Seems like Intel might use 10nm only on low power mobile and server CPUs and keeps using 14nm for Client desktop CPUs until 7nm, due to the high cost and margins.
Actually it seems that reluctance has ended or they had enough time to do the port while the time was too short for kabys and cofees.I bet Rocket Lake is 14nm Skylake core as well. There is no reason to think otherwise. Simple math. Anything newer was developed for 10nm and Intel is reluctant to port it for the past several years.
Interesting point is what clocks and TDP will have Rocket or even Comet Lake 8/10 core CPUs.
Based on what inputs?So I used one of those chip yield calculators... based on that I would put the O/U on Intel's current defect rate for 10 nm at 3.5/sqcm (super awful); and the defect rate when the 8121U launched at 8-9.
Yes, should be PRQ."Icelake 10nm Qual(ification?) Q2"
Yeah they clarified in the call - Icelake will be qualifying for production this quarter.Yes, should be PRQ.
oofwhich basically meant products are moving through the factory 2x faster (improved cycle time)
0*20 is still 0.also the raising volume comment is just more volume of 10nm products this year than they expected.
Fair enough, if you can't increase yields you can still crank up the speed and move more wafers through. It's still low volume, but it's twice™ the low volume.They also commented on the "Factory Velocity" comment which basically meant products are moving through the factory 2x faster (improved cycle time), also the raising volume comment is just more volume of 10nm products this year than they expected.
Think this might backfire, as in taking shortcuts?Yeah they clarified in the call - Icelake will be qualifying for production this quarter.
They also commented on the "Factory Velocity" comment which basically meant products are moving through the factory 2x faster (improved cycle time), also the raising volume comment is just more volume of 10nm products this year than they expected.
No, making some steps faster.as in taking shortcuts?
That we don't know.If this could always be done fairly easily, why was it never implemented?
So possibly 10nm in name only.No, making some steps faster.
They use Co and Ru extensively in 10nm, and these are ass for cycle times.
That we don't know.
No, the very same node, just with new tooling to cut the cycle time on some steps.So possibly 10nm in name only.
A sign of Rocket Lake?View attachment 5536
https://tweakers.net/nieuws/151984/...l-in-2021-nog-desktop-cpus-op-14nm-maakt.html
No 10nm Desktop-CPU until 2022? If it's true then... well, RIP Intel.
So the pic says:
4-core Y in 2019 - Amber Lake 14 nm
4-core Y in 2020 - Comet Lake 14 nm
4-core Y in 2021 - Tiger Lake 10 nm
However, I saw a different table elsewhere:
4-core Y in 2019 - Amber Lake 14 nm
4-core Y in 2019 (limited) - Ice Lake 10 nm
4-core Y in 2019 to 2020 - Comet Lake 14 nm
4-core Y in 2020 - Tiger Lake 10 nm
https://wccftech.com/intel-desktop-mobile-cpu-roadmap-leak-14nm-comet-lake-10nm-ice-lake-tiger-lake/
OK but then to ask a stupid question:Because there are 2 roadmaps in the latter article: desktop and mobile
OK but then to ask a stupid question:
What is desktop Y series?