7nm Process - Game changer?

[Troezar]

I was just reading this article about 7nm and EUV. Thought it might be of interest to the more nerdy of you

https://www.pcgamesn.com/intel-amd-7nm-cpu-euv

Initial costs could be high but the use of 13.5nm light rather than 193nm could have a lot of benefits. I didn't realise that light below a wavelength of 193nm is absorbed by the air.

 

[Cold_And_Miserable]

EUV doesn't seem remotely viable or practical requiring "bits of tin" to generate the 13.5nm light. All the equipment has to operate in vacuum which sounds like an utter nightmare to avoid contamination and the light is near impossible to manipulate with lenses or mirrors. I doubt they'll ever get it to work.

 

[Troezar]

EUV doesn't seem remotely viable or practical requiring "bits of tin" to generate the 13.5nm light. All the equipment has to operate in vacuum which sounds like an utter nightmare to avoid contamination and the light is near impossible to manipulate with lenses or mirrors. I doubt they'll ever get it to work.

IBM seem pretty optimistic about it and are already looking at its use below 7nm. Sounds like there's been a lot of effort put into it over the last 10 years. Quite remarkable the about of time and money it takes to produce a CPU. Something I think most of us take for granted.

 

[Rroff]

I think 7nm with any kind of advanced lithography is going to be expensive for a very long time.

It is potentially a game changer though for a lot of things especially in combination with some other advances it will make it possible to move away from monolithic cores in a way that isn't really possible currently for certain applications.

 

[Troezar]

I think 7nm with any kind of advanced lithography is going to be expensive for a very long time.

It is potentially a game changer though for a lot of things especially in combination with some other advances it will make it possible to move away from monolithic cores in a way that isn't really possible currently for certain applications.

Is that because it is harder to produce small cores with the longer wavelengths?

 

[Meaker]

Is that because it is harder to produce small cores with the longer wavelengths?

It affects the resolution of the individual pathways and details. The smaller the wavelength the smaller transistor you can make. With current wavelengths it's like etching a delicate pattern into wood with a butter knife. This wants to go back to using a scalpel.

 

[Minusorange]

I was just reading this article about 7nm and EUV. Thought it might be of interest to the more nerdy of you

https://www.pcgamesn.com/intel-amd-7nm-cpu-euv

Initial costs could be high but the use of 13.5nm light rather than 193nm could have a lot of benefits. I didn't realise that light below a wavelength of 193nm is absorbed by the air.

Surely Graphene transistors is more of a game changer than making them smaller in current form ?

https://www.sciencedaily.com/releases/2017/06/170613145144.htm

CPU's capable of terahertz instead of gigahertz is truly game changing, I think graphene will throw a spanner into Moores law

 

[Jumper118]

7nm will not change any games. The devs are still lazy slackers who just throw half baked rubbish up on early access so they don't have to pay proper beta testers

 

[Troezar]

7nm will not change any games. The devs are still lazy slackers who just throw half baked rubbish up on early access so they don't have to pay proper beta testers

Perhaps a tad harsh. I see large corporations dictating what is produced with unrealistic deadlines and profits in mind. The independents are left with the scraps or crowd sourcing. Definitely developing the hard way.

This attitude permeates society in the west at the moment. It would be nice to see some pride in creating a good product without trying to screw every last penny out of the customers.

 

[Doobedoo]

https://www.overclock3d.net/news/cpu_mainboard/amd_details_future_7nm_cpu_and_gpu_designs/1

 

[itchy]

They wont get past 1nm and even at 1nm its very difficult.

https://arstechnica.com/science/2016/10/nanotubes-atomically-thin-material-smallest-transistor-ever/

Lucky if we see 5nm silicon.

 

[4K8KW10]

7nm will not change any games. The devs are still lazy slackers who just throw half baked rubbish up on early access so they don't have to pay proper beta testers

7nm will change the monopoly in the CPU business. Intel will be doomed, AMD will thrive

 

[ToOo]

Lucky if we see 5nm silicon.

This... there is a significant chance that 7nm will be the last high power silicon node we see for commercial parts. While 5nm and 3nm are theoretically possible the questions of manufacturability (cost of fabs now seems to double for every node shrink) and performance benefit (leakage is serious problem at this geometry) leaves a question mark over the future of silicon because it might make more sense at that point to transition over to germanium or graphene.

 

[Troezar]

IBM, GF and Samsung already have a 5nm GAAFET chip created using EUV so we seem to have a little more mileage left in silicon. 50% more transistors on a given area than 7nm.

After that though who knows...

 

[d_brennen]

Dr Lisa Su has been confident of the viability of 7nm. AMD pretty much need it to work out of the gate to stay competitive.

 

[d_brennen]

After that though who knows...

Glue!

 

[itchy]

https://semiengineering.com/10nm-versus-7nm/

Food for thought.

 

[Troezar]

https://semiengineering.com/10nm-versus-7nm/

Food for thought.

Whatever happens there are going to have to be some major changes in the way they manufacturer chips in the next 10 years. Could be very interesting. I can see PCs lasting longer and longer though as the rate of improvement slows and the price creeps up. It's not like there are really any killer apps that are forcing people to upgrade their old tech.

 

[David Bisset]

Surely Graphene transistors is more of a game changer than making them smaller in current form ?

https://www.sciencedaily.com/releases/2017/06/170613145144.htm

CPU's capable of terahertz instead of gigahertz is truly game changing, I think graphene will throw a spanner into Moores law

How far does your signal propagate during a clock cycle at that kind of frequency? Not far at all! Fractions of current die sizes. You will have some interesting challenges keeping bits synchronised.

 

[Rroff]

How far does your signal propagate during a clock cycle at that kind of frequency?

That is quite an interesting consideration.