Discussion Intel current and future Lakes & Rapids thread

Page 130 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Bouowmx

Golden Member
Nov 13, 2016
1,138
550
146
The Phantom canyon config is so weird.

U processors don't have PCIE lanes, so PCIE devices go through PCH, as the images show: limited to x4 (unlike desktop, which allegedly gets x8). Normally, a H processor would be more suitable.
2021 is time for next-gen NVIDIA or even Intel GPU, but I can see why Turing is named, as the presentation was given 'today'.
 
  • Like
Reactions: thigobr

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
There's another NUC called Ghost Canyon coming out, which I believe will offer a 3.0 x16 slot. Uses Coffee Lake-H Refresh. This Phantom Canyon would be an alternative.

Also the Thunderbolt is on the CPU die (for now?) which has 4 lanes.
 

FanlessTech

Member
Oct 25, 2015
33
19
51
From what I understand, the Phantom Canyon NUC is still a work in progress, very fluid situation with multiple concepts / form factors floating around.
The Ghost Canyon NUC is pretty much finalized and is the next big release (it's going to rock the SFF world).
 
  • Like
Reactions: Ottonomous and Glo.

ondma

Platinum Member
Mar 18, 2018
2,718
1,278
136

They must be on 10nm, right, considering the names!! (Phantom, Ghost, get it??)

In any case, I can see a nuk for a desktop or enterprise use. I occasionally use a mass spec lab with instruments very crowded together, lots of noise from the fans, and lots of heat produced.. Seems like a nuk would be great for this. Not sure I see the point of a "gaming" nuk though.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
Hmm, Prime95 is also AVX512 if I recall correctly?

As of 29.5 yes.

They must be on 10nm, right, considering the names!! (Phantom, Ghost, get it??)

Phantom yes, Ghost no. But yeah, ba dum tsh.

I guess the question would also be why they wouldn't use the 35 W 6 core Rocket Lake U instead.

Um, when is Rocket Lake U even going to be ready for the market? That comes well after IceLake-U/Y's limited release. I think 2020?
 

SAAA

Senior member
May 14, 2014
541
126
116
10++ nm can't come soon enough.
If Tiger Lake is just another ~5% IPC on top of Ice who cares for 5 GHz, 4.5 will suffice, at 3.6 it's already as good as top of the line CPUs:

https://cpu.userbenchmark.com/Compare/Intel-0000-vs-Intel-Core-i9-9900K/m863395vs4028
https://cpu.userbenchmark.com/Compare/Intel-0000-vs-AMD-Ryzen-7-3800X/m863395vs4047

Only thing left is power consumption.. and more cores.
Current Ice lake die would grow by what, 20-25% adding 4 more cores? Cut the IGP in half and go for 12 cores monolithic, that's the only way.
 

OriAr

Member
Feb 1, 2019
63
35
91
10++ nm can't come soon enough.
If Tiger Lake is just another ~5% IPC on top of Ice who cares for 5 GHz, 4.5 will suffice, at 3.6 it's already as good as top of the line CPUs:

https://cpu.userbenchmark.com/Compare/Intel-0000-vs-Intel-Core-i9-9900K/m863395vs4028
https://cpu.userbenchmark.com/Compare/Intel-0000-vs-AMD-Ryzen-7-3800X/m863395vs4047

Only thing left is power consumption.. and more cores.
Current Ice lake die would grow by what, 20-25% adding 4 more cores? Cut the IGP in half and go for 12 cores monolithic, that's the only way.

Math suggests TGL is 17% improvement in IPC over ICL, which is straight up nuts (Intel improving IPC by almost 40% in little more than a year after zero improvement in 4 years is pretty insane).

If RCL is WC like I suspect it is.... 5+ GHz WC is gonna be straight up crazy performance.
 

mikk

Diamond Member
May 15, 2012
4,112
2,108
136
Current Ice lake die would grow by what, 20-25% adding 4 more cores? Cut the IGP in half and go for 12 cores monolithic, that's the only way.


10 cores with a 32 EU Gen12 iGPU as far as we know.
 

Khato

Golden Member
Jul 15, 2001
1,202
240
106
Math suggests TGL is 17% improvement in IPC over ICL, which is straight up nuts (Intel improving IPC by almost 40% in little more than a year after zero improvement in 4 years is pretty insane).
That's what happens when you have IP development gated behind process. Just because there hasn't been a product release doesn't mean that IP development has stopped.

It'll be interesting to see where Intel goes with number of cores across the product line. I've yet to notice a tangible benefit to going beyond quad core for anything other than video transcode. Gaming is at least making markedly better use of additional cores, but gains beyond 6 cores with SMT are still minimal. And that shift only occurred due to the necessity of the PS4/XBox One CPU architecture - the next generation of consoles shifting away from an anemic CPU architecture is going to remove the requirement to highly optimize games for multi-threading.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
Gaming is at least making markedly better use of additional cores, but gains beyond 6 cores with SMT are still minimal.

With the increased adoption of low level APIs like Vulkan and DX12, you can expect that trend to increase rather than decrease. As an example, Ubisoft has finished porting over their AnvilNext engine to DX12, which is their main engine for the Assassin's Creed series and the Ghost Recon titles. Ghost Recon Breakpoint will be the first title they release which will run with DX12. Bethesda is also making it a point to port all of their engines over to Vulkan.

And that shift only occurred due to the necessity of the PS4/XBox One CPU architecture - the next generation of consoles shifting away from an anemic CPU architecture is going to remove the requirement to highly optimize games for multi-threading.

I would hope not. Why waste the amazing progress the gaming industry has had over the years with increased parallelism? There are many incentives to continue on that path with next gen consoles, like more games targeting 60 FPS, more realistic physics simulation etcetera.

Improving multithreading even more would certainly help them to achieve those goals.
 

Khato

Golden Member
Jul 15, 2001
1,202
240
106
I would hope not. Why waste the amazing progress the gaming industry has had over the years with increased parallelism? There are many incentives to continue on that path with next gen consoles, like more games targeting 60 FPS, more realistic physics simulation etcetera.

Improving multithreading even more would certainly help them to achieve those goals.
I doubt they're going to waste the progress made. But if the increased performance of the next generation consoles allows them to meet performance goals using coarse grained/simple multithreading approaches that take a quarter of the time/require less experienced engineers to write? If there's no incentive to extract as much parallelism as possible, they won't.

It's definitely an interesting tangential subject to CPU core counts for the mainstream. Product planning is certainly a responsibility I wouldn't care for - build a product with more cores than maistream workloads make use of results in a marked reduction in margins while not having enough cores means lost sales to the competition. Personally, I've yet to encounter anything that left me wishing for more than 8 cores and hence always wonder at those clamoring for more.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
Math suggests TGL is 17% improvement in IPC over ICL, which is straight up nuts (Intel improving IPC by almost 40% in little more than a year after zero improvement in 4 years is pretty insane).

You can't tell the turbo clock based upon what it says. All it says is that the average is 3.6.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
I should add that if Tigerlake is using chiplets, they would have an incentive to not hold it back if it's ready because of the better yield they would get.
 

mikk

Diamond Member
May 15, 2012
4,112
2,108
136
You can't tell the turbo clock based upon what it says. All it says is that the average is 3.6.


Yes but the scores on this site are really good so early in development, it beats the best i7-8565U devices on userbenchmark. The ES samples from Icelake 1 year before launch were much slower clocked, look at this from 11 months ago: https://www.userbenchmark.com/UserRun/10940550

10nm is clearly in much better shape now. I still wonder if Tigerlake is made on 10nm++ or really on 10nm+ as Intel claimed.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
Yes but the scores on this site are really good so early in development.

That was what I was sort of getting at - Tigerlake looks like it is further along than you would think. When you will see it is another story, but as I mentioned Intel has an incentive to not hold it back.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
Cut the IGP in half and go for 12 cores monolithic, that's the only way.

If Tigerlake is not monolithic, then cutting the iGPU in half won't help them. I strongly suspect that, in light of Intel's yield problems on 10nm, Tigerlake will use EMIB to connect CPU and SoC die/iGPU in order to reduce die sizes and improve yields. It may just be that they can't reliably produce dice larger than 4c on 10nm in any significant quantity (see delays for Icelake-SP).

Rocketlake is a different issue altogether. Willow Cove implemented on 14nm++(+) as Rocket Lake would be an interesting product. Perhaps too little/too late given that it'll be a 2020 product, but still a step up from Coffeelake/Comet Lake.
 

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
Math suggests TGL is 17% improvement in IPC over ICL, which is straight up nuts (Intel improving IPC by almost 40% in little more than a year after zero improvement in 4 years is pretty insane).

I mean I have said it before. Intel obviously didn't stop cpu uarch development just because of their process issues. A huge jump was to be expected due to what on the desktop will probably end up being half a decade of skylake-cores.
 
  • Like
Reactions: coercitiv