By: blue (blue.delete@this.blue.com), July 31, 2021 11:34 am
Room: Moderated Discussions
Intel can claim what they want, but AMD is making vastly denser CPUs with similar clocks and similar/better performance per watt and area despite not using these specialized cells nearly as much.
Seems like Intel's density is either BS, or both their architects and process team are terrible...
Their 14nm "wide" Knights Landing architecture was ~682mm with ~7.1b transistors, or a bit over 10 million/mm^2.
Zen1, aimed at higher performance dense cores at the "it's 20nm class node renamed" was ~213mm with ~4.8b transistors. A bit over 22 million/mm^2.
Intel supposedly is twice as dense, so are we saying the cache is what is giving Zen over 4x the effective density? That seems pretty insane to me.
That being said, public SoCs are loser to 1.8 than 1, so idk about the whole "1.8 is overrated". Because if the total SoC is closer to 1.8 than 1....
PLEASE, inform me into what silly things I missed due to me being the idiot that I am.
Seems like Intel's density is either BS, or both their architects and process team are terrible...
Their 14nm "wide" Knights Landing architecture was ~682mm with ~7.1b transistors, or a bit over 10 million/mm^2.
Zen1, aimed at higher performance dense cores at the "it's 20nm class node renamed" was ~213mm with ~4.8b transistors. A bit over 22 million/mm^2.
Intel supposedly is twice as dense, so are we saying the cache is what is giving Zen over 4x the effective density? That seems pretty insane to me.
That being said, public SoCs are loser to 1.8 than 1, so idk about the whole "1.8 is overrated". Because if the total SoC is closer to 1.8 than 1....
PLEASE, inform me into what silly things I missed due to me being the idiot that I am.
