TSMC's 7nm+ EUV is in production, improves performance by 10%

[mongeese]

Something to look forward to: At TSMC’s recent technology forum in Hsinchu, Taiwan, the company's CEO confirmed new details regarding N7+, TSMC’s refined 7nm node employing EUV (extreme ultraviolet) lithography. Most importantly, N7+ chips are already being produced and 7nm production will ramp up by 150% this year to one million total units.

Despite how young 7nm technologies are, having yet to appear in a single laptop or desktop, they are already superseded by 7nm+.

N7+ has identical yield rates to N7 and will steadily improve, while also offering a 20% increase to transistor density. There’s also a 10% performance uplift or 15% power efficiency increase. AMD will take advantage of the former in their fourth-gen Ryzen which they’ve confirmed to use TSMC’s 7nm+, while Huawei will most likely take advantage of the latter in their mobile SoC flagships due out in devices at the end of the year.

While this is all pretty good news for the consumer, it also means that Intel’s got some challenges ahead. They’ve long touted that their 10nm process is equivalent to TSMC’s 7nm, but what about 7nm+? And while some 10nm products will appear in laptops later this year, key market segments, like desktops, high-performance laptops, and server parts won’t receive 10nm hardware for up to three years.

As a bonus, TSMC’s 5nm EUV technology is coming along swimmingly. They’re already bringing equipment into Fab 18, their first 5nm facility, which will begin production in the first quarter of 2020. The land next door is reserved for a new 3nm facility, which is on track to begin outputting hardware in 2022. Things couldn’t be much better for TSMC and their partner AMD right now.

Image Credit: Laura Ockel on Unsplash

Permalink to story.

https://www.techspot.com/news/80237-tsmc-7nm-production-improves-performance-10.html

 

[quadibloc]

I'm very glad to hear that TSMC has gotten EUV working. Presumably Intel will get to this point pretty soon as well. If it is only six months behind, this will not be too big a deal.

But AMD's timing is not that good. Given that the Ryzen 3000 series is expected to correct the deficiency of not having as much floating-point muscle as Intel chips, it would have been better if they had brought them out sooner, in 14 nm... and then they could have waited a little, and gone to this node as soon as it was available. Instead, it looks like they'll only get to 7nm+ when it's old news.

One can bet that Intel won't make that mistake with its process nodes.

 

[Vulcanproject]

I'm very glad to hear that TSMC has gotten EUV working. Presumably Intel will get to this point pretty soon as well. If it is only six months behind, this will not be too big a deal.

But AMD's timing is not that good. Given that the Ryzen 3000 series is expected to correct the deficiency of not having as much floating-point muscle as Intel chips, it would have been better if they had brought them out sooner, in 14 nm... and then they could have waited a little, and gone to this node as soon as it was available. Instead, it looks like they'll only get to 7nm+ when it's old news.

One can bet that Intel won't make that mistake with its process nodes.

I think you're a little out of the loop. Intel were at least two years ahead on process as late as 2015. Skylake was on tri-gate 14nm in 2015 while AMD were still building garbage on 28nm in 2016. Now in 2019, a mere four years later, Intel are suddenly 6 months behind when Zen 2 launches.

AMD's timing couldn't be better. There isn't a bad time to be leading. They handed everything they have got to TSMC who are streaking ahead. First to 7nm, then by the time Intel have 10nm working TSMC will deliver 7nm+ according to these reports.

Intel have to finally fix 10nm for full production by the end of this year, and they better be on track for their TSMC 5nm equivalent in 2021 otherwise expect AMD to take away huge swathes of Intel market share in the next 2-3 years.

Expect it anyway for the end of this year and 2020 if Zen2 is as good as leaks suggest thus far.

I have been all Intel for over a decade now. Saying that the security problems are even starting to scare me off Intel. I was also one who dismissed the likes of Spectre last year because they seemed minor. It is proven they were- relative to the latest much more serious issues now appearing.

They have a lot of hurt coming, let's hope they get a grip and start back to thinking about the consumer with a little humble pie.

 

[pencea]

AMD's timing is not that good.s.

Huh?

I'm very glad to hear that TSMC has gotten EUV working. Presumably Intel will get to this point pretty soon as well. If it is only six months behind, this will not be too big a deal.

But AMD's timing is not that good. Given that the Ryzen 3000 series is expected to correct the deficiency of not having as much floating-point muscle as Intel chips, it would have been better if they had brought them out sooner, in 14 nm... and then they could have waited a little, and gone to this node as soon as it was available. Instead, it looks like they'll only get to 7nm+ when it's old news.

One can bet that Intel won't make that mistake with its process nodes.

I think you're a little out of the loop. Intel were at least two years ahead on process as late as 2015. Skylake was on tri-gate 14nm in 2015 while AMD were still building garbage on 28nm in 2016. Now in 2019, a mere four years later, Intel are suddenly 6 months behind when Zen 2 launches.

AMD's timing couldn't be better. There isn't a bad time to be leading. They handed everything they have got to TSMC who are streaking ahead. First to 7nm, then by the time Intel have 10nm working TSMC will deliver 7nm+ according to these reports.

Intel have to finally fix 10nm for full production by the end of this year, and they better be on track for their TSMC 5nm equivalent in 2021 otherwise expect AMD to take away huge swathes of Intel market share in the next 2-3 years.

Expect it anyway for the end of this year and 2020 if Zen2 is as good as leaks suggest thus far.

I have been all Intel for over a decade now. Saying that the security problems are even starting to scare me off Intel. I was also one who dismissed the likes of Spectre last year because they seemed minor. It is proven they were- relative to the latest much more serious issues now appearing.

They have a lot of hurt coming, let's hope they get a grip and start back to thinking about the consumer with a little humble pie.

Very well said.

I'm looking forward to jumping back onto AMD's side. The last time I've used an AMD rig was way back during the Athlon 64 days. The Pentium 4's was much less desirable compared to their offerings.

Hopefully Zen 2 really does live up to expectations.

 

[Evernessince]

I'm very glad to hear that TSMC has gotten EUV working. Presumably Intel will get to this point pretty soon as well. If it is only six months behind, this will not be too big a deal.

But AMD's timing is not that good. Given that the Ryzen 3000 series is expected to correct the deficiency of not having as much floating-point muscle as Intel chips, it would have been better if they had brought them out sooner, in 14 nm... and then they could have waited a little, and gone to this node as soon as it was available. Instead, it looks like they'll only get to 7nm+ when it's old news.

One can bet that Intel won't make that mistake with its process nodes.

I don't see how a 7nm+ refresh next year is old news. It's the same thing AMD did with 1st gen zen and the node bump provided a nice increase in between Zen 1 and Zen 2.

One can very much bet that Intel has already made numerous blunders with it's 10nm node. I don't know where you get the idea that Intel won't mess up again when they've already had 10nm delayed for 3 years at this point.

 

[7thnm]

TSMC has very little volume on 7nm+, most are on regular 7nm. There is more expected for 5nm or 6nm. EUV volume is still negligible. Not enough throughput.

 

[Eric Klien]

"Despite how young 7nm technologies are, having yet to appear in a single laptop or desktop," Here are a couple items that have 7nm in them: iPhone XS and XS Max, iPhone XR, iPad Mini (2019), iPad Air (2019), and AMD Radeon VII. The last item does appear in desktops.

 

[Eric Klien]

TSMC has very little volume on 7nm+, most are on regular 7nm. There is more expected for 5nm or 6nm. EUV volume is still negligible. Not enough throughput.

TSMC is currently mass producing the A13 processor for a little company called Apple using the 7nm+ process. The volume for this order is far from negligible. This is not the only order for 7nm+ that TSMC is processing. They are also currently mass producing the Kirin 985 for a little company called Huawei. TSMC bought more EUV machines than every other company on the planet combined this year to make this happen.

 

[Markoni35]

Next stop: Gamma rays. For atom-sized transistors.

 

[7thnm]

TSMC is currently mass producing the A13 processor for a little company called Apple using the 7nm+ process. The volume for this order is far from negligible. This is not the only order for 7nm+ that TSMC is processing. They are also currently mass producing the Kirin 985 for a little company called Huawei. TSMC bought more EUV machines than every other company on the planet combined this year to make this happen.

So that's the thing, there are only so many tools. CEO said the revenue would be negligible, so it can't be major.