times more area than the L3 cache. Also, look at how dense the transistors are in the GPU area, looks as dense or maybe even more dense than the cache. It wouldn't surprise me if graphics are a source of manufacturing problems in addition to L3 at this point.
Good news? (Score:2)
And what kind of problem you can have on a fab that is not a "defect density issue"?
In a related question, can I declare Moore's law dead already, or is there some current fab upgrade that isn't delayed by at least 18 months?
Re: (Score:1)
'And what kind of problem you can have on a fab that is not a "defect density issue"?'
Systematic flaws, like all horizontal wires are printing 10% narrower than intended or the effective dielectric constant at a particular layer trails off evenly from the center of a wafer to the edges.
Getting in touch with my inner Grammar Nazi (Score:2)
... because of a manufacturing issue that leaves it with too high a density of defects.
Sorry, after a long time, I have to put myGrammar Nazi hat on for (I think) the first time. It's not just you, this is just the example that tipped me over the edge - much like the "leaves it with too high [of] a" phrasing leaves the reader tipping off into.... what?
This type of construction has become endemic in conversation in the last few years, and I'm sorry, but it's cumbersome, ungainly, unsightly, and painful to hear or see. Perhaps, just perhaps, if I say something, this bad practice will lose so
Re: (Score:2)
It's too short a season to grapple with so harsh a critique of this minor a transgression.
Sorry. I guess that was a bit like scratching a chalkboard, but I personally rather like this particular grammatical construct. It's efficient and it front loads the subjective point the author is trying to make, making comprehension easier. Compare: "The season is too short to grapple with a critique that's so harsh of a transgression that's this minor."
Re: (Score:2)
It's too short a season to grapple with so harsh a critique of this minor a transgression. :)
Wow, you managed to get three of them into a single sentence!
How about "The season is too short to spend it grappling with such a harsh critique of a transgression this minor".
Our brains read predicively, constructing the most probable usage as we go. (There was an article on slashdot about this recently). In this case, I would say that "of this minor a transgression" is first interpreted by our brains first as re