As said above, the title of the article may be incorrect, but next year AMD will most likely be on 7nm EUV or 6nm, and with Intel still being on 10nm in 2020, in fact desktop parts will only appear by then, I'd say the conclusion / result is the same, Intel will probably only be able to catch up to AMD in 2021. By then it will probably be AMD on 6nm or 5nm and Intel on 7nm.
Originally Posted by guttheslayer
Hold up, you forgot the 15% IPC increase for AMD took them 2 years from 2017 to 2019. They didnt increase IPC every year. Leading me to believe the next 15% jump for Ryzen is in 2021, and will be head to head with Intel "7nm" node.
In fact in 2021 we might see Ocean Cove with 2.5D stacking for desktop chip on 7nm? That will be a very interesting year ahead.
And the article should change name to "overtake" AMD 7nm chip.
Zen+ did increase IPC, by around 2.5% - 3% due to the several cache and memory latency optimizations. But Zen+ is a refinement of the original Zen design, as AMD said after Ryzen 1000 series launched back in 2017, the way the CPUs came out was the worst case scenario possible, so there was room for it to be polished, and it was with Zen+. The lower cache latency and the more gracious boost behaviour of Zen+ were originally scheduled for the original release, but didn't make it in time. The difference here is that next year, if the roadmaps are to be believed, we won't be getting a Zen 2+, but Zen 3, so a more substantial upgrade to the arch is to be expected. In fact, some of the improvements scheduled for Zen 3 were implemented early in Zen 2, which hopefully means that they are executing so well that they had more time to do the next round of improvements early too, and thus fill the gap with improvements that would originally go into Zen 4.