What the fuck went so wrong?

Hmm i wonder what went wrong

Attached: original.jpg (960x550, 128K)

Imagine what kind of hell furnance a 10 GHz chip would be.

Attached: everything will be fire.png (506x609, 410K)

meanwhile tsmc already looking into 5nm

Attached: mld2zlcp78731.jpg (1200x1077, 238K)

Attached: Untitled.png (601x1664, 634K)

one layer of transistors multiply layers for metal wire.

Pure cancer

>moore's law
that faggot revised several times his "law" which was a topic of discussion often in his radio appearances. there's no scientific evidence behind this shit, it's just an estimation which is/was kept alive by amd, ibm, hpe, motorolla as intel's contributions to this shit, aside from a decade of lithography lead, is coming late to the party by 2 decades (OoO, superscalar, multicore, etc) or slowing themeselves down (x87, ia-64).

Also only one layer of "14nm+++"
Rest is done on older processes to cut cost.

"not the end of moore's law"
litterly it ended in like 2011-2015 depending how you measure it, there is a time limit, # transistors double every 2 years.
Its been dead for a long time.

>OoO
OwO

Why don't they keep it at like 30nm so cirtuits are more spread out and get less hot and can run at 7-8 GHz?

Attached: 161E5AA7E4924066843B876F4C0CBAD1.jpg (701x1024, 89K)

"intel marketing predicts 10ghz"
"Intel engineers see CPU consuming 1 megawatt at 10ghz"

Two things:

Firstly Intel put all of their eggs into a 10nm basket. And their design structure was dependent on the new process node. So we had years of refreshed products instead of anything new.

Which segways into point 2, intel kept products priced at the same segments despite chips getting smaller and decent yeilds. If you wanted 8 cores in 2016 you had to pay over a thousand dollara just for a cpu not to mention moving onto a more expensive HEDT platform.

Intels stagnation let AMD actually take a large portion of marketshare despite zen 1 being %15-20 behind skylake single core performance. If the 6700k was a hex core instead of four zen would have been another massive disappointment. And would have only been mildly relevant in hedt and server marketa due to the dramatic difference in cores per dollar.

It's not meant as a science law.

It's a guide for the industry.
Specifically: work on new chips designs start years before the processes needed to make those chips exist.
Ideally the design gets finalized just as the new process is ready.

>their design structure was dependent on the new process node.

And that new node required euv lithography, which got delayed and delayed.

Other manufacturers based their new nodes on good old immersion lithography instead.
Which works, but won't get cheaper over time like euv should.

Nothing went wrong. Physics blocked their path.

>7.5nm
>2024
HAHAHAHAHAHA

Death of competition is not a good thing. Intel being put back in their place is a good thing.

You do realize that if Intel would actually bite the dust, there would be at least a dozen new that pop up in its place and you'd have more competition than you ever even had in the 80's?

Intel would become an IP company worst case scenario, after all they still own half of the x86 ISA. What would more realistically happen is intel would sell off their fabs or spin them off into separate subsidiaries and just become a design house like amd is. This is all assuming intels 10nm falls through and their 7nm never sees the light of day.

They still are a massive company after all and still the biggest player in networking chips and technology, not to mention they still have the mobile market on lock.