they're dreaming, once intel is back in the circle THEY ARE DONE

lynx29If they can build a 3nm fab and it's already done, why build a 5nm one Arizona where construction just started... I don't care who the customer is, it doesn't make sense... lol

Cost and capacity. There are a lot of foundries that manufacture chips on bigger nodes. You don't do it unless you have a reason to because it costs money and time to go with the latest and greatest. A prime example are PCH chips and the IO chiplet on Ryzen CPUs. Bigger process than the cores, but that's okay.

By 2022 i'm still looking to buy a 5800X and RTX 3070.

okbuddythey're dreaming, once intel is back in the circle THEY ARE DONE

Dude, Intel was never in fab business.

TheDeeGeeBy 2022 i'm still looking to buy a 5800X and RTX 3070.

Sounds trolling.

lynx29I still don't get it. If you are a for-profit business, then you go balls to the walls as soon as your engineers give you the thumbs up, and you rake in the $$$$$ If 3nm is greenlit, and it apparently is, then you go full pedal to the medal on it on a new construction project. The world doesn't have enough natural resources left to keep playing milking games.

Well, the vast portion of the global semiconductors manufacturing is done on NON-bleeding-edge manufacturing process.
Most is 14nm, 22nm, 32nm, etc.

medi01Dude, Intel was never in fab business.

I think they are trying to get into it though:

www.eetimes.com/fab-emphasis-questioned-as-u-s-chip-lobby-sounds-off/#

Moore’s law lives on with one more transistor shrink :)
Can we get to 1nm node?

gmn 17Moore’s law lives on with one more transistor shrink :)
Can we get to 1nm node?

Basically the cost of designing leading node is doubling every two year. So we are seeing something like the inverse of Moore's Law. But yea, eventually we'll get there.

bencrutzactually TSMC still have older nodes occupied pretty handsomely, see the blue and grey portion for their 300mm fabs:



there's a lot of things that won't benefit much on bleeding edge node, like analog circuit, RF etc

Maybe you missed what I was saying, once an order is complete these numbers will diminish unless new orders are made. Back filling orders included.

gmn 17:)
Can we get to 1nm node?

Well, since those guys who built the pyramids have been on sub-sub-sub ym for the past 100 years or so, I would say YES, we can & then some ! hahahaha :)

gmn 17Moore’s law lives on with one more transistor shrink :)
Can we get to 1nm node?

1nm TSMC is a given.
I mean, marketing name.
7N transistor size is 22nm mind you.
Intels 14nm is 24nm.

lynx29If they can build a 3nm fab and it's already done, why build a 5nm one Arizona where construction just started... I don't care who the customer is, it doesn't make sense... lol

It's to get access to US government contracts as well as avoid some potential trade restrictions. The US has some strict rules on who can produce what electronics go into certain government sectors. IIRC, certain US military equipment have chips and the like produced on older but mature nodes strictly in US fabs. TSMC having a fairly modern node fab in America will help secure some of that juicy slice of the pie, as well as helping to shrink some of the less-vital components (the kinds that don't need to be as hardened as older nodes). Moreso for components that could use the efficiency gains to reduce power or improve performance, all in a smaller package.

In fact, TSMC's Arizona Fab is the only one on the leading edge (not bleeding edge; that's 3nm and smaller) in the US; Intel is the only other US foundry close enough with their 14+^14 nm, WIP 10nm, and planned 7nm nodes. GloFo's US fabs only go down to 14nm, after having given up on 7nm despite having more potential than TSMC's 7nm (mostly due to money and a lack of partner investment). The rest of the US fabs (TI, IXYS, Infineon, Medtronics, etc) are mostly for larger and older nodes, with Texas Instruments mostly dominating the US military market last I recall.

its official: TSMC does not "print chips", it prints money. :roll:

interesting, 3nm tech is great and many manyfactor competitions thouse.
this time cash took more who get thouse 1st but also numbers of them.

amd .nvidia sure be 1st of line...i think nvidia is this time little bit wiser...

ARFSounds trolling.



Well, the vast portion of the global semiconductors manufacturing is done on NON-bleeding-edge manufacturing process.
Most is 14nm, 22nm, 32nm, etc.

Most are not even that small. You don't need the expensive nodes for dishwashers, microwaves, cheap mp3 players and so on. Even the Raspberry Pi SoC, which is way way overkill for that kind of stuff, only recently moved to 28nm. Xilinx FPGAs were on 65nm, only this year moving some new models to 20nm. And that is a top tier FPGA vendor. I'd say like 90% of the market is 65nm and larger nodes.

lynx29If they can build a 3nm fab and it's already done, why build a 5nm one Arizona where construction just started... I don't care who the customer is, it doesn't make sense... lol

I am looking forward to 3nm products. I am also looking forward to desktop workstations that will use 5nm. It was said many times 5nm will be used for pc's for a few years longer than it will be used for mobile. Mobile for obvious reasons takes advantage of smaller node's, more than desktop parts. or I should say there is more need for smaller nodes in mobile than desktop. either way, the near future tech looks interesting.

lynx29If they can build a 3nm fab and it's already done, why build a 5nm one Arizona where construction just started... I don't care who the customer is, it doesn't make sense... lol

A factory is just an empty building until you put manufacturing tools in it. This isn't a "3nm factory". This is a factory which is planned to host 3nm manufacturing tools and produce wafers at that level of etching. Possibly even 450mm wafers instead of 300mm. Likewise for Arizona. It's 5nm till they decide to re-tool for 3nm. By the time Arizona comes online (planned for 2024), 5nm will be commoditised, so it'll still serve customers that require more mainstream fabrication (maybe Intel will fail to produce 7nm and 5nm and outsource it to TSMC, who knows?). There will still be plenty of demand for 5nm... It just won't carry the premium you'd want it to.

I'd be worried that the plant simply does not come to pass, since Trump is no longer in charge and I doubt Biden will pursue a competitive agenda against companies trading with China.

medi01Say exactly 0 reliable sources.

Pretty common knowledge, Apple had exclusity on all the new nodes for years at this point

SovsefandenPretty common knowledge

Pretty common knowledge is that DLSS 2 upscaling "produces images indistinguishable from native" which, nevertheless, is an obvious BS.

Apple produces large number of much smaller chips, than AMD.
Bar facts, even on expectations level, AMD seems to be much bigger a customer of TSMC than Apple is.

medi01Pretty common knowledge is that DLSS 2 upscaling "produces images indistinguishable from native" which, nevertheless, is an obvious BS.

Apple produces large number of much smaller chips, than AMD.
Bar facts, even on expectations level, AMD seems to be much bigger a customer of TSMC than Apple is.

DLSS 2.0 is magic, I know it's hard to accept as an AMD fanboy

And no, AMD is not TSMC's biggest customer, Apple is.

That's why Apple uses 5nm SoCs now, while AMDs Ryzen 5000 and Radeon 6000 products are still 7nm

Apple gets priority.

medi01Bar facts, even on expectations level, AMD seems to be much bigger a customer of TSMC than Apple is.

Are bar facts the ones you hear after a few beers?

SovsefandenDLSS 2.0 is magic

How some brains work, is magic.
It is the better TAA, works lovely with lines (grass, hair, eyebrows), but wipes out fine detail, adds blur.

That is why the most common image shared when hyping this good, but nowhere as good as hyped tech, is that female face, on which there is barely any texture to spot the loss of details, but there are eyebrows/eyelashes for line improvement to kick in.

Yay, magic



Only RT makes it more magical (RT is on on the... left, cool, ain't it? That blured floor, absolutely lovely!):

Caring1Are bar facts the ones you hear after a few beers?

You sure sound like someone who can bring in numbers and beat the poor opponent to the punch.
I'll wait for them to popup.

medi01How some brains work, is magic.
It is the better TAA, works lovely with lines (grass, hair, eyebrows), but wipes out fine detail, adds blur.

That is why the most common image shared when hyping this good, but nowhere as good as hyped tech, is that female face, on which there is barely any texture to spot the loss of details, but there are eyebrows/eyelashes for line improvement to kick in.

Yay, magic



Only RT makes it more magical (RT is on on the... left, cool, ain't it? That blured floor, absolutely lovely!):





You sure sound like someone who can bring in numbers and beat the poor opponent to the punch.
I'll wait for them to popup.

Yep magic, as long as it's v2.0 or higher - THATS WHY IT GETS PRAISED EVERYWHERE MAH DUDE.

www.nvidia.com/da-dk/geforce/news/death-stranding-nvidia-dlss-2-0/

Sliders for you

I have tried all DLSS 2.0 games on my 3080 and I know why AMD fanboys are in denial - they will have nothing like this anytime soon, sad bue true

DirectML will never bring what DLSS does, and Nvidia will support DirectML too

:laugh:

Nvidia is going to dominate so hard in Cyberpunk, you know, the biggest game release in years - Can't wait to max this game out on my 3080 at 1440p with Ray Tracing and DLSS 2.0 - AMD users can only dream

SovsefandenYep magic, as long as it's v2.0 or higher -

It's 2.0 quality mode, my friend.
And you can see images are blurred even on those sites that are hyping it.

Sovsefandenon my 3080

My condolesces. Did you undervolt it yet?

SovsefandenSliders for you

Oh, so kind of Jensen Huang to hide the uncomfortable pics, ain't it?
But I'll go to, let me generous, overclock3d review, hyping the hell out of it, oh, look at this, from the page 2 of the review:


Look what happens to the bush



When magic was applied: :D



Now, to be clear here:

It is expected from temporal anti-aliasing.
It does improve the lines (again, grass, eyebrows, hair) you could see it on the right of the same picture.

But the point is: it is NOWHERE close to what it is hyped to be. Not even freaking remotely.
It is just the best TAA derivative we have, not more, not less.

www.overclock3d.net/reviews/software/dlss_2_0_in_death_stranding_-_nvidia_4k_performance_trump_card/2

medi01It's 2.0 quality mode, my friend.
And you can see images are blurred even on those sites that are hyping it.


My condolesces. Did you undervolt it yet?


Oh, so kind of Jensen Huang to hide the uncomfortable pics, ain't it?
But I'll go to, let me generous, overclock3d review, hyping the hell out of it, oh, look at this, from the page 2 of the review:


Look what happens to the bush



When magic was applied: :D



Now, to be clear here:

It is expected from temporal anti-aliasing.
It does improve the lines (again, grass, eyebrows, hair) you could see it on the right of the same picture.

But the point is: it is NOWHERE close to what it is hyped to be. Not even freaking remotely.
It is just the best TAA derivative we have, not more, not less.

www.overclock3d.net/reviews/software/dlss_2_0_in_death_stranding_-_nvidia_4k_performance_trump_card/2

You are clueless, so funny when AMD fanboys are raging :D

I have tried DLSS 2.0 myself, because I can actually afford new and relevant hardware