Discussion Leading Edge Foundry Node advances (TSMC, Samsung Foundry, Intel) - [2020 - 2025]

[DisEnchantment]

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.


N7 performance is more or less understood.


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

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

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.

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

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.

---------------------------------------------------------------------------------------------------------------------------------------------------


FEEL FREE TO CREATE A NEW THREAD FOR 2025+ OUTLOOK, I WILL LINK IT HERE

 

[johnsonwax]

This is the dumbest argument. Someone pined for a big GPU, was pointed out that Apple has reasonably large GPUs, then ginned up an argument about unit share to save face, and everyone followed him down that hole.

 

[511]

This happens many times lol

 

[Josh128]

Apple watches are fraction of iPhones.

iPhone is about 220 million per year. Apple has sold 22-25 million mac units. Now the iPhone data is 2022 and mac units are 2025 data. But it is roughly 1 out of 10. Even if you only include the M3 Max, it's not going to be anywhere near 100x, nevermind 1000x and that's a very contrived way of calculating things.

Bulls**t. Show the numbers. Theres no way I'll believe that of the approximately 50% of the smartphone owners on the planet, more than 1 out of every 20 or 30 have an M3 Max system, unless you provide some kind of proof. I believe its significantly less than 1 out of every 100, closer to or even exceeding 1 out of 1000.

Also, so you are ignoring ipads, disregarding watches, and saying the majority of those Mac units are the Pro and Max SOCs? Do you know why I made the claim in the first place?

 

[Josh128]

This is the dumbest argument. Someone pined for a big GPU, was pointed out that Apple has reasonably large GPUs, then ginned up an argument about unit share to save face, and everyone followed him down that hole.

Now you are just being facetiously ignorant to what I posted. I didnt "pine" for a big GPU, thats not even remotely close to being true. Look what started it all, its right there for everyone to read. The entire argument started about die size, yield and cost. I didnt gin up anything. The 1000x claim was hyperbole, but the intent was to show that Apple sells more silicon, at higher yields, than anyone else due to the fact the vast majority of the SOCs it produces are small.

 

[fastandfurious6]

Interesting that Fmax increase is only mentioned for N2X and not at all for N2P. Wouldnt it be something if all the 6.5GHz hoopla caused by MLID is for a product coming in 2027 and Zen 6 doesnt move the bar much at all and ends up at ~5.8GHz.

does fmax really matter anymore?

desktops: get gear for OC

mobile: constrained by thermals anyway

?????????

 

[511]

does fmax really matter anymore?

Most of Apple's gain since M1 is Fmax mostly they have gained approximately 40% from Fmax alone 3.2 -> 4.45 Ghz

 

[branch_suggestion]

does fmax really matter anymore?

Qualcomm is quite literally about to drop the first 5Ghz AA64 core ever.
You sure bet it does.

 

[MoistOintment]

Intel allegedly increases their High-NA orders

https://twitter.com/x/status/1970472523612508563

 

[adroc_thurston]

Interesting that Fmax increase is only mentioned for N2X and not at all for N2P. Wouldnt it be something if all the 6.5GHz hoopla caused by MLID is for a product coming in 2027 and Zen 6 doesnt move the bar much at all and ends up at ~5.8GHz.

man you're really desperate for it to suck.
almost adorable.

alas, fmax goes up. that's the rule.

 

[Josh128]

man you're really desperate for it to suck.
almost adorable.

alas, fmax goes up. that's the rule.

Oh noes...Im up around 100% on my AMD holdings and I'd rather it be 200% or more. I'm just over the whole hype train thing, it always reels you in, then destroys your dreams.

Zen 2: 16C, 5GHz 🤡
Zen 3: Almost no crazy rumors, yet still awesome. This was peak times.
Zen 4: +25%-30% IPC 🤡🤡
Zen 5: +40% IPC 🤡🤡🤡
Zen 6: >6.5GHz ❓❓❓

 

[adroc_thurston]

Zen 2: 16C, 5GHz 🤡

They got to 4.7 which was close nuff.

Zen 3: Almost no crazy rumors, yet still awesome. This was peak times.

That's because the FUD for Z3 was it being low-mid single digit IPC.

Zen 4: +25%-30% IPC 🤡🤡

no?

Zen 5: +40% IPC 🤡🤡🤡

Turns out it's mostly clocks. A lot of clocks.

Zen 6: >6.5GHz ❓❓❓

Yeah?
That's what you get from a double-wham node shrink.

Extra nice for mobile since it gets a proper CCD now, should be the first real freq bump in mobile since CZN-HX.

 

[Joe NYC]

BTW, how many memory channels is that?

It seems like 3 external and 1 (optional) internal

 

[adroc_thurston]

BTW, how many memory channels is that?

what. you mean Glymur?

It seems like 3 external and 1 (optional) internal

the what

 

[Darkmont]

BTW, how many memory channels is that?

It seems like 3 external and 1 (optional) internal

It’s 12 16-bit LPDDR5X channels forming a 192b bus

 

[Joe NYC]

what. you mean Glymur?

the what

Looks like 2 external and 1 internal, running at fast speeds.

 

[adroc_thurston]

Looks like 2 external and 1 internal, running at fast speeds.

the what?
It's a 192b part, but 192b (the whole membus) is enabled only for a single MoP SKU.

 

[Joe NYC]

It’s 12 16-bit LPDDR5X channels forming a 192b bus

1 channel is internal, Memory on Package of 48 GB

 

[Joe NYC]

the what?
It's a 192b part, but 192b (the whole membus) is enabled only for a single MoP SKU.

"Channel" in PC is 64 bits. Why is this news to you?

 

[adroc_thurston]

"Channel" in PC is 64 bits.

Channel for LPDDR is 16b exactly.

 

[Joe NYC]

Sorry about derailing this thread, meant to post it in Qualcomm thread.

 

[Geddagod]

no?

Who was this source I wonder lol

Turns out it's mostly clocks. A lot of clocks.

Fmax stayed the same T-T

 

[adroc_thurston]

Fmax stayed the same T-T

Not in server.

 

[Darkmont]

"Channel" in PC is 64 bits. Why is this news to you?

Channels in LPDDR are 16, arranged in a 32/64/128 bit module. They're not even 64 bits in Desktop anymore.

 

[Josh128]

They got to 4.7 which was close nuff.

4.7GHz wasnt really real. All Zen 2 owners know this. It couldnt sustain more than ~4.5 ST in Cinebench. Zen 2 actual max ST clock under any kind of sustained load whatsoever was 4.375 - 4.475 for all the SKUs. You could sometimes see 4.6 and 4.7 registered in HWinfo if you wiggled the mouse or dragged a window around the desktop.

no?

Most definitely yes. MLID and RedGamingTech pushed the hell out of the +25% IPC for Zen 4 narrative until it was clear that was FUD. This is from HotHardware.

 

[adroc_thurston]

4.7GHz wasnt really real

Oh it was.
Later bins did it just fine.

MLID and RedGamingTech

You could've not picked worse ebeggars.
Try harder.