Discussion Intel current and future Lakes & Rapids thread

Page 161 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
So what is unclear is the level they can achieve above what they have. My current hypothesis is for desktop chips, considering that ice lake was 900MHz lower than cometlake boost of 4.8, which is 500MHz less than single core boost on desktop, that we will see for desktop frequencies in the mid-4GHz range, similar to AMDs. If that is correct, then we are entering an IPC race.

I do expect Rocket to boost to 5 Ghz. It'll draw a crapton of power however.
 
  • Like
Reactions: ajc9988

mikk

Diamond Member
May 15, 2012
4,111
2,105
136
They won't give yield numbers because it is likely still in the high teens to low 20s. Anand tech noted the number of SKUs is 1/3d their normal stack. There is low volume available of 10nm+ based ice lake chips from ODMs and OEMs.

I don't need numbers, I'm just saying that with Icelake the yields are better for 10+ and probably gets better month for month.
 

ajc9988

Senior member
Apr 1, 2015
278
171
116
I do expect Rocket to boost to 5 Ghz. It'll draw a crapton of power however.
Rocket is likely 14nm++++ with a 10nm GPU, potentially. So being on 14nm while reducing back to 8 cores, I do agree.

That, however, may not be enough to overcome the IPC changes of Zen 3 and Zen 4, although I do feel it will be competitive, just a couple percent behind on IPS (IPC*Frequency).
 

ajc9988

Senior member
Apr 1, 2015
278
171
116
I don't need numbers, I'm just saying that with Icelake the yields are better for 10+ and probably gets better month for month.
While comparing it to TSMC's process with equivalent density, you have a drastic disparity on yields, even if improving. Then you have my analysis on frequency increases at the voltage knee showing after the first architecture change and iterative node.

This is why I don't understand why people think Intel will magically significantly change their uplift. Sure, as a percentage the increase is greater on frequency than 14nm, but each 100MHz beyond this will be much harder is what I'm suggesting.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Er, I never thought I'd be saying this, but @liahos1 is right you know... Tiger Lake will be better in terms of ST and iGPU performance than Renoir, and will clock notably better than Ice Lake as well... the biggest problem with 10nm is the yields, not the clock-ability anyway. ICL-U was a little on the rushed side, so clocks worse than it should.

AMD will be in the lead for MT performance and maybe power efficiency with Renoir (7nm+LPDDR4X will rein in idle power draw very significantly), but let's not kid ourselves and pretend Renoir is a full on spanking.

Also, anyone that thinks Zen 3 will be the same as Willow Cove IPC-wise is kidding themselves. It'll be Sunny Cove level, maybe a little higher depending on workload, but I highly doubt it will flat out beat Willow.

That's just for mobile though. Intel are still in a rough patch for desktop mind you.
I'm sure he has valid points sometimes - like everybody else on this planet. This however, 'you guys are delirious with hatred', is not one of them for me and not something after which I, for example, would wanna continue the conversation. I read the post he reacted to with this and I saw zero hatred there, and just as much delirium...
 

ajc9988

Senior member
Apr 1, 2015
278
171
116
I'm sure he has valid points sometimes - like everybody else on this planet. This however, 'you guys are delirious with hatred', is not one of them for me and not something after which I, for example, would wanna continue the conversation. I read the post he reacted to with this and I saw zero hatred there, and just as much delirium...
I'm just astonished at the claims.

If Zen 2 is 6-7% IPC over coffee refresh, and ice lake with sunny cove is 18% IPC, sunny would be 12% IPC over Zen 2, estimated (although looking at the anandtech article of IPC, 7-9% was seen over Zen 2 3900X, let's kick in the couple percent for the desktop variant, just in case).

We then would have 5-10% on Willow cove according to estimates from leaks, so let's use 7%, the industry standard before the recent large increases on IPC. So we are at 19% over Zen 2 for Willow cove.

RedGamingTech's leak suggest 10-12% IPC increase on integer and 17% on the average for mixed workloads, with as high as 50% in some floating point operations. If that 17% number is true, that gives Intel a 2% IPC lead over Zen 3.

To be fair, Zen 3 mobile won't come out until Q1 2021 most likely. But also betting fair, we didn't really see commercially available laptops with ice lake until Q4 2019, meaning a similar scenario may happen for 2020 tiger lake.

This means frequency will be the make or break performance metric for that round of devices. With that, on other considerations like power efficiency, Intel may still be the preferred platform. But sticking to performance, that is pretty close. If Intel's boost is increased to single core of 4.0-4.1, while AMD's is around 3.8, Intel will have a 5% lead in ST performance.

But AMD could have more cores and a higher all core boost, which would give them the MT work horse win.

That suggests a fairly tight competition even in mobile regarding tiger lake and Zen 3.

In desktop, it really is a different story. 10-core skylake on 14nm++++ (who cares if three, four, or five iterations at this point) will square off against Zen 3, which would have over 20%, approaching 25%, IPC advantage. If it can reach 4.4GHz all core, that would be about the equivalent of Intel at 5.2-5.4. Suddenly, AMD could take more ST workloads while also having more cores.

Server side, Intel hadn't shown much to make them even really in contention with Zen 2 release, so Zen 3 continues the beating.

So on content alone, it does seem to be wishful thinking.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
I'm just astonished at the claims.

If Zen 2 is 6-7% IPC over coffee refresh, and ice lake with sunny cove is 18% IPC, sunny would be 12% IPC over Zen 2, estimated (although looking at the anandtech article of IPC, 7-9% was seen over Zen 2 3900X, let's kick in the couple percent for the desktop variant, just in case).

We then would have 5-10% on Willow cove according to estimates from leaks, so let's use 7%, the industry standard before the recent large increases on IPC. So we are at 19% over Zen 2 for Willow cove.

RedGamingTech's leak suggest 10-12% IPC increase on integer and 17% on the average for mixed workloads, with as high as 50% in some floating point operations. If that 17% number is true, that gives Intel a 2% IPC lead over Zen 3.

To be fair, Zen 3 mobile won't come out until Q1 2021 most likely. But also betting fair, we didn't really see commercially available laptops with ice lake until Q4 2019, meaning a similar scenario may happen for 2020 tiger lake.

This means frequency will be the make or break performance metric for that round of devices. With that, on other considerations like power efficiency, Intel may still be the preferred platform. But sticking to performance, that is pretty close. If Intel's boost is increased to single core of 4.0-4.1, while AMD's is around 3.8, Intel will have a 5% lead in ST performance.

But AMD could have more cores and a higher all core boost, which would give them the MT work horse win.

That suggests a fairly tight competition even in mobile regarding tiger lake and Zen 3.

In desktop, it really is a different story. 10-core skylake on 14nm++++ (who cares if three, four, or five iterations at this point) will square off against Zen 3, which would have over 20%, approaching 25%, IPC advantage. If it can reach 4.4GHz all core, that would be about the equivalent of Intel at 5.2-5.4. Suddenly, AMD could take more ST workloads while also having more cores.

Server side, Intel hadn't shown much to make them even really in contention with Zen 2 release, so Zen 3 continues the beating.

So on content alone, it does seem to be wishful thinking.
I wouldn't bet huge piles of money on ICL being more power efficient than Zen2 mobile. As for tiger lake coming out in 1 year? Now that is the real wishful thinking.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
I wouldn't bet huge piles of money on ICL being more power efficient than Zen2 mobile. As for tiger lake coming out in 1 year? Now that is the real wishful thinking.

Tiger Lake should be coming soon actually, first half of next year if not Q1. Intel has a pretty strong incentive to not hold it back if it has improvements that improves yield over Icelake.
 

naukkis

Senior member
Jun 5, 2002
701
569
136
Nobody noticed that those Sunnycove-cores seems to be terrible inefficient? Their multicore speed seems to be actually worse than that of 12nn AMD Zen+cores. With that kind of performance/watt server versions of that Sunnycove will be useless.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Tiger Lake should be coming soon actually, first half of next year if not Q1. Intel has a pretty strong incentive to not hold it back if it has improvements that improves yields over Icelake.
Are we talking about the same tiger lake? The one with the Xe graphics? Well, then... let's hold our breath... ;)
 
  • Like
Reactions: ajc9988

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
Are we talking about the same tiger lake? The one with the Xe graphics? Well, then... let's hold our breath... ;)

DG1 is likely to be announced soon too, yes. But you are only getting the 128 EU die. Don't know if they will bother with a desktop card of it.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
Nobody noticed that those Sunnycove-cores seems to be terrible inefficient? Their multicore speed seems to be actually worse than that of 12nn AMD Zen+cores. With that kind of performance/watt server versions of that Sunnycove will be useless.

That's 10 nm for you.
 
  • Like
Reactions: ajc9988

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
its 100 dollars more than the ryzen sku lol. Tigerlake clocks will improve as will IPC and GPU performance. you guys are delirious with hatred

Exactly. The review clearly shows no reason to buy the ryzen based one even if it's $100 less. But I would pay that $100 alone for the battery life not to mention the better performance. Graphics on a laptop don't really matter to me. Gaming on one isn't ideal anyway.
 

ajc9988

Senior member
Apr 1, 2015
278
171
116
Tiger Lake should be coming soon actually, first half of next year if not Q1. Intel has a pretty strong incentive to not hold it back if it has improvements that improves yield over Icelake.
Who said Intel is holding back. Show me your basis for thinking 10nm++ will be ready Q1. If it is not in volume right now to build up inventory for Q1, then it is not happening. Show me tiger lake was taped out 9 months ago on a process iteration that was just a glimmer back then.

If being reasonable, it is obvious the just released ice lake and the comet lake chips on mobile will take us into the summer of 2020. I'm betting no real tiger lake volume until Q4 2020 or later.
 
  • Like
Reactions: spursindonesia

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
Who said Intel is holding back. Show me your basis for thinking 10nm++ will be ready Q1.

That's the rumor. Icelake is more late because of the bad yields than Tigerlake being early. Tigerlake's design was presumably completed a long time ago. The ++ doesn't mean much from an fab perspective, it's the same tooling I think.

Say March or something. You might see the Comet Lake with LPDDR4X then as well.
 

ajc9988

Senior member
Apr 1, 2015
278
171
116
That's the rumor. Icelake is more late because of the bad yields than Tigerlake being early. Tigerlake's design was presumably completed a long time ago. The ++ doesn't mean much from an fab perspective, it's the same tooling I think.

Say March or something. You might see the Comet Lake with LPDDR4X then as well.
Show me a link to the rumor so I can evaluate the claim, please. Or cite to the outlet covering it, such as I did crediting redgamingtech for IPC on zen 3, anandtech's IPC numbers for ice lake, or semi-accurate for yields.

I don't care that the core design of Willow lake was done long ago, or that intel's ramp on ice lake volume was Q4 when "released" this summer. Show me the tape out. Show me anything to support your claim, please.

Edit: so on the mobile leaked roadmap, it does say Q2 2020, even though the SIPP roadmap literally shows Q2 2021. This suggests consumer availability somewhere in between, like Q4 2020, which was my stated estimate.
 
Last edited:

naukkis

Senior member
Jun 5, 2002
701
569
136
That's 10 nm for you.

That SOC seems to be fine -only CPU cores are extremely power hungry. Is that all pointing to Intel 10nm or is there also problem with those cpu cores? If it's about cores expect Intel to axe all Sunnycove products except those that have only few cores.