Discussion Leading Edge Foundry Node advances (TSMC, Samsung Foundry, Intel)

Page 38 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
TSMC's N7 EUV is now in its second year of production and N5 is contributing to revenue for TSMC this quarter. N3 is scheduled for 2022 and I believe they have a good chance to reach that target.

1587737990547.png
N7 performance is more or less understood.
1587739093721.png

This year and next year TSMC is mainly increasing capacity to meet demands.

For Samsung the nodes are basically the same from 7LPP to 4 LPE, they just add incremental scaling boosters while the bulk of the tech is the same.

Samsung is already shipping 7LPP and will ship 6LPP in H2. Hopefully they fix any issues if at all.
They have two more intermediate nodes in between before going to 3GAE, most likely 5LPE will ship next year but for 4LPE it will probably be back to back with 3GAA since 3GAA is a parallel development with 7LPP enhancements.


1587739615344.png

Samsung's 3GAA will go for HVM in 2022 most likely, similar timeframe to TSMC's N3.
There are major differences in how the transistor will be fabricated due to the GAA but density for sure Samsung will be behind N3.
But there might be advantages for Samsung with regards to power and performance, so it may be better suited for some applications.
But for now we don't know how much of this is true and we can only rely on the marketing material.

This year there should be a lot more available wafers due to lack of demand from Smartphone vendors and increased capacity from TSMC and Samsung.
Lots of SoCs which dont need to be top end will be fabbed with N7 or 7LPP/6LPP instead of N5, so there will be lots of wafers around.

Most of the current 7nm designs are far from the advertized density from TSMC and Samsung. There is still potential for density increase compared to currently shipping products.
N5 is going to be the leading foundry node for the next couple of years.

For a lot of fabless companies out there, the processes and capacity available are quite good.
 

FlameTail

Golden Member
Dec 15, 2021
1,875
982
106
Hey guys, I am very interested in Samsung Foundry.

Any updates on Samsung 4nm (4LPX, 4LPE)?

There were reports at the start of the year that Samsung 4nm was having yields as low as 20%.

I guess this has improved by now ?
 

FlameTail

Golden Member
Dec 15, 2021
1,875
982
106
Ok, so I found these two images. Golden Nuggets!
 

Attachments

  • chrome_screenshot_1667296584832.png
    chrome_screenshot_1667296584832.png
    647.8 KB · Views: 56
  • chrome_screenshot_1667296742755.png
    chrome_screenshot_1667296742755.png
    685.6 KB · Views: 57
  • Like
Reactions: Vattila

FlameTail

Golden Member
Dec 15, 2021
1,875
982
106
At the same website they talk about the yields of leading edge nodes of Samsung. And the "reports" of 20-30% yields were false. And now are much more improved.

Source ?

Exact figures aside, it isn't hard to believe they were having poor parametric yields, because evidence being that the chips that came out of the fab underperformed. Obviously, yields improve over time, so I am not sure what you are pointing to.
 

biostud

Lifer
Feb 27, 2003
18,193
4,674
136
Anyone looking for a new job?

Intel Foundry Services Chief Steps Down | Tom's Hardware (tomshardware.com)

a6bab19e357a027706c33bc71578c431.jpg
 

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
It seems Tower Semiconductor will be taking charge of IFS.
China might disagree. If China disagrees then Intel would have to think hard before risking 30% revenue of their total revenue from there.
Long term prospect for American companies in China is quite uncertain. EU/JP/SK/TW will benefit.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
They are going to spin off their foundry services?

But why? Isn't having your own foundry an advantage?
They've publicly pushed back hard at speculation that they will spin off IFS, but that doesn't stop the rumors. Bear in mind it's just people observing Intel's actions under a lens of "if they were to spin off the fabs, this is something they'd do". But that not-so-coincidentally has a lot of overlap with running IFS in general.
 

FlameTail

Golden Member
Dec 15, 2021
1,875
982
106
Nvidia 5000 series on 3nm gaa .. looks promising

As long as their 3nm doesn't underperform and become a flop.

I feel like 3nm is Samsung's last chance to fight back against TSMC and stay competitive. If their 3nm flops, it will be end of an era for Samsung Foundry as a cutting edge fab.

I am basing this on the fact that ASML CEO once stated -as I remember- that eventually one player will have to drop out of the leading edge foundry race.

Edit: As I said, this is my opinion personally, and I might be terribly wrong on this count. If you disagree, tell me what you think🤔
 
Last edited:

Henry swagger

Senior member
Feb 9, 2022
356
235
86
As long as their 3nm doesn't underperform and become a flop.

I feel like 3nm is Samsung's last chance to fight back against TSMC and stay competitive. If their 3nm flops, it will be end of an era for Samsung Foundry as a cutting edge fab.

I am basing this on the fact that ASML CEO once stated -as I remember- that eventually one player will have to drop out of the leading edge foundry race.
Thats bs
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
They are going to spin off their foundry services?

But why? Isn't having your own foundry an advantage?

The more TSMC wafers they take, the more their own fabs will become a lodestone about their neck. 10nm was a disaster for the company in numerous ways, despite the fact that they eventually salvaged some working products from that fiasco. If Intel 4 (and Intel 3) turns out to be anywhere near as bad, it could take the entire company with it. Isolating their fabs from every other part of the business gives them a chance to shift losses and preserve some part of the company.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
N3/E F12P8 is almost done
1669234740782.png

N3/E F18P4/5/6 already done. F18P7 progressing well.
1669234902300.png

N3/E F18P8 seems not much progress. Instead F14P8 progressing much better.

1669235164924.png



All in all for N3 (excluding F18P4 which conducted risk N3 production)
  • F18P5/6 structurally completed (already committed to HVM 2H22).
  • F18P7 structurally to be completed.
  • F18P8 started but seems to be halted or at least slow progression
  • F12P8 structurally almost completed (Rumored to handle Intel only wafers)
  • F12P9 is halted (Rumored to handle Intel only wafers)

Interesting notes
  • Contrary to earlier rumors, F12P8 seems to be a full sized Fab similar to the F18 series instead of the previous rumors of being sized around F14 series. F18 series are designed to handle 30-40K wpm.
  • F12P9 indeed halted, suggesting Intel delayed or reduced the planned intake of N3 wafers or industrialization deferred.
By end of 2H22, TSMC should have 60-80K wpm N3 capacity from F18P5/6. Next year, another 60-80K wpm from F12P9 and F18P7. 2024 probably F18P8 will come online but so far F12P9 seems not started at all. 200K WPM by 2024 at the very least.

BTW, incredible speed of the construction of F12P8.

Will check back in 6 months on F18P8 and F12P9.

*Google Maps take many quarters to update satellite imagery for obvious reasons. Live data is actually costly.
 

FlameTail

Golden Member
Dec 15, 2021
1,875
982
106

Looks like Samsung going full steam ahead with 3nm. Still, the yields situation is concerning. I hope they can fix it.
 

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
"Second generation" 3nm node without a working first generation node sounds, to me, like they are brewing up a version of their 3nm node that removes the most troublesome part and adds a few tweaks learned trying to get vanilla 3nm going as a fall back in case they can't get 3nm to acceptable yields.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
"Second generation" 3nm node without a working first generation node sounds, to me, like they are brewing up a version of their 3nm node that removes the most troublesome part and adds a few tweaks learned trying to get vanilla 3nm going as a fall back in case they can't get 3nm to acceptable yields.


The big unknown is what is causing the low yields. Is it related to the GAA transistor stack (in which case "tweaking" may not be enough to make much of a difference) or is it something where relaxing the design rules a bit will help (which it seems is what TSMC did with N3E)

This may be why TSMC is taking a more measured approach with N2, with the first GAA node offering only a minimal shrink from the N3 family. Trying to turn too many knobs too far and at once caused much of Intel's 10nm woes, and may be what is biting Samsung with 3GAE. TSMC's N3E may be a less dire example of the same thing, hence the stepwise N2 rollout.
 
  • Like
Reactions: Ajay

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Not sure why anyone is surprised by 3GAE. There were rumours Samsung would keep that node internal, making 3GAP the first GAAFet node available for wafer orders.
 
  • Like
Reactions: Thibsie

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
The internal node was the 4nm one.
backport.png

When it was called this in the 2017 roadmap:
"4LPP (4nm Low Power Plus): 4LPP will be the first implementation of next generation device architecture – MBCFET structure (Multi Bridge Channel FET). MBCFET is Samsung’s unique GAAFET (Gate All Around FET) technology that uses a Nanosheet device to overcome the physical scaling and performance limitations of the FinFET architecture."
- https://news.samsung.com/global/sam...ith-comprehensive-process-roadmap-down-to-4nm
Based on (ARM Logic+Memory) compiler data:
7nm LPP/SF7: 1.0x perf + 1.0x power + 1.0x area
5nm LPE/SF5E: 1.11x perf + 0.85x power + 0.71x area
4nm LPE/SF4E: 1.2321x perf + 0.714x power + 0.5396x area
3nm GAE/SF3E: 1.416915x perf + 0.5355x power + 0.4047x area
[Same core, Same memory :: Real core -> Real PPA]

The bad yield rumors aren't coming from Korea but Taiwan apparently. It is really silly since Samsung announced the 11? future fabs in Texas, specifically for 3GAE and lower.
[11 new chipmaking fabs that would create 10,000 jobs and account for an investment of about $192 billion.] <-- multiple sources, with different calculated values.
 
Last edited:
  • Like
Reactions: Henry swagger