C'mon people who know... make this site actually useful for rumors.
12 replies (most recent on top)
@Iup0aPQ-1tuf Ah as I suspected 10nm a flaming disaster and will never yield volume or revenue that intel needs. Who are the TMG monkeys that are running it, they need all be fired. Wonder what the backup plan is, limp along and alone on 14nm while apple and Samsung ramp there 10nm and 7nm what an egg in the face and a kick in the groin to Intel and TMG
CannonLake will most likely be only a proof-of-concept for 10nm, maybe some limited sampling and low volume for limited customers very late in 2017/first half of 2018, the yields on the new process will not allow anything more. The "real" 10nm product will be IceLake and I don't think they'll be able to get it out to market until mid to late 2018. Moore's Law is coming to an end soon ...
@Iup0aPQ-1kjw Nothing adds more power / performance pop tha moving a design to the next node.
If Intel is doing both Canonlake and Coffee than it says that either 10nm isn't performing or they aren't close to delivering volumes and need to supply volume with a big new design on 14nm. Everything points to a disaster on the process /product here
Process has hit the wall at 10nm. Intel will be lucky if it gets a 10nm product out to market in 2018.
It is now a commodity market, and a race to the bottom in terms of pricing. Intel will not be able to compete in this market with their vertical cost structure and closed, monopolistic market strategy.
The end is coming fast, faster than expected ...
The reports on the internet say that there are going to be Cannon Lake and Coffee Lake products at the same time.
Wow we have both Kaby Lake and Coffee Lake on 14nm. The end is clearly here.
I keep hearing nothing adds more value to intel than moving to another process node as transistor shrinks, you decrease their size and the value is worth all the billions spent on R&D and more billions on the manufacturing.
To produce what effectively is two more designs to substitute the late 10nm is not only astounding it is the clearest sign yet the Moore's Law is dead. If intel has to slip it and actually produce two more generation on 14nm it means the 10nm process is really in big trouble.
The two new products are using the same process as the last two thus the die size and power will both likely go up, as nothing is for free.
Tick Tock Tick Tock, looks like the last two rounds signify the end of Moore's Law, what happened to this think called Canon Lake, the new design on 10nm.. just thrown away, what a train wreck that they had to cancel it. Somebody in TMG really should be fired because of this slip!
"Coffee Lake" available to drink at Starbucks.
So in sum, absolutely nothing concrete.
Thanks for the try anyways.
Apple for a while was "happy" using someone else's design than decided to design it themselves.
If you look at the benchmarks for the latest Aps processor and do the scale up you'd argue that Apple likely can do their own high performance Mac chip, but think it is only the fact the volumes are much lower that why not let BK bend over and apply lots of KY. We all know the fruit company gets special treatment at Intel, and if they were to ever lose it then it would be a signal to every box manufacture that they can do as well going and getting an ARM processor and porting the OS. Than the place would implode in a year or less, if it doesn't next year with 10nm. Google chrome box running cloud subscription MS/Outlook out of the Azure/AWS on ARM cloud, LOL
Coffee Lake looks to me like a product primarily to keep Apple happy while 10nm deals with uncertainty and yields. Apple buys the processors with GT3 graphics. BK will do anything to keep Apple, they are a premium brand, and now that they have struggled through Intel's BS during 14nm they are using that leverage to get their own tock.
Uhm, "leading edge" and "IoT" shouldn't even belong in the same sentence.
"Leading edge" and "memory" don't, either.
LOL high end chips that sell for 1000 aren't the big movers, today in client and mobile the average ASAP for the most part on volume is less than 200 bucks or so. You need that price point to move chips. When you get to anything more than 4 cores and 20 watts you are talking always plug in.
You hear about this thing called all day on once charge, always-on, always connected is what people want. You see the wall huggers in airports or the accessories that extend your phones battery life.
PCs whether more powerful or whatnot is chasing after the wrong direction.
Intel needs something to fill it's leading edge factories and get pricing at more than 10000/wafer at a minimum, ain't happening with IoT, memory, yeah new high end might but the volume is so low they can't fill four 10nm factories ( D1D, D1C, Ireland and the holy lands )
Now the reality is they only need two 10nm factories with the business they have, that is if they hit the yields they were used to in the past