PCs With Cannon Lake CPUs Won't be on Sale Until the Very End of 2019

DooKey

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
13,552
In yesterdays conference call Intel confirmed that Cannon Lake will be out just in time for the 2019 holidays. We already knew they had delayed mass production until 2019, but waiting till the end of next year is huge. As a result Intel is hoping 14nm chips like Whiskey Lake and Amber Lake will tide their client market over for the next 16 to 18 months. I'm sure those chips will be fine performers, however, AMD is releasing 7nm chips next Spring and that's bound to capture some good PR. Good luck Intel.

So, this isn’t another delay as such, just a confirmation of exactly when in 2019 the launch will happen – but it’s kind of a disappointment seeing as the timeframe sounds pretty much as late as it can be (with consumer PCs launching by December).
 
Tiburon1186 said:
Maybe they will incorporate a hardware fix for specter?
Click to expand...
I'm hoping we get at least some hardware changes that make any future fixes for Spectre variants less expensive in terms of performance. I don't want to buy a chip in the future that gets more and more bogged down with firmware fixes as the years go on. But I also feel that perhaps AMD may have a better handle on this. The Zen CPUs have been more resilient than Intels chips, and AMD has made statements about additional hardware measures against this style of attack.

The 9990K 8c/16t looks tempting, but I will bide my time, I'm really in no rush being on a 7770K, and I'll see what AMD brings to the table next year. Seems to me it'll be pretty awesome.
 
how many pluses should be allowed to get added to the end of a node before we get skeptical? 1? 2? 14nm+++ inbound!
 
DPI said:
Good time to buy.
Click to expand...
I haven't followed their stock too closely. Will a new CEO be able to turn around the company? I bought some MSFT shortly before Steve Ballmer left. It's had a really nice rebound - I've sold half of it with a tidy profit. I'm playing with the dealer's money on what I have left (for now, just holding.)
These big companies get lax and lazy. You can't necessarily blame them - there wasn't a lot of competition, money was flowing in, so why bother? Problem is, there is always a disruptive outside force that comes along with a superior product.
If Intel can start firing on all cylinders, get out a superior product to AMD CPU's - they can come back roaring. A few other areas of interest for Intel: storage and GPU. Both of storage and GPU seem like fertile grounds for growth.
 
steakman1971 said:
I haven't followed their stock too closely. Will a new CEO be able to turn around the company? I bought some MSFT shortly before Steve Ballmer left. It's had a really nice rebound - I've sold half of it with a tidy profit. I'm playing with the dealer's money on what I have left (for now, just holding.)
These big companies get lax and lazy. You can't necessarily blame them - there wasn't a lot of competition, money was flowing in, so why bother? Problem is, there is always a disruptive outside force that comes along with a superior product.
If Intel can start firing on all cylinders, get out a superior product to AMD CPU's - they can come back roaring. A few other areas of interest for Intel: storage and GPU. Both of storage and GPU seem like fertile grounds for growth.
Click to expand...
That likely depends on what you mean by turn around, at least in the short term. As an outsider it's hard to know exactly what is wrong with Intel's 10nm process, but they've always claimed that the process node works, but yields are to low. Prior to AMD's competitiveness, I generally took that to mean: "we have no competition and the yields are to low to support our very high margins." That is to say, they could turn the 10nm tap on with lower than desired yields, but choose not too because they could still make great profits reusing 14nm. I'm not saying Intel's 10nm doesn't have problems, it does, but rather that until this news, they did a good job giving the impression it's under control and why change if there's no competition? That aligned well with their architecture/functional segmentation which unnecessarily drove customers to the next higher tier of product in many cases. It was clear Intel was playing a margins game and investing big into 10nm (despite it's problems, Intel was very aggressive with the targets).

However, now Intel does have competition in the lucrative enterprise/cloud segment and in the near future might be facing competition on the lower-power laptop/desktop end as well, having already ceded smartphones/tablets to ARM architectures. Any delay at this point can only be interpreted as, IMHO, yields are so bad they can't reasonably turn the tap on because it would be a financial disaster, far worse than loosing significant market share. The damage could be anywhere between, it would crush margins to the point our shareholders would revolt to Intel can't even make money on each unit. The margins are so high, it's unlikely they can't make money today on 10nm. Perhaps it's a gamble that they can recover from any real damage as the enterprise segment shifts to heterogeneous architectures which Intel has done a good job positioning themselves for (e.g. FPGA, integrated 100+ Gbps I/O fabric, tailored AI co-processors, etc.).

IMHO, that's the only bright spot for Intel right now. AMD is still playing last decades game, but it's a game that hasn't quite played out just yet. AMD was very wise and needs all those PCIe lanes/cores to support this shift in the enterprise/cloud markets; just take a look at the newest AWS instances. The next stage in this game is really cool, purpose built processors using EMIB or interposers, etc. Intel isn't dead yet, but they are in big trouble as far as their core business practices are concerned; making one basic architecture and then scaling it up and down while charging big prices with massive margins (and segmenting the living daylights out of the markets) won't work for the future. However, maybe if Intel can position itself just right with the heterogeneous processors they can keep margins super high, but they've got to change.
 
T
LuxTerra said:
That likely depends on what you mean by turn around, at least in the short term. As an outsider it's hard to know exactly what is wrong with Intel's 10nm process, but they've always claimed that the process node works, but yields are to low. Prior to AMD's competitiveness, I generally took that to mean: "we have no competition and the yields are to low to support our very high margins." That is to say, they could turn the 10nm tap on with lower than desired yields, but choose not too because they could still make great profits reusing 14nm. I'm not saying Intel's 10nm doesn't have problems, it does, but rather that until this news, they did a good job giving the impression it's under control and why change if there's no competition? That aligned well with their architecture/functional segmentation which unnecessarily drove customers to the next higher tier of product in many cases. It was clear Intel was playing a margins game and investing big into 10nm (despite it's problems, Intel was very aggressive with the targets).

However, now Intel does have competition in the lucrative enterprise/cloud segment and in the near future might be facing competition on the lower-power laptop/desktop end as well, having already ceded smartphones/tablets to ARM architectures. Any delay at this point can only be interpreted as, IMHO, yields are so bad they can't reasonably turn the tap on because it would be a financial disaster, far worse than loosing significant market share. The damage could be anywhere between, it would crush margins to the point our shareholders would revolt to Intel can't even make money on each unit. The margins are so high, it's unlikely they can't make money today on 10nm. Perhaps it's a gamble that they can recover from any real damage as the enterprise segment shifts to heterogeneous architectures which Intel has done a good job positioning themselves for (e.g. FPGA, integrated 100+ Gbps I/O fabric, tailored AI co-processors, etc.).

IMHO, that's the only bright spot for Intel right now. AMD is still playing last decades game, but it's a game that hasn't quite played out just yet. AMD was very wise and needs all those PCIe lanes/cores to support this shift in the enterprise/cloud markets; just take a look at the newest AWS instances. The next stage in this game is really cool, purpose built processors using EMIB or interposers, etc. Intel isn't dead yet, but they are in big trouble as far as their core business practices are concerned; making one basic architecture and then scaling it up and down while charging big prices with massive margins (and segmenting the living daylights out of the markets) won't work for the future. However, maybe if Intel can position itself just right with the heterogeneous processors they can keep margins super high, but they've got to change.
Click to expand...



The lower yields has more problems than just profit margins. If the yields are low, then bigger chips will have a zero yield, therefore you can only produce smaller chips, as Intel is currently doing with it. With low yields, it means the defective parts on the wafer is high. With bigger chips the chances of those defective parts land right inside a CPU becomes too high.
 
Johan Steyn said:
T




The lower yields has more problems than just profit margins. If the yields are low, then bigger chips will have a zero yield, therefore you can only produce smaller chips, as Intel is currently doing with it. With low yields, it means the defective parts on the wafer is high. With bigger chips the chances of those defective parts land right inside a CPU becomes too high.
Click to expand...
True, larger dies certainly exacerbate the situation. I imaging Intel isn't too happy with it's monolithic dies on 10nm given AMD kicking off the core wars.
 
You must log in or register to reply here.
Back
Top