Pemalite said:
haxxiy said:
Well, we will see if they can tape out their designs this year yet. As I mentioned to JEMC, I just don't think they can, because somewhere, somehow, there will surely be delays. Even Intel and Samsung have their setbacks and end up not matching up to their roadmaps, AMD and Global Foundries even more so. Even Zen was late by +-6 months, after all, same with Vega 10, and that's on a somewhat mature process, not one which still is mostly on paper.
|
Just because Zen was late, doesn't mean Zen+ will be. Seperate teams.
We don't know if Zen+ will be using Global Foundries 7nm anyway. It might continue to use Samsungs 14nm Finfet at Global Foundries or might not use either and use a 12nm process, if Global Foundries tools it up appropriately.
Chances are we will see chips drop before tax time next year. And there is a plethora of reasons for that.
One is to get in-on the OEM upgrade cycle.
|
I do assume there will be an improved Zen on 14 nm next year, with improved memory compatibility and higher clock speeds. Since their 7 nm architecture has been renamed Zen 2, AMD might choose to call this improved design Zen+. But I don't think AMD has the money to redesign their chips intended for 7 nm back to 14 nm, like Nvidia did with Maxwell. As for 12 nm... it seems even less likely since it's a node without Finfets, that would probably be even harder to redesign their Zen chips around. It will be worth the wait for the 7 nm node even if it is months late, the front end of line is bound to be much denser than Intel's 14 nm. It will be close to Intel's 10 nm than GloFo's 14 nm was to Intel's 14 nm.
