• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Discussion Intel current and future Lakes & Rapids thread

Page 197 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

uzzi38

Golden Member
Oct 16, 2019
1,817
3,654
116
Same with the GPU, AMD falls behind Intel and suddenly it's not relevant anymore.
Tiger Lake taking the lead for a few months is irrelevant, as iGPUs are. I won't be claiming any different with any of the APUs after Renoir, Van Gogh included.

I will be keeping these comments in mind though, because I get the feeling your stance on this will be rather different come later this year.
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
Because Renoir won't be able to match Intel in singlethread against Tigerlake
Intel can't even match a smartphone cpu in single thread on the same testing conditions of the smartphone, how about eight of them?

Same with the GPU, AMD falls behind Intel and suddenly it's not relevant anymore.
I think the 3 years old vega is still technical superior (and in performance too) to anything intel have or something has changed that i'm not whereabouts?
 
  • Like
Reactions: lobz

mikk

Diamond Member
May 15, 2012
3,253
1,066
136
Intel can't even match a smartphone cpu in single thread on the same testing conditions of the smartphone, how about eight of them?
So you finally switched into a troll mode after you realized you lost this debate, as expected. This is a poor effort from you.
 
  • Haha
Reactions: pcp7

IntelUser2000

Elite Member
Oct 14, 2003
7,615
2,476
136
I counted 36 Icelake-based laptops.

On the Q4 earnings release they said 44 devices, so there's a few that I didn't find. Some are likely models only available for a certain country.

Though unorthodox, there's a good possibility we'll see Comet/Ice/Tigerlake existing at the same time, maybe even Whiskeylake.

They've been putting Icelake into $299 laptops, so it could be relegated to value consumer devices with premium devices going for Tigerlake.

Maybe Icelake will also go into smaller package devices? The "N" variants such as 1030NG4 features a significantly reduced pin count.

I think the 3 years old vega is still technical superior (and in performance too) to anything intel have or something has changed that i'm not whereabouts?
I disagree.

There's certainly huge room to improve for both camps. Renoir's iGPU isn't at all area efficient to have a 150mm2 7nm die.
 

uzzi38

Golden Member
Oct 16, 2019
1,817
3,654
116
I expect Renoir to still be faster than Tiger in actual games, despite the lower than expected CU count from AMD.
I don't. But I don't know what to expect right now, because things are a little less rosy in Tiger Lake-land than ideal.

But if you care about iGPU crowns (which is pointless for the reasons I mentionned before) neither of those two are taking it with this generation of products anyway, so w/e. In the end, it doesn't matter, and whereas AMD are switching to RDNA2 soon and beyond past that, Intel's using Gen12LP till 2022 and beyond.

I find it laughable that somebody here thinks others here don't care about it because AMD will lose to Tiger Lake with Renoir. Intel will be losing the iGPU crown for another 1 and a half to 2 years until Meteor Lake-U releases, doesn't change the fact it's a worthless crown to have.
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
So you finally switched into a troll mode after you realized you lost this debate, as expected. This is a poor effort from you.
Ok let me explain it in your own language what you say:

According to you and some posters here, one intel eight core sixteen thread cpu is slower than one intel four core eight thread cpu.

That's your logic. How about that to start an debate?
 

Exist50

Senior member
Aug 18, 2016
331
352
136
Ok let me explain it in your own language what you say:

According to you and some posters here, one intel eight core sixteen thread cpu is slower than one intel four core eight thread cpu.

That's your logic. How about that to start an debate?
Where did anyone claim that?
 

DrMrLordX

Lifer
Apr 27, 2000
17,856
6,824
136
@mikk

Sorry multiquote screwed up quoting. In any case, I do not know that anyone is really underestimating or overestimating Tiger Lake's CPU performance since we have so little to go on right now (and what little we have isn't all that impressive). You can't tell me that a 4c part is going to compare all that well in CPU performance to an 8c part that has already been on the market for ~6 months by the time Tiger Lake hits the streets. AMD may not have 8c/16t on all their Renoir parts, but they do have one (actually two if you count the HS part), which is something Intel didn't even bother trying on 10nm. I expect many Intel buyers to be looking at 6c 14nm parts more-seriously than Tiger Lake.

8c on a lappie is kind of a gimmick, but it sort of isn't. Getting desktop-like performance out of a mobile CPU is really nice. Getting that with a uarch you launched 6 months ago is even better.

We have almost zero ice lake parts and you are already comparing yet to be released in one or even two years parts from intel with amd current gen?
There's quite a bit of Ice Lake out there, at least in terms of models available. And Tiger Lake is going to be out this year, probably one year after Ice Lake officially launched last year. How long it takes to roll out those units is anyone's guess (hopefully it goes better than the Ice Lake-U rollout).
 

lobz

Golden Member
Feb 10, 2017
1,818
2,353
136
@mikk
8c on a lappie is kind of a gimmick, but it sort of isn't. Getting desktop-like performance out of a mobile CPU is really nice. Getting that with a uarch you launched 6 months ago is even better.
Exactly. Everyone could have said this about 4 cores when all intel had were 2 cores, same for 6 core laptops etc. Since intel did those first, they were the pinnacle of tech innovation. Now that AMD has the only 8-core U laptop, some people say that suddenly it doesn't make sense because TGL is coming with 4 cores.
 

coercitiv

Diamond Member
Jan 24, 2014
4,546
6,262
136
I sense CML-U+MX350 will give TGL-U a run for it's money. Normally they would control pricing to make this kind of combo less appealing, but Renoir will likely limit Intel's movements and CML-U is still the likely candidate for price-cuts & volume.

Say what you will about Skylake, it will probably prove to be the most resilient core in Intel's history, and I'm not talking about security. I think I'll start affectionately calling it Hodor.
 

mikk

Diamond Member
May 15, 2012
3,253
1,066
136
Where did anyone claim that?
Nobody claimed this but he won't understand it. I guess everyone agrees that Renoir-U 8/16 will be faster than TGL-U 4/8 when multithreading really matters, how much is the question. The point is this 8/16 CPU won't be faster everywhere because Intels Willow Cove cores are faster per core, and only one SKU gets the fully enabled 8 core. Therefore the situation with AMD years ago was different, AMD was slower in singlethread and multithread with a much worse battery life and only did have a better GPU.
 

jpiniero

Lifer
Oct 1, 2010
10,111
2,367
136
Took another look at the Tiger Lake die shot, and the die size increase versus Ice Lake looks almost entirely due to the cores, perhaps the additional cache. IGP size looks about the same as Skylake GT2 (~40 mm2)
 

Exist50

Senior member
Aug 18, 2016
331
352
136
Steering back from the iGPU bickering, what are people's expectations for Willow Cove clocks in Tiger Lake U?
 

uzzi38

Golden Member
Oct 16, 2019
1,817
3,654
116
Steering back from the iGPU bickering, what are people's expectations for Willow Cove clocks in Tiger Lake U?
4.4-4.5GHz single, 3.9-4.1GHz all-core for the top end.

Depends on whether or not TVB and/or TB3 are present.
 

vstar

Member
May 8, 2019
41
36
61
4.4-4.5GHz single, 3.9-4.1GHz all-core for the top end.

Depends on whether or not TVB and/or TB3 are present.
Do you expect the base clocks to be significantly higher than what we've seen in the ES leaks?

I'm hoping for
~2.5 GHz for cTDP 15W (vs 2.3 GHz ES)
~3.0 GHz for cTDP 28W (vs 2.7 GHz ES)
 

uzzi38

Golden Member
Oct 16, 2019
1,817
3,654
116
Do you expect the base clocks to be significantly higher than what we've seen in the ES leaks?

I'm hoping for
~2.5 GHz for cTDP 15W (vs 2.3 GHz ES)
~3.0 GHz for cTDP 28W (vs 2.7 GHz ES)
No. I'm not even entirely convinced the 2.3GHz base clocks are 15W. The top end 15W ICL-U SKU - the 1065G7 - has a base clock of 1.3GHz. The 28W SKU has a base of 2.3GHz.

Currently I'm more inclined to believing the 2.3GHz was just an early sample 28W chip and the 2.7GHz is closer to final.

Furthermore, I doubt the difference between 15W and 28W will be so little like the 500mhz you suggested.

With all that in mind, if I'm honest I'm expecting about 2GHz and 2.8GHz for 15W and 28W respectively. But I mean, this is pretyy much a shot in the dark, so take it how you will.
 
  • Like
Reactions: vstar

uzzi38

Golden Member
Oct 16, 2019
1,817
3,654
116
I'm guessing you meant 3.9-4.1 GHz all-core for 6 seconds.
If Ice Lake-U is any idea, then what I mean is 3.9-4.1GHz for however long the boost duration is assuming PL2 is set to 45W or so. ICL-U in CB15 clocks at about 3.3GHz at 35W, and this is a non-AVX workload, so I'd assume that at the bare minimum 40-45W might be needed, more for AVX workloads.
 
  • Like
Reactions: lobz

IntelUser2000

Elite Member
Oct 14, 2003
7,615
2,476
136
Do you expect the base clocks to be significantly higher than what we've seen in the ES leaks?
The -Y chip will probably beat the ICL-U 15W in base clocks lol.

The significance of base clocks are diminished nowadays. You can see laptops using any CPU going below base clocks provided the compute load is great enough(Prime 95 + Furmark for one).
 
  • Like
Reactions: uzzi38

uzzi38

Golden Member
Oct 16, 2019
1,817
3,654
116
The -Y chip will probably beat the ICL-U 15W in base clocks lol.

The significance of base clocks are diminished nowadays. You can see laptops using any CPU going below base clocks provided the compute load is great enough(Prime 95 + Furmark for one).
Agreed on both points. AMD's APUs have this issue especially - and it's not even with heavy compute workloads. When stressing the CPU and iGPU at the same time even iwth just a regular game, you'll often find yourself below base at 15W.

Afaik Ice Lake exhibits this behaviour in AVX512 workloads, not sure about mixed, I haven't checked clocks in mixed workloads.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,615
2,476
136
Afaik Ice Lake exhibits this behaviour in AVX512 workloads, not sure about mixed, I haven't checked clocks in mixed workloads.
Whiskeylake and Amberlake does that too. Amberlake even at 7W can drop to 500MHz depending on implementation.

It also depends a lot on system implementation. The XPS 13 ICL can maintain the minimum 1.3GHz, but on battery it probably drops to 1.1-1.2GHz.

Even with AVX-512, there's a noticeable difference between a code that's designed to stress the FPU(matrix multiplication for example) and applications that use the instruction for real workload such as x265. The load on the CPU is not black and white and can vary in a hundred different ways between.

Icelake only has support for the instruction and not the double throughput so the impact will likely be minimal compared to AVX-256. I don't think we'll see dual AVX-512 support in client for a while.

My experience with the i7 Ivy Bridge XPS 12 suggests that its not at all realistic running such power virus workloads(such as Prime 95 plus Furmark). Most laptops/ultrabooks can maintain better than base frequencies quite readily when running either of them.

Just in web browsing workloads the performance is greatly impacted by power management settings, thermals, and even the condition of the battery. Get a DTR or better yet, get a desktop.
 
Last edited:
  • Like
Reactions: spursindonesia

ASK THE COMMUNITY