Discussion Intel current and future Lakes & Rapids thread

Page 233 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

mikk

Diamond Member
May 15, 2012
4,111
2,105
136
Definitely much faster than Icelake but impossible to say by how much, depending on the map and scene the fps can differ.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
I'm getting sick of the leaks. It'll perform how it's supposed to perform. Even pre-release benchmarks, the one they did with Icelake. You are not DIYing these CPUs, and it'll perform based on what laptop you get.

Icelake's sweet spot is 20W: https://pc.watch.impress.co.jp/docs/column/hothot/1259311.html

15 to 20W results in 30-40% gain, while going from 20W to 25W is only 5-7%.

Oh and another thing, if you go by NBC's result for the Gen 11 graphics, lots of their tests are using the Dell Inspiron 14 which performs quite a bit lower(one reason being it uses DDR4-2666) or HP Pavilion 14, which uses single-channel DDR4-3200 memory.

I'm starting to suspect some people saying the Gen 11 underperforms in games are using those under-optimal systems for comparison. Something like the Yoga 940, or Surface Laptop 3 performs 25-40% better. Sometimes, the differences are greater than 50%.
 
Last edited:

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,572
146
Thought I'd post my own experience with my 8300H + 1050 laptop... but while it performs about the same as Renoir in this game for averages (and this is with dedicated memory and the GPU hitting 1800mhz in game!) the frametimes are beyond poor. Consistent stuttering would be putting it lightly.

So unfortunatly, this isn't quite the game for a comparison vs MX or the mobile GTX1050... but looking at it differently it's a verys trong first showing for Tiger Lake.

Though I'm surprised by the use of DX11 over DX12 - the DX12 implementation of BFV (after many updates) runs smoother on most hardware now than the DX11 version.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
And this adds to the thread how?
I'm sorry for not writing an essay about it at first, but here it goes: by all accounts TGL seems to be a beast, but the moment this guy starts to talk about it out if nowhere, it makes me nervous and think there's something not going as planned.
Hope it's just paranoia!
 

liahos2

Banned
Jan 3, 2020
11
7
51
Thought I'd post my own experience with my 8300H + 1050 laptop... but while it performs about the same as Renoir in this game for averages (and this is with dedicated memory and the GPU hitting 1800mhz in game!) the frametimes are beyond poor. Consistent stuttering would be putting it lightly.

So unfortunatly, this isn't quite the game for a comparison vs MX or the mobile GTX1050... but looking at it differently it's a verys trong first showing for Tiger Lake.

Though I'm surprised by the use of DX11 over DX12 - the DX12 implementation of BFV (after many updates) runs smoother on most hardware now than the DX11 version.
i think someone commented on twitter that FRAPS only in DX11?
 

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,572
146
i think someone commented on twitter that FRAPS only in DX11?
Yeah but MSi Afterburner and CapFrameX both exist and let you use fancy other tools like frame-time plots!

Oh wait, that would be too much info I guess and most improtantly, we're getting into some real nitty gritty benchmarks there :p

Jokes aside, yeah there are two other readily available and easy to use solutions than Fraps, so idk if that's actually a good reason. For all we know it could just be their DX12 driver isn't quite where they want it to be yet.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
You weren't very optimistic about numbers improving this much, though hehe

Actually I kinda did. They were saying 2x over Icelake for a year now. But its good to see the claims confirmed.

Though we're not out of the woods yet.

AMD laptops aren't the only ones configured with single channel memory:

seems like a 70% uplift over icelake G7. nice

Update: According to some in the know(such as Andreas Schlling), the map is demanding and the 1065G7 only gets 12-13 fps.

So Battlefield V must be one of those games where its weak on Gen 11, but fixed on Xe.
 
Last edited:

majord

Senior member
Jul 26, 2015
433
523
136
Someone testing a 4800U or whatever on the same campaign at 25w would be handy.

Adored tested a 4900H and got 25fps AVG.
 
  • Like
Reactions: lightmanek

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Last edited:
  • Like
Reactions: lightmanek

Cardyak

Member
Sep 12, 2018
72
159
106
Yea that would be interesting for sure.
How much could those willow Cove Cores be tipping the scales? ....

Still largely unknown by this point, but most early indicators from leaks and rumours point to Willow Cove having ~7% IPC increase over Sunny Cove, which would mean a ~25% increase over Skylake.

Golden Cove looks to be the one where things will get interesting. The Intel road map from the Architecture day in late 2018 mentioned the term "Strong IPC gain", and also had the following image:

GoldenCove.JPG
 
Last edited:
  • Like
Reactions: french toast

inf64

Diamond Member
Mar 11, 2011
3,685
3,957
136
Still largely unknown by this point, but most early indicators from leaks and rumours point to Willow Cove having ~7% IPC increase over Sunny Cove, which would mean a ~25% increase over Skylake.

Golden Cove looks to be the one where things will get interesting. The Intel road map from the Architecture day in late 2018 mentioned the term "Strong IPC gain", and also had the following image:

View attachment 23523
It's reasonable to expect that Golden Cove might have the same % jump over Willow Cove as Sunny Cove had over Skylake - so I assume around 18-20% is to be expected. The problem for intel is that AMD is not standing still at all, Zen3 is supposed to have major IPC uplift over Zen2(~>20% rumored) and Zen4 will bring another round of IPC jump over Zen3 - probably in the range of Zen+ to Zen2, which is ~15%.

Since Zen2's IPC is ahead of Skylake's by some ~7% in everything except gaming (due to latency advantage Skylake has over CCX/chiplet design of Zen2), so I think that the cumulative impressive jump to Golden Cove ,from Skylake, might be just about enough to equal Zen4:
Skylake - 1x
Zen2 - 1.07x (see AT deep dive of Zen2 and spec 1T tests)
SC - 1.185x
WC - 1.26x (rumored/guesswork)
Zen3 - 1.28x (rumored/guesswork)
GC - 1.5x (rumored/guesswork)
Zen4 - 1.47x (rumored/guesswork)

Clock speed will def. play a role in the overall performance and it looks like TSMC will have a process node advantage over intel when Zen4/GC launch. It's entirely possible that we have a re-run of Zen2 vs Skylake situation where the IPC is fairly close (win some, lose some) but clock speed, power draw and core count might all go to AMD's side. There is a high likelihood that both intel and AMD will hit a clock wall at around 4.9/5Ghz with GC/Zen4 so power, heat, core count and features will be a deciding factor. And price of course.
 
Last edited:

vstar

Member
May 8, 2019
46
39
61
5 GHz on Intel 10nm in 2021? Why do I not see that happening?
Looks like there's good evidence for the TGL-U chips hitting boost speeds of ~4.6-4.7 GHz. Given that, I don't think that 5 GHz is unlikely on their third(?) iteration of the 10 nm process next year, especially for desktop chips at higher TDPs. Although I am not sure about the power consumption, but it should be somewhat more efficient than the 14nm process.
 
  • Like
Reactions: french toast

vstar

Member
May 8, 2019
46
39
61
Some new information on the TGL-U SKUs:

Looks like most of these will be getting the full 96 EUs!


Intel Core i7-1185G7 4 x 3000 MHz, 15 W
Intel Core i7-1165G7 4 x 2800 MHz, 15 W

EDIT: A couple of Acer TGL-U laptops were listed by a Dutch retailer, with a release date of 7/27.
 
Last edited:
  • Like
Reactions: mikk