Intel Delays 10nm Cannon Lake Processors Again

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
13,500
Intel has announced they are not going to ship 10nm processors in high volume this year. This move seems to confirm rumors that their 10nm process still isn't up to steam and they need more time to get things right. While I'm all for getting things right I'm also impatient to get my new shiny 10nm processors and I wish Intel would get their act together sooner than later. They better hurry up before their competitors release next generation process processors. You can read the earnings report here.

Intel is on solid footing, in other words, though pesky challenges remain in manufacturing its next-generation 10nm parts. Krzanich acknowledged as much during an earnings call, attributing the delay to difficulties in getting 10nm yields to where they need to be. So rather than push to ship 10nm in volume this year, Intel is giving itself some additional time to sort things out.
 
So we will get more marketing crap like 10nm class processors or 10nm like performance in the meantime.
 
Why do I read this as "AMD isn't taking the market share we were worried about so we're going to push this out just like what NVDA is doing on the graphics card side"
 
Intel said:
In the first quarter, Intel saw strong performance from data-centric businesses, which accounted for approximately half (49%) of Intel's revenue, an all-time high. The Data Center Group (DCG) achieved growth in all market segments and saw increasing adoption of Intel® Xeon® Scalable processors, including for artificial intelligence workloads. Non-Volatile Memory Solutions Group (NSG) revenue grew 20 percent as strong demand for storage continued. The Programmable Solutions Group (PSG) won new designs with server OEMs adding Intel’s field programmable gate array (FPGA) acceleration to their data center offerings, and strong demand from retail and video customers drove first-quarter growth in the Internet of Things Group (IOTG). The Client Computing Group (CCG) continued its strong execution and introduced a new lineup of highperformance mobile products including the 8th Gen Intel® Core™ i9 processor and a new Intel Core platform extension that brings together the benefits of 8th Gen Intel Core processors with Intel® Optane™ memory. We continue to make 14 nm process optimizations and architectural innovations in both data center and client products that will be coming this year. Intel is currently shipping low-volume 10 nm product and now expects 10 nm volume production to shift to 2019.

Interesting they don't bring up consumer market.

I wonder how much of this revenue is because server markets beefing up due to spectre/meltdown?

I have never seen these kind of number increases out of intel before. So something is suspect.
 
Which will come first? Intel 10nm desktop CPU's, or AMD 7nm desktop CPU's? Anyone want to wager?
 
Why do I read this as "AMD isn't taking the market share we were worried about so we're going to push this out just like what NVDA is doing on the graphics card side"
Yeah Intel wants you to believe engineering new stuff and processes is difficult, when in reality those guys have just been playing golf for the last 15 years. Now if the AMD engineers could just leave the hookers and blow alone for one weekend we'd have 200% IPC gains by summer.
 
Until Intel complete redesigns the chip , Cannon lake is no better than any Intel chip before it because of the specter and meltdown fixes will bring a performance penalty. Ill be riding out my chip for a long run this time
 
Yeah Intel wants you to believe engineering new stuff and processes is difficult, when in reality those guys have just been playing golf for the last 15 years. Now if the AMD engineers could just leave the hookers and blow alone for one weekend we'd have 200% IPC gains by summer.

Am I the only one that read the article?

Krzanich acknowledged as much during an earnings call, attributing the delay to difficulties in getting 10nm yields to where they need to be. So rather than push to ship 10nm in volume this year, Intel is giving itself some additional time to sort things out.
 
Am I the only one that read the article?

Are you arguing against my point, or your own? The article is also more than the two sentence excerpt in the OP.
Intel is on solid footing, in other words, though pesky challenges remain in manufacturing its next-generation 10nm parts. Krzanich acknowledged as much during an earnings call, attributing the delay to difficulties in getting 10nm yields to where they need to be. So rather than push to ship 10nm in volume this year, Intel is giving itself some additional time to sort things out.
By your logic, the reason Tesla is having Model 3 "production issues" is due to no competition in the EV market.
 
Am I the only one that read the article?

Face saving talk. Read this article. Intel was supposed to ship 10nm in 2016, it's 2018 now, with Intel forced to delay, delay, over and over with no end in sight. Intel just hired Jim Keller, and restructured their manufacturing division not because everything is fine and dandy, but cause Intel completely fucked up and needs to start "gluing" chips together to make up for yields.

Meanwhile, TSMC is set to begin volume 7nm (equivalent somewhat to 10nm) production. Remember, Intel likes to boast their foundry tech is ahead of everyone else, well that's looking less and less true.

If Intel can't get 10nm by the time AMD starts running 7nm Zen 2 at GlobalFoundries, it will be bloody for Intel.
 
Last edited:
Face saving talk. Read this article. Intel was supposed to ship 10nm in 2016, it's 2018 now, with Intel forced to delay, delay, over and over with no end in sight. Intel just hired Jim Keller, and restructured their manufacturing division not because everything is fine and dandy, but cause Intel completely fucked up and needs to start "glueing" chips together to make up for yields.

Meanwhile, TSMC is set to begin volume 7nm (equivalent somewhat to 10nm) production. Remember, Intel likes to boast their foundry tech is ahead of everyone else, well that's looking less and less true.

If Intel can't get 10nm by the time AMD starts running 7nm Zen 2 at GlobalFoundries, it will be bloody for Intel.

I appreciate the anandtech article, it definitely paints a different picture than the PCGamer article. Is GlobalFoundries able to do 7nm CPU yet? If I remember, GF has an exclusive supply arrangement with AMD for CPU's.
 
It's all fine and dandy to me. Haven't owned an AMD processor since the FX-53, held onto that till the Core 2 Quad dropped.

I'm loving what I'm seeing with the Zen+ releases this year. Waiting for the Threadripper 2 to drop which I believe is Q3. I intend to go with the lower core variant whenever it drops, I just want the quad channel RAM more than anything.

It's exciting times, just wish the Radeon end of things would give Nvidia a run for it's money.
 
I appreciate the anandtech article, it definitely paints a different picture than the PCGamer article. Is GlobalFoundries able to do 7nm CPU yet? If I remember, GF has an exclusive supply arrangement with AMD for CPU's.

Not yet, if my memory of recent news article is correct, so AMD may still be behind Intel if GF trips up again, but this is cutting it close. Current plans according to Anandtech have GF beginning production 2H 2018. As is, 10nm has been delayed for volume in 2019. As for the supply arrangement, I think it applies to all wafers, GPU or CPU. One particular note, AMD has the option to get wafers from other foundries in return for paying a fee to GF. Already, AMD has said they are making a Navi 7nm machine learning oriented GPU at TSMC thats in currently in sampling. But I think the vast majority of CPUs and GPUs will be GF. Console chips are made at TSMC through.
 
AMD will now easily take more market share with their 7nm processors. Taking all the server market share, also have thier APU's in intel cpu's now. AMD making quite the comeback, they will be king of the hill soon. AMD's 2700X already whooping the dogshit out of 8700k in multi-threaded applications. Lovin it.
 
AMD will now easily take more market share with their 7nm processors. Taking all the server market share, also have thier APU's in intel cpu's now. AMD making quite the comeback, they will be king of the hill soon. AMD's 2700X already whooping the dogshit out of 8700k in multi-threaded applications. Lovin it.
Everybody better start buying Intel & Nvidia again then to keep up the competition. :p
 
I just upgraded to an 8700K, my God... it's like stepping into a Ferrari when coming from my old 3570K. I'm good for the next five years. Intel... take your time, I'm patient.
 
Probably to have Keller take a look at them before they launch them.
 
My understanding is that at about 10nm, quantum mechanics become a huge problem and to expect AMD to have similar issues with 7nm. Well, that's what the resident on his way to a PhD electrical engineer claims.
The conspiracy theorist in me thinks that Intel just wants to release around the same time as Zen 2 to rain on their parade.
There is also the hope they just might be working out meltdown and specta.
 
My understanding is that at about 10nm, quantum mechanics become a huge problem and to expect AMD to have similar issues with 7nm. Well, that's what the resident on his way to a PhD electrical engineer claims.
The conspiracy theorist in me thinks that Intel just wants to release around the same time as Zen 2 to rain on their parade.
There is also the hope they just might be working out meltdown and specta.

10nm seems like a pain, Nvidia also ditched it.
 
No shit? Isn't this like the upteenth time now? This kinda makes sense why they are bring the God of making shit happen, Jim Keller in.

+1, this is the guy that made Athlon64 and Zen

This is great for consumers. Intel needs it's nose bloodied and AMD needs the market share.

Even if AMD brings 7nm quickly(still in prototype stage I heard) and performs better than Intel 10nm. It won’t make a dent. This would be the Athlon/P4 debate all over again. Consumers are trained to equate “Intel” with “best”. It would take multiple consistent product cycles of beating Intel to get people to notice.
 
Probably to have Keller take a look at them before they launch them.

He could do that. But that likely isn't his role. That's like hiring an outside consultant to come in and say "here evaluate all our windows code"

What he will more than likely do is analyze various performance metrics and look for avenues to lower consumption or increase ipc or both. He'll recommend design strategies like how the pipeline will be set up as well as prediction algorithms.

I will say though Jim Keller is pretty much all that and a bag of chips given his history.
 
Yeah Intel wants you to believe engineering new stuff and processes is difficult, when in reality those guys have just been playing golf for the last 15 years. Now if the AMD engineers could just leave the hookers and blow alone for one weekend we'd have 200% IPC gains by summer.

200% IPC gains if we ignore the laws of physics and the "IPC wall"

https://hardforum.com/threads/when-did-golden-age-of-computing-end.1931057/

+1, this is the guy that made Athlon64 and Zen

Current chips aren't made by a single person but by a team. Moreover, he wasn't lead architect of the teams that made those you mention, the lead architects are

K8 --> Fred Weber

Zen --> Mike Clark

https://hardforum.com/threads/jim-keller-joins-intel.1959438/#post-1043602816
 
Maybe internally, Intel has been directing its R&D at future Apple SoC foundry business over their own Crap Lake cores. Selling Apple modems and fabbing their iPhone/iPad/future Mac SoC is probably a bigger business than its x86 processors at this point. It's not like they've had to answer a challenge from AMD up until this point.
 
  • Like
Reactions: ecktt
like this
Back
Top