Intel accidentally published some info on the Intel 4 process

Lakados

[H]F Junkie
Joined
Feb 3, 2014
Messages
10,361
1655097217886.png


1655097228035.png


1655097237447.png


It was apparently only up for a very short time before being taken down, but according to this guy who originally found it: https://twitter.com/witeken
It could feasibly see a 20% clock speed increase at the same power/heat over the Intel 7 process.
They also leaked a shot of a Meteor Lake Die
1655097550158.png

1655097562639.png

This would indicate that Intel has caught up to TSMC with their TSV and Interposer packaging.

Looking forward to seeing what team Blue can bring to the table in 2023.
 
But don't call them chiplets, they're tiles, okay?
 
But caught up to what, TSMC’s last gen?
The Intel 4 process is basically on par with the TSMC 3 process. With the delays at TSMC, Intel will likely have their 4 process online before TSMC’s 3 process. This will temporarily return the crown to Intel, and if their 18A process remains on schedule they will take it back again then.
 
Just because it may be smaller doesnt mean much if it is a badly designed process, just ask intel how their 10nm process worked out. They were ahead for a long time until that happened, and now they have to work hard just to keep up with a similar size process.
 
Just because it may be smaller doesnt mean much if it is a badly designed process, just ask intel how their 10nm process worked out. They were ahead for a long time until that happened, and now they have to work hard just to keep up with a similar size process.
10nm was too ambitious, they invented and patented many technologies there that they planned on implementing in 10nm but turns out is just mechanically impossible to do until you get down to 3 and smaller. It also resulted in them inventing a completely new alloy which is partially what makes their 18 processes possible. 10nm was too many engineers going "this would be so cool" and not enough people saying "yeah but should we!". The very definitions of inflated egos at work.
 
The Intel 4 process is basically on par with the TSMC 3 process. With the delays at TSMC, Intel will likely have their 4 process online before TSMC’s 3 process. This will temporarily return the crown to Intel, and if their 18A process remains on schedule they will take it back again then.
If Intel's 4 works out, then yeah. If it's a mess like 10/7, then the par-value doesn't matter, cause they've still got to refind their mojo. We'll see how things turn out in 2023, I guess.
 
If Intel's 4 works out, then yeah. If it's a mess like 10/7, then the par-value doesn't matter, cause they've still got to refind their mojo. We'll see how things turn out in 2023, I guess.
Everything indicates that it is working fine at this stage and is even ahead of schedule, I'm not worried there and neither is Intel. 10/7 was the outlier certainly not the norm.
 
The Intel 4 process is basically on par with the TSMC 3 process. With the delays at TSMC, Intel will likely have their 4 process online before TSMC’s 3 process. This will temporarily return the crown to Intel, and if their 18A process remains on schedule they will take it back again then.

I hear the opposite from people I know in the semiconductor industry. I worked in semi til 6 years ago, glad to be out.
 
Well, it's what TSMC told their shareholders, so if TSMC is saying that Intel is going to leapfrog them then I kind of believe it.

Leapfrogging doesn’t happen in semiconductor. You don’t fail 3rd grade and then skip to 5th. Ask Global Foundries how their leapfrog plan worked.
 
Leapfrogging doesn’t happen in semiconductor. You don’t fail 3rd grade and then skip to 5th. Ask Global Foundries how their leapfrog plan worked.
Intel didn’t fail anything, they just invented and designed a lot of stuff that didn’t work for 10nm but when used at smaller sizes works very well. They also had the money to work with IBM who is the company who’s actually designed their 20 and 18 processes. Leap frogging happens all the time, your analogy doesn’t work nor apply here.
 
Intel didn’t fail anything, they just invented and designed a lot of stuff that didn’t work for 10nm but when used at smaller sizes works very well. They also had the money to work with IBM who is the company who’s actually designed their 20 and 18 processes. Leap frogging happens all the time, your analogy doesn’t work nor apply here.

I’ve worked with IBM, Intel and TSMC - there’s no magic bullet in high volume semiconductor manufacturing, you have to execute in so many areas, not just design. TSMC excels at this, and things like defect reduction just get harder with each node. If you can’t control 7 nm sized particles how will you do better when you need to control 3nm particles? Design is exciting and Intel and IBM are better at it than TSMC, but that’s just one part of the puzzle. If you can’t manufacture this stuff with a usable yield, you’ll go the way of IBM and Global Foundries.

The marketing talk is nice but nothing has changed at Intel, they’re just as arrogant and stubborn as ever, so I’ll believe it when I see it.
 
Last edited:
I’ve worked with IBM, Intel and TSMC - there’s no magic bullet in high volume semiconductor manufacturing, you have to execute in so many areas, not just design. TSMC excels at this, and things like defect reduction just get harder with each node. If you can’t control 7 nm sized particles how will you do better when you need to control 3nm particles? Design is exciting and Intel and IBM are better at it than TSMC, but that’s just one part of the puzzle. If you can’t manufacture this stuff with a usable yield, you’ll go the way of IBM and Global Foundries.

The marketing talk is nice but nothing has changed at Intel, they’re just as arrogant and stubborn as ever, so I’ll believe it when I see it.
Better start believing then. You are a fool if you think any single company can stay in the lead forever.
 
Better start believing then. You are a fool if you think any single company can stay in the lead forever.

Have you been inside a TSMC fab? I have, and it’s not something another company can just replicate, especially here in the US. All this talk of new fabs in the US to compete with TSMC is a joke. In a way I hope they fail cause the work life is inhumane.
 
Better start believing then. You are a fool if you think any single company can stay in the lead forever.
Yet people do this for Intel all the time. To the point that people think it's weird just to say "You know what? I'll reserve opinions until after the actual product is out there". Imagine that being controversial lol.
 
Back
Top