Read more.Quote:
Reduced multipatterning when compared to 10nm, helps it stay on track with original plans.
Printable View
Read more.Quote:
Reduced multipatterning when compared to 10nm, helps it stay on track with original plans.
Only a single 7nm fab?
Maybe a secret jump to 5nm but don't want to tip off the competition?
Or sheer uncertainty? Given the competition is already on 7nm, maybe building multiple expensive fabs isn't good sense when the competition has it sewn up and they're well behind? I suspect you may be right and they're pumping a load into 5nm in the hope of writing this off as a really bad few years, speeding up progression and regaining the lead.Quote:
Originally Posted by Tabbykatze
Intel's 10nm is reportedly as dense as the competition's 7nm so they aren't quite as far behind as it sounds.
Their 7nm tech being notably different to the 10nm is weird. Wonder if they'll end up being complimentary instead of a simple replacement?
I look at that picture and can't help thinking that 22nm was 2011, 14nm was 2013, 10nm was meant for 2015, 7nm for 2017 and 5nm for 2019. At least that's what it looks like to me.
Are they making a couple of tweaks and rebranding their failed 10nm as 7nm? As said above, intels 10nm is similar to their competitions 10nm.
A single fab being dedicated to it doesn't make them seem *that* optimistic.
Not quite, Intel's 10nm is almost equivalent to everyone else's 7nm apart from one thing: everyone else's 7nm works.Quote:
Originally Posted by Stukes
Intel’s 14nm process is equivalent to other's 10nm. also intel's 10nm is equivalent to other's 7nm/8nm.
(intel 10nm has 100MTr/mm2, TSMC 7nm has 67MTr/mm2 due to backend)
no one uses full 7nm yet. tsmc uses multipatterning of 10nm process to reach 7nm. samsung didnt want it so they went straight to 7nm EUV which takes time to develope.
intel can do risk production of 10nm but yield rate is bad atm cuz of monolithic chip design.
Zing!Quote:
Originally Posted by DanceswithUnixNot quite, Intel's 10nm is almost equivalent to everyone else's 7nm apart from one thing: everyone else's 7nm works.Quote:
Originally Posted by Stukes
Repeat of that old story of tortoise catching up with the sleeping hare, only that Intel failed to learn from previous round and got caught again with the competition.
Depends on how you measure equivalent. Transistor density - i.e. the thing that affects Intels costs and has no direct effect on the end user, yes. Performance - voltage/frequency/power consumption - no. It's more like the competition's 10nm. i.e. on the metrics that those using the Processors should care about, Intel's 10 nm is nothing special.Quote:
Originally Posted by DanceswithUnix
Pre empting those that try to suggest that users should care about transistor density due to cost: Yes but Intel can only (and absolutely will) charge what people are willing to pay. Peoples willingness to pay is not connected to Intel's costs.