Intel ends Tick-Tock cycle and delays 10nm production to 2017
Click here to post a comment for Intel ends Tick-Tock cycle and delays 10nm production to 2017 on our message forum
Denial
Bummer. Intel needs to get EUV up and going. Pretty crazy that a company with that much money and arguably the best engineers can't figure this out.
Matt26LFC
Well like this wasn't obvious. Which is basically why I'm going to get on skylake now, nothing is coming for at least the next 2 years thats going to be interesting. It won't be until 10nm Canonlake arrives, and even then Canonlake is mostly just a die shrink of Skylake. Could be 2018 and 10nm with a whole new Architecture before we see something amazing for CPUs in the Desktop space.
At least Z170 chipset will have all the latest features etc Only thing I could think that might be missing from Z170 Platform is PCIe 4.0, perhaps that'll be a feature that Intel use to sell Kaby Lake when it arrives next year, PCIe 4.0 for the mainstream, but limit it to 16 lanes so's not to undercut Skylake-E sales.
Denial
From Anandtech:
waltc3
Yes, "tick-tock" is like "Moore's Law," one of those generally understood truths that's false...;)
tsunami231
sound like skylake will be what i going after then
Megabiv
Same here
Pete J
....that proves it's not a trivial problem.
You get problems with quantum tunneling and other weird physics that only a few people in the world understand. So what happens when they get to 10 nm? 8 nm? 6 nm? Silicon chip atoms are ~0.2nm. We are running into physical limits.
abula
Intel liked so much what they gain with Devils Cannon that they want to repeat this trend with Kabylake, and milk us even more between real ticks and tocks.
Denial
I don't really get the ... I know it's not a trivial problem. That's why I think it's crazy.
Deasnutz
It was an absolute bitch for them to get on 14nm, let alone 10nm. 14nm is fine for a 3rd generation, as it will still have room for improvement.
They obviously need time for engineering, I'm sure they are considering many directions.
Loobyluggs
It's probably more like you can't reduce manufacturing costs and increase yields ATMIT and therefore doesn't make sense to progress.
umeng2002
IBM just made the first 7nm chip, so I doubt Intel needs more time for R&D.
I think it's just a delay since they're so far ahead of AMD. Why push things out the door when you can stretch technology out and make more money until AMD get a little closer, at least.
-Tj-
Exactly.. CannonLake was first made to be Z170 compatible, but now they decided it will have new chipset, more $$$$ this way. And for backwards compatibility they make some ripoff KabyLake, just so Z170 users won't be left out..
I really hope AMD Zen will kick Intels ass so they will stop with this ripoff, either release proper tock in same socket like in the past or don't bother with something half-made :P
Denial
Uh, Intel may be competing against AMD at the desktop level, but their focus has been on mobile for a few years now. They aren't competitive there yet and they definitely need die shrinks to become competitive. They definitely aren't holding back.
-Tj-
Well I was only referring to desktop die shrinks, it looks like they're milking it to the max.. I agree on mobile part.
Chillin
It was barely a "chip", it had a few transistors on it was all, nowhere nearly as complicated as these CPU monsters. It was also not done in a way that's suitable for manufacturing, it was a proof of process demonstrator.
Pete J
Ah, gotcha. I just get a little tired of people who think that throwing more money and people at a problem will automatically solve it. Not to mention those that think a company have some sort of prerogative to not make money.
icedman
Die shrinks arent even exciting anymore at this point it seems all they do is increase the cost and add 5% performance increase.
Clouseau
Has software really been keeping up with the times lately? Why go down the same road AMD does and release an architecture that is so far ahead of software development that would render it irrelevant in the current market? Hardware needs to stay ahead of software development to keep pushing the development; cannot get to far ahead of the curve at the same time. Like has been said, AMDs market share has very little impact on Intel's decisions. Intel does what is in the best interest of Intel. They are aligned very well to the current state of the market.
k3vst3r
Made in a lab versus mass production in factory, even IBM said they would have to figure out if it's possible to mass produce using their new technique.
