Monday, October 14th 2019
Intel Scraps 10nm for Desktop, Brazen it Out with 14nm Skylake Till 2022?
In a shocking piece of news, Intel has reportedly scrapped plans to launch its 10 nm "Ice Lake" microarchitecture on the client desktop platform. The company will confine its 10 nm microarchitectures, "Ice Lake" and "Tiger Lake" to only the mobile platform, while the desktop platform will see derivatives of "Skylake" hold Intel's fort under the year 2022! Intel gambles that with HyperThreading enabled across the board and increased clock-speeds, it can restore competitiveness with AMD's 7 nm "Zen 2" Ryzen processors with its "Comet Lake" silicon that offers core-counts of up to 10.
"Comet Lake" will be succeeded in 2021 by the 14 nm "Rocket Lake" silicon, which somehow combines a Gen12 iGPU with "Skylake" derived CPU cores, and possibly increased core-counts and clock speeds over "Comet Lake." It's only 2022 that Intel will ship out a truly new microarchitecture on the desktop platform, with "Meteor Lake." This chip will be built on Intel's swanky 7 nm EUV silicon fabrication node, and possibly integrate CPU cores more advanced than even "Willow Cove," possibly "Golden Cove."
The HardwareLuxx article making these explosive revelations attributes the sudden change in Intel's plans to the company not being able to scale clock-speeds of "Ice Lake" high enough to establish product leadership. It feels "Skylake," which has IPC parity with "Zen 2," has enough scalability and clock-speed headroom to stay competitive with AMD at high clock-speeds. The company will augment next-generation uncore (revamped memory controllers, support for PCIe gen 4.0, Gen12 iGPU, etc.), with "Skylake" CPU cores, over time. Other areas where Intel could grow its mainstream desktop silicon is cache rebalancing similar to its HEDT chips, and implementing the Mesh Interconnect to maintain low latencies as core-counts enter two-figures.
Interestingly, 10 nm "Ice Lake" remains on Intel's enterprise roadmap, where the company appears more desperate not to cede market-share to AMD, especially as businesses around the world set their 5G plans rolling, springing a cycle of hardware updates in the data-center. 2020 could see the introduction of Xeon Scalable processors based on 10 nm "Ice Lake" microarchitecture with "Sunny Cove" CPU cores. In 2021, the company will introduce the "Sapphire Rapids" Xeon processor with even more cores and larger I/O connectivity, spearheaded with PCI-Express gen 5.0.
Update Oct 15th: Intel has released a statement, denying these claims, read more here.
Source:
HardwareLuxx.de
"Comet Lake" will be succeeded in 2021 by the 14 nm "Rocket Lake" silicon, which somehow combines a Gen12 iGPU with "Skylake" derived CPU cores, and possibly increased core-counts and clock speeds over "Comet Lake." It's only 2022 that Intel will ship out a truly new microarchitecture on the desktop platform, with "Meteor Lake." This chip will be built on Intel's swanky 7 nm EUV silicon fabrication node, and possibly integrate CPU cores more advanced than even "Willow Cove," possibly "Golden Cove."
Interestingly, 10 nm "Ice Lake" remains on Intel's enterprise roadmap, where the company appears more desperate not to cede market-share to AMD, especially as businesses around the world set their 5G plans rolling, springing a cycle of hardware updates in the data-center. 2020 could see the introduction of Xeon Scalable processors based on 10 nm "Ice Lake" microarchitecture with "Sunny Cove" CPU cores. In 2021, the company will introduce the "Sapphire Rapids" Xeon processor with even more cores and larger I/O connectivity, spearheaded with PCI-Express gen 5.0.
Update Oct 15th: Intel has released a statement, denying these claims, read more here.
148 Comments on Intel Scraps 10nm for Desktop, Brazen it Out with 14nm Skylake Till 2022?
and from what I can see in the tpu review 9600k is not only faster than 3900x in games but in a whole bunch of other typically single thread heavy office tasks.other ones use multi threading at least to some extent so it's a pointless comparison.
that said,I can see ryzen 4000 deliver really good performance.with 3000 they made good improvements so hopefully they'll continue on the right track.
AM5/DDR5 will be 2021 at the earliest.
When it comes to Zen 3 then current leaks point to unified 32MB L3 for one chiplet (currently each CCD within a chiplet has it's own 16MB cache) and 3-way SMT on desktop and 4-way in servers. Supposedly also 200Mhz clock bump and 8-10% IPC uplift. So more meaningful upgrade than Zen+ was to Zen but not as major as Zen 2 was to Zen+.
Besides considering the massive AM4 install base i would not rule out the possibility of cross socket compatible Ryzen 5000 series that has both DDR4 and DDR5 capable memory controller built in.
But they will be in very limited supply, so game over man.
AMD is on the verge of changing sockets after 2020 and folks are freaking the F out! :confused:
Food for thought:
2022 , wow not great news so were getting two more years of optimised skylake.
I expect triple digit % gains in CPU market share.
It will get really spicy if they push ML to 2023.
That said, ZEN 3 will be on Socket AM4. ZEN 4 may see Socket AM4+. But 99% of the user base that builds there own PCs usually always upgrade there CPUs and Motherboards together. To get to my point, if AMD releases a new socket, that doesn't make AM4 obsolete. Never did in the past and never will in the future, LOL I think Jim Keller will deliver but it will take 4 - 5 years from when he was hired. The question is when and like you said, density/process issues may also be a limiting factor overall.
AMD has the perfect opportunity to continue to deliver and increase its market share while Intel is struggling.
So why would AMD release a new socket for just one year when DDR5 (and therefore AM5) is most likely slated for 2021.
It makes absolutely no sense... They'll just release X670 and a new B board on AM4 and those'll be the last AM4 parts.
I mean, they could, if they can make it compatible with Zen4 and DDR5 and Zen 3(DDR4). But what are the likes of that.
They're both working on on die cach but Intel can not change their designs as easily as their competition, not just Amd here.
www.guru3d.com/news-story/rumor-zen-3-sees-yet-another-ipc-significant-gain.html
For both Intel and AMD a big L4/5 cache has to be on the cards and that plus extra internal cache should accelerate IPc gains a bit.