Intel announces 10nm Cannon Lake NUC codename "Crimson Canyon" with AMD Radeon Graphics launching in September 2018

[AngryBeaver]

19 hours ago, cj09beira said:

the thing is that from what we know they are behind tsmc and maybe on par with glo fo but newer rumors points to intel scaling down 10nm as they aren't able to improve yields to acceptable levels, while in a few months tsmc will be producing 12-16 core cpus for amd. i have no doubt that they invest a lot into their products, but this time they messed up and there is no way around it. everyone is tackling the same problem, intel has been able to be ahead for a long time, but this time they wont be able too and it will hurt

Well my first response got killed by the work DLP... So second attempt.

 

So first off intel was quoting I think a 50-70% increase in transistor density moving to 10nm from 14nm. Even if they need to scale that back I fully expect to see them regain that as they move towards 10nm+ and the process becomes more streamlined and yields increase. If you look at how they compare 14nm++ to 12 nm you will see they have a really good design and it works well. 10nm should have no problem competing with or even pulling ahead of 7nm as it moves to 10nm+

 

Next, I don't think intel is ever worried about getting their first. They are not going to lose much market share with prebuilts and data centers. Which honestly is where most of the money is. Then you have to look at them saying it will be 2 years until it is ready... in the cpu market that is nothing. Most people do not upgrade cpu's more often than 3-4 years. Most leased machines are 2-4 years and servers can be sooner or much longer depending on the company and use. I mean I personally play to upgrade to a 9700k and pass down my current 8700k system. I can get a nice overclock on it probably 5.3-5.5 ghz and then sit on it for 2 years.

 

I guess what I am trying to say is that the only market intel MIGHT see a hit on is the custom pc market, which honestly is a very small portion of where they make their money. Even in this market I doubt we will see a huge shift and by the time most people are looking to upgrade their cpus the new 10nm stuff should be hitting the market. I think people are missing the bigger picture here and not looking at how this realistically plays out.

[cj09beira]

2 minutes ago, AngryBeaver said:

Well my first response got killed by the work DLP... So second attempt.

 

So first off intel was quoting I think a 50-70% increase in transistor density moving to 10nm from 14nm. Even if they need to scale that back I fully expect to see them regain that as they move towards 10nm+ and the process becomes more streamlined and yields increase. If you look at how they compare 14nm++ to 12 nm you will see they have a really good design and it works well. 10nm should have no problem competing with or even pulling ahead of 7nm as it moves to 10nm+

 

Next, I don't think intel is ever worried about getting their first. They are not going to lose much market share with prebuilts and data centers. Which honestly is where most of the money is. Then you have to look at them saying it will be 2 years until it is ready... in the cpu market that is nothing. Most people do not upgrade cpu's more often than 3-4 years. Most leased machines are 2-4 years and servers can be sooner or much longer depending on the company and use. I mean I personally play to upgrade to a 9700k and pass down my current 8700k system. I can get a nice overclock on it probably 5.3-5.5 ghz and then sit on it for 2 years.

 

I guess what I am trying to say is that the only market intel MIGHT see a hit on is the custom pc market, which honestly is a very small portion of where they make their money. Even in this market I doubt we will see a huge shift and by the time most people are looking to upgrade their cpus the new 10nm stuff should be hitting the market. I think people are missing the bigger picture here and not looking at how this realistically plays out.

the thing is, in early 2019 if a company wants a 48 core per socket cpu only amd will be able to provide, and 10nm will take a long time until the yields are good enough for the bigger dies as they are much harder to make, even then amd will still have the advantage as they will have much lower costs, when intel can finally start making those cpus amd will already have passed qualification of the 48 core cpus and will be selling as much as they produce mean while the real volume production of epyc starts now and we should start hearing about it in the winter, the market where amd will hit hard is not the consumer market its the server market, epyc really is mostly a proof that amd can do it, zen 2 will take it to another level as intel wont be able to compete against it for a while. from the specifications we got from tsmc intel and global foundries, intel's 10nm is about as dense as anyone elses 7nm but they are having a harder time getting it to work and from what we ear will scale back meaning they will be behind everyone else at least on density and with 7nm being a high performance node intel's single threaded advantage will be mostly gone, so sure 10nm will compete but it will be very late to the party, i bet intel will again resort to illegal practices to try and reduce amd's progress. (sorry for being all over the place)

[Brooksie359]

16 minutes ago, AngryBeaver said:

Well my first response got killed by the work DLP... So second attempt.

 

So first off intel was quoting I think a 50-70% increase in transistor density moving to 10nm from 14nm. Even if they need to scale that back I fully expect to see them regain that as they move towards 10nm+ and the process becomes more streamlined and yields increase. If you look at how they compare 14nm++ to 12 nm you will see they have a really good design and it works well. 10nm should have no problem competing with or even pulling ahead of 7nm as it moves to 10nm+

 

Next, I don't think intel is ever worried about getting their first. They are not going to lose much market share with prebuilts and data centers. Which honestly is where most of the money is. Then you have to look at them saying it will be 2 years until it is ready... in the cpu market that is nothing. Most people do not upgrade cpu's more often than 3-4 years. Most leased machines are 2-4 years and servers can be sooner or much longer depending on the company and use. I mean I personally play to upgrade to a 9700k and pass down my current 8700k system. I can get a nice overclock on it probably 5.3-5.5 ghz and then sit on it for 2 years.

 

I guess what I am trying to say is that the only market intel MIGHT see a hit on is the custom pc market, which honestly is a very small portion of where they make their money. Even in this market I doubt we will see a huge shift and by the time most people are looking to upgrade their cpus the new 10nm stuff should be hitting the market. I think people are missing the bigger picture here and not looking at how this realistically plays out.

Yeah Intel said they would lose market share to amd in the server market so saying that only the custom pc market is going to be hit is just not true. If you look at the deals AMD has made in the server market already it is clear that they are making good headway. 

[thorhammerz]

On 8/19/2018 at 2:29 PM, asus killer said:

still releasing dual cores in late 2018! Sweet mother of god 

Pentiums and Celerons till exist  (and will continue to exist, so long as the market for PCs that only need to run light office tasks still survives).

[DrMacintosh]

Wow, it took them the entire die space of a i3 to get a 10nm chip that was sellable. They have a long way to go.