Monday, June 13th 2022

Intel 4 Process Node Detailed, Doubling Density with 20% Higher Performance

Intel's semiconductors nodes have been quite controversial with the arrival of the 10 nm design. Years in the making, the node got delayed multiple times, and only recently did the general public get the first 10 nm chips. Today, at IEEE's annual VLSI Symposium, we get more details about Intel's upcoming nodes, called Intel 4. Previously referred to as a 7 nm process, Intel 4 is the company's first node to use EUV lithography accompanied by various technologies. The first thing when a new process node is discussed is density. Compared to Intel 7, Intel 4 will double the transistor count for the same area and enable 20% higher performing transistors.

Looking at individual transistor size, the new Intel 4 node represents a very tiny piece of silicon that is even smaller than its predecessor. With a Fin Pitch of 30 nm, Contact Gate Poly Pitch of 50 nm between gates, and Minimum Metal Pitch (M0) of 50 nm, the Intel 4 transistor is significantly smaller compared to the Intel 7 cell, listed in the table below. For scaling, Intel 4 provides double the number of transistors in the same area compared to Intel 7. However, this reasoning is applied only to logic. For SRAM, the new PDK provides 0.77 area reduction, meaning that the same SoC built on Intel 7 will not be half the size of Intel 4, as SRAM plays a significant role in chip design. The Intel 7 HP library can put 80 million transistors on a square millimeter, while Intel 4 HP is capable of 160 million transistors per square millimeter.
Regarding individual transistor layers, Intel had some fun with them as the company now put 16 layers of metal for logic in the Intel 4 node, compared to 15 in the Intel 7 PDK. The gate is made from Tungsten, while the first four metal layers are made from Copper with Cobalt Cladding, ensuring better performance than plain Copper. For the remaining layers, pure Copper is used for interconnecting the wiring.

In the manufacturing process of Intel 4, the company is using EUV for the first time and achieving great results. The number of steps to manufacture the chip is now reduced, and it is reported that the number of masks for etching is reduced by 20% compared to Intel 7, thanks to more powerful EUV lithography tools. If not for EUV, the number of masks for this type of transistor would need to jump by 30% as the multi-patterning process is very complicated.

And last but not least comes the performance of these new transistors. At iso-power of 0.65 Volts, Intel reports a 21.5% gain in sustainable frequency with Intel 4 compared to Intel 7. At higher power like 0.84 Volts, the curve flattens to about 10% improvement. For iso-frequency tests, the company reports 40% lower power consumption measured at 2.1 GHz transistor switching speed.
First products based on the new node are set to arrive sometimes in 2023, with Meteor Lake-P designs hitting the shelves first. Other SKUs are expected to follow and we are eager to see how good this semiconductor manufacturing node turns out to be.
Add your own comment

15 Comments on Intel 4 Process Node Detailed, Doubling Density with 20% Higher Performance

#2
ppn
Whats the point. I expect 16 angstrom (Å), already. nothing less than, including the SOC, the GPU and the PCH tiles shoud be on the sub 2 Å nodes.

And why is the Soc so big, usually the memory controller is as big as a dual core.
Posted on Reply
#3
Vayra86
ppnWhats the point. I expect 16 angstrom (Å), already. nothing less than, including the SOC, the GPU and the PCH tiles shoud be on the sub 2 Å nodes.

And why is the Soc so big, usually the memory controller is as big as a dual core.
Whats the point of a shrink with unparalleled efficiency?

This is the shrink we have all been waiting for.., finally moving away from DUV. Its the whole reason TSMC is ahead of the curve right now.
Posted on Reply
#4
Nanochip
ppnWhats the point. I expect 16 angstrom (Å), already. nothing less than, including the SOC, the GPU and the PCH tiles shoud be on the sub 2 Å nodes.

And why is the Soc so big, usually the memory controller is as big as a dual core.
The formidable Taiwan competitor (and partner since Arc is built by TSMC) hasn't even volume ramped 3nm, their 2nm is delayed, possibly out to late 2024 or even 2025, and you expect 1.6 nm? Intel has finally introduced EUV and a node that was very delayed and many thought won't ever happen is about to volume ramp. This is good progress.
Posted on Reply
#5
ncrs
NanochipThe formidable Taiwan competitor (and partner since Arc is built by TSMC) hasn't even volume ramped 3nm, their 2nm is delayed, possibly out to late 2024 or even 2025, and you expect 1.6 nm? Intel has finally introduced EUV and a node that was very delayed and many thought won't ever happen is about to volume ramp. This is good progress.
Since this is a PR slide show I'll believe Intel when I see the chips using it on shelves.
Their 10nm process also had slides like those, and it still isn't going well as indicated by their inability to make huge Xeon dies in high volumes: www.techpowerup.com/295694/intel-xeon-sapphire-rapids-volume-shipping-delayed-again-company
It's also the second time Sapphire Rapids got delayed.
Don't get me wrong, I love competition, and wish Intel well, but they have a history of spamming PR campaigns to mask failures.
Posted on Reply
#6
bonehead123
YADA, YADA, YADA........

All that techno-babble, yet notta nuthin new or betta available yet to install into my rig, so therefore, it just moar V*A*P*O*R*warez to me....

@AleksandarK... was the above an actual press release, some copy/paste of speculations from some unknown insider, or just a compilation of random info from across the interwebz ? just wondering :)
Posted on Reply
#7
Nanochip
ncrsSince this is a PR slide show I'll believe Intel when I see the chips using it on shelves.
Their 10nm process also had slides like those, and it still isn't going well as indicated by their inability to make huge Xeon dies in high volumes: www.techpowerup.com/295694/intel-xeon-sapphire-rapids-volume-shipping-delayed-again-company
It's also the second time Sapphire Rapids got delayed.
Don't get me wrong, I love competition, and wish Intel well, but they have a history of spamming PR campaigns to mask failures.
That's a fair point and we did get 14nm++++++ for years. So a dose of skepticism is certainly healthy. But when 10nm was delayed, where were the papers at the IEEE symposium detailing the 10 nm process? This time around feels different. They seem to have a bit more confidence this time around. They haven't even released Raptor Lake and are already seemingly excited about Meteor Lake and Arrow Lake. That certainly wasn't the case in the 14nm++++ days. They weren't as excited by Comet Lake or Rocket Lake. Hell Rocket Lake was junked in 6 months. We'll see if Meteor Lake launches on time next year or if it's delayed...

But one thing to keep in mind is the move from 14nm->10 nm was an extremely ambitious step where Intel was trying to accomplish 2.7x density scaling from the node shrink, without using EUV, and instead it used multi patterning. EUV was not ready yet... Now is different. EUV is ready as shown off by TSMC , and with 10nm/Intel7 -> Intel 4 ... intel is using EUV and has a lower density scaling target of only 2x. And apparently, EUV allows them to simplify the lithography process from 4 or more steps to one step.

So the factors and considerations of the node shrink this time around are different from 14->10.
Posted on Reply
#8
Wirko
bonehead123YADA, YADA, YADA........

All that techno-babble, yet notta nuthin new or betta available yet to install into my rig, so therefore, it just moar V*A*P*O*R*warez to me....
Why can't we have year X+2 technology in the year X, for cheap, and while we're at that, it should be self-cooling too?
Posted on Reply
#9
Nanochip
bonehead123YADA, YADA, YADA........

All that techno-babble, yet notta nuthin new or betta available yet to install into my rig, so therefore, it just moar V*A*P*O*R*warez to me....

@AleksandarK... was the above an actual press release, some copy/paste of speculations from some unknown insider, or just a compilation of random info from across the interwebz ? just wondering :)
What are you using in your Rig? Do you also feel that Zen 4 (which is also not yet available) is also Vaporware?
Posted on Reply
#10
AleksandarK
News Editor
bonehead123@AleksandarK... was the above an actual press release, some copy/paste of speculations from some unknown insider, or just a compilation of random info from across the interwebz ? just wondering :)
...Today, at IEEE's annual VLSI Symposium...
Official Intel presentation.
Posted on Reply
#11
ModEl4
Very good news overall but i expected more regarding SRAM scaling, TSMC went from 7nm to 5nm with 1.8X more logic density and 1.35X SRAM density advantage.
Intel achieved going from Intel7 to Intel4, 2X logic density and 1.3X SRAM density.
They should focus more regarding SRAM density scaling, they will need it in the future!
Posted on Reply
#13
bonehead123
NanochipWhat are you using in your Rig? Do you also feel that Zen 4 (which is also not yet available) is also Vaporware?
Doesn't matter what I am using now (which is working fine for me atm) what matters to me is what I can/will be able buy & install in my next machine.... AND WHEN.... and YES, until I can go out/online & buy something that was announced or shown in some slide show or press release at a trade show, it remains Vaporware to me....

But if you really must know, my system specs are shown in my profile :)
Posted on Reply
#14
Minus Infinity
Intel loves them powepoint slides. Too bad products are already slipping including Raptor Lake and Sapphire Rapids and only expect Mobile Meteor Lake for a long time next next.
Posted on Reply
#15
AdmiralThrawn
Dear god the ammount of heat that thing will psuh.
Posted on Reply
Add your own comment
May 21st, 2024 17:30 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts