Discussion Intel current and future Lakes & Rapids thread

[mikk]

Looks like the 10nm node is pretty much unsolvable in regard to clocks/power consumption.

Wait for 10+(+) Tigerlake, it looks much better.

 

[vstar]

Wait for 10+(+) Tigerlake, it looks much better.

Don't they also have a 10++(+) node coming out next year with SPR and ADL?

 

[OriAr]

Don't they also have a 10++(+) node coming out next year with SPR and ADL?

Yes they do.

 

[DisEnchantment]

Some sad news about Intel's 10nm nodes (courtesy of WCCFTech):



Faster than Coffee Lake gen9/10 CPUs will only be released in 2022 which means AMD has all the chances to gain an even greater market share with its Ryzen 4000 CPUs. Looks like the 10nm node is pretty much unsolvable in regard to clocks/power consumption.

Let's just hope they will have a working fast and efficient 7nm node by 2022.

Dang,
I have been reading about this on twitter as well as on Tom's

Intel Says Process Tech to Lag Competitors Until Late 2021, Will Regain Lead with 5nm

It looks like 2020 and 2021 are going to be long years for Intel.

www.tomshardware.com

If true, it is devastating.
Two years is a long time in Tech Industry.

I really don't wish for Intel to fall this behind.
In 2022 Intel will have to fight GAA from Samsung/TSMC. Samsung already have GAA test devices with a target of 30%+ performance process over their next gen 5nm process
GAA will bring decent gains similar to the move to FinFET.
I hope all the people they have been assembling lately will help them.

 

[maddie]

Dang,
I have been reading about this on twitter as well as on Tom's

Intel Says Process Tech to Lag Competitors Until Late 2021, Will Regain Lead with 5nm

It looks like 2020 and 2021 are going to be long years for Intel.

www.tomshardware.com

If true, it is devastating.
Two years is a long time in Tech Industry.

I really don't wish for Intel to fall this behind.
In 2022 Intel will have to fight GAA from Samsung/TSMC. Samsung already have GAA test devices with a target of 30%+ performance process over their next gen 5nm process
GAA will bring decent gains similar to the move to FinFET.
I hope all the people they have been assembling lately will help them.

Yeah, so let's see here.

TSMC advancing faster than Intel. Check
Intel 1-2 generations behind at present. Check
Intel will catch up in 2 years. Dream on

 

[Exist50]

Some sad news about Intel's 10nm nodes (courtesy of WCCFTech):



Faster than Coffee Lake gen9/10 CPUs will only be released in 2022 which means AMD has all the chances to gain an even greater market share with its Ryzen 4000 CPUs. Looks like the 10nm node is pretty much unsolvable in regard to clocks/power consumption.

Let's just hope they will have a working fast and efficient 7nm node by 2022.

Just because Intel won't be at process parity doesn't mean we won't see any advancement at all. I don't know how you reached that conclusion.

 

[JasonLD]

Yeah, so let's see here.

TSMC advancing faster than Intel. Check
Intel 1-2 generations behind at present. Check
Intel will catch up in 2 years. Dream on

Though in 3-4 years, TSMC, Samsung, and Intel's progress will all be dependent on ASML's progress on their EUV equipment. I expect all of them will be on same level in terms of process nodes by then. I expect quality of 3D stacking and packaging will be more important.

 

[maddie]

Though in 3-4 years, TSMC, Samsung, and Intel's progress will all be dependent on ASML's progress on their EUV equipment. I expect all of them will be on same level in terms of process nodes by then. I expect quality of 3D stacking and packaging will be more important.

So you think lithography machines is all, or at least the majority of chipmaking?

 

[JasonLD]

So you think lithography machines is all, or at least the majority of chipmaking?

Yes, definitely. 3nm and beyond will need higher NA EUV system which isn't really due until 2023~2024. Going 3nm on current EUV system could be quite a risk, and TSMC is trying to stay with finfet on their 3nm. If they can achieve that, it will be a quite a achievement, but it is still unknown at this point.

 

[DrMrLordX]

Wait for 10+(+) Tigerlake, it looks much better.

We've all been looking at it, but there's sparse results right now.

 

[DisEnchantment]

So you think lithography machines is all, or at least the majority of chipmaking?

Yes, definitely. 3nm and beyond will need higher NA EUV system which isn't really due until 2023~2024. Going 3nm on current EUV system could be quite a risk, and TSMC is trying to stay with finfet on their 3nm. If they can achieve that, it will be a quite a achievement, but it is still unknown at this point.

Just stating that obvious but there is so much more to EUV process tech than just EUV machines.
Masks, pellicles, cell structures, patterning, resistance/capacitance management, cell libraries, interconnects and so much more... all of these need to be designed besides having the EUV machines.
Samsung and TSMC had the same ASML machines but the outcome is different.

 

[coercitiv]

Samsung and TSMC had the same ASML machines but the outcome is different.

Plot twist.

 

[JasonLD]

Just stating that obvious but there is so much more to EUV process tech than just EUV machines.
Masks, pellicles, cell structures, patterning, resistance/capacitance management, cell libraries, interconnects and so much more... all of these need to be designed besides having the EUV machines.
Samsung and TSMC had the same ASML machines but the outcome is different.

Oh, definitely. However, All of the above things you mentioned are done within the specifications of what current EUV machines are capable of, on sub-7nm and any smaller nodes after. At least for the decade, TSMC, Samsung, and Intel will have to follow the ASML’s roadmap basically.

 

[maddie]

Oh, definitely. However, All of the above things you mentioned are done within the specifications of what current EUV machines are capable of, on sub-7nm and any smaller nodes after. At least for the decade, TSMC, Samsung, and Intel will have to follow the ASML’s roadmap basically.

Which was basically true for each earlier state of lithography, and yet we had widely different architectures and performance levels. All engineers, in all fields, work within the limits of the available materials and tools.

 

[Markeyse]

Though in 3-4 years, TSMC, Samsung, and Intel's progress will all be dependent on ASML's progress on their EUV equipment. I expect all of them will be on same level in terms of process nodes by then. I expect quality of 3D stacking and packaging will be more important.

I'm with you on this. I think 3D packaging is really the next step. After 7nm, what more is there? It is time to stack them chips!

 

[jpiniero]

Dell XPS 13 9300 with bright 16:10 display, but disappointing CPU performance

We just received the entry-level spec SLU of the brand-new Dell XPS 13 9300. The highlight is the return of the 16:10 display, but the processor performance is worse compared to the previous model.

www.notebookcheck.net

Looks like Dell got some crap tier i5 1035g1 chips. Performance is what you would maybe expect from a 15 W i5 but it is running at 25 W. So performance on the previous XPS models with Whiskey is faster.

 

[vstar]

https://videocardz.com/newz/intel-alder-lake-s-to-feature-16-cores-125-150w-tdp-and-pcie-4-0

Some new information on ADL-S. Looks like Intel is planning a heterogeneous mix of cores for the desktop (8 big + 8 small). Really interested in why they chose this option and not to go with more 'BIG' cores instead.

 

[mikk]

This might explain the big socket. I wonder what big little architecture they will use, possibly Gracemont.

 

[jpiniero]

https://videocardz.com/newz/intel-alder-lake-s-to-feature-16-cores-125-150w-tdp-and-pcie-4-0

Some new information on ADL-S. Looks like Intel is planning a heterogeneous mix of cores for the desktop (8 big + 8 small). Really interested in why they chose this option and not to go with more 'BIG' cores instead.

It's probably two 4+4 chiplets plus an IGP chiplet. The 125 maybe 150 W TDP at base does suggest that they might indeed backport Alder Lake to 14nm. Given when it'd be released (likely two years from now) you would think they would consider 7 nm.

Wouldn't surprise me if the 6+0 was Rocket Lake's CPU chiplet.

 

[OriAr]

It's probably two 4+4 chiplets plus an IGP chiplet. The 125 maybe 150 W TDP at base does suggest that they might indeed backport Alder Lake to 14nm.

Wouldn't surprise me if the 6+0 was Rocket Lake's CPU chiplet.

Alder Lake is 10nm, that much is certain. If they backported Golden Cove to 14nm it'd have gotten another codename.
Regarding what is the exact configuration... At this point I think a big cores die + a small cores die but my guess is as good as anything else.
This is legitimately the biggest change in desktop since the introduction of x86-64, maybe even bigger.

 

[mikk]

It's probably two 4+4 chiplets plus an IGP chiplet. The 125 maybe 150 W TDP at base does suggest that they might indeed backport Alder Lake to 14nm.

There is also a 8+8+1 variant with a TDP of only 80W TDP in this picture! I doubt this is doable with 14nm because Golden Cove is probably bigger than Willow Cove.

 

[ondma]

Seems strange on the desktop though. I would think big/little would be more suited for mobile, where you could operate on the small cores to save power until a heavy load required the big cores. I think the problem is that yields/power consumption on 10 nm is still not what it should be, so they cant make a 12 or 16 core with manageable power consumption or yields.

 

[vstar]

This might explain the big socket. I wonder what big little architecture they will use, possibly Gracemont.

Golden cove + Gracemont would make sense.

Do we have any predictions of how Gracemont cores would perform?

 

[DrMrLordX]

How odd.

edit: so maybe Gracemont for the small cores. Still odd.

 

[OriAr]

Golden cove + Gracemont would make sense.

Do we have any predictions of how Gracemont cores would perform?

Gracemont should have IPC between SKL to Sunny Cove.
Not too shabby at all, especially if it's towards the higher end of the scale.