Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Intel promises 7 nm products in 2021; first product will be a GPU
https://www.anandtech.com/show/14312/intel-process-technology-roadmap-refined-nodes-specialized-technologies
After the disaster that is Intel's 10 nm process node, how long will it take for them to be competitive again? 10 nm probably isn't going to do it, especially with the claims of sharply reduced volume. 7 nm could be a chance for Intel to be competitive again, and they're saying that's coming in 2021 with a GPU targeted at data centers.
Of course, by 2021, Samsung and TSMC will be on to 5 nm or so. But those process nodes won't necessarily be better than Intel's 7 nm, as process node naming schemes are more than a little dubious. For that matter, TSMC's new 7 nm node isn't necessarily better than Intel's 10 nm node would have been if they had been able to deliver what they claimed. Which, of course, they didn't.
Of course, that leaves open the question of whether Intel's schedule will slip. And even if they do manage to push something out the door on 7 nm in 2021, it doesn't necessarily follow that it will be followed closely by beefy Xeon parts. But with AMD Rome coming so soon, it should be interesting to see how much traction AMD can get in the server market when they have a process node advantage not merely for the first time ever, but likely to be sustained for years.
After the disaster that is Intel's 10 nm process node, how long will it take for them to be competitive again? 10 nm probably isn't going to do it, especially with the claims of sharply reduced volume. 7 nm could be a chance for Intel to be competitive again, and they're saying that's coming in 2021 with a GPU targeted at data centers.
Of course, by 2021, Samsung and TSMC will be on to 5 nm or so. But those process nodes won't necessarily be better than Intel's 7 nm, as process node naming schemes are more than a little dubious. For that matter, TSMC's new 7 nm node isn't necessarily better than Intel's 10 nm node would have been if they had been able to deliver what they claimed. Which, of course, they didn't.
Of course, that leaves open the question of whether Intel's schedule will slip. And even if they do manage to push something out the door on 7 nm in 2021, it doesn't necessarily follow that it will be followed closely by beefy Xeon parts. But with AMD Rome coming so soon, it should be interesting to see how much traction AMD can get in the server market when they have a process node advantage not merely for the first time ever, but likely to be sustained for years.
Comments
Intel still holds an advantage on both of those, for now, even without 10nm. That may not last through the year but I am thinking it will hold for another generation at the least.
AMD is a lot small company with a lot fewer resources - Bulldozer was for AMD a nearly decade long disaster of much larger magnitude than 10nm and the survived to come back from that.