Intel 10nm Ice Lake to Quantitatively Debut Within 2019

[lexluthermiester]

Shareholders...

Good point.

 

[GoldenX]

The continued problems for 10nm is a sad tale, since Intel is sitting on a new faster architecture they can't use (yet).

But in all the Zen(2) hype, it's easy to forget that despite all these troubles, Intel still have the fastest core performance, and is quite likely going to do well against Zen 2 for the mainstream platform. The only "hole" will be the lack of a direct competitor to AMD's expected 12-core CPU, but that shouldn't matter a lot. Coffee Lake (v2) will not be inferior when Zen 2 launches, the only sad part is that Intel will be stagnant and seemingly have nothing new in this segment this year.

I'm more curious in what lies ahead. We still don't really know much about Comet Lake-S, and as far as I can see there are three possibilities;
1) Just another iteration of Skylake with no real IPC changes, but minor tweaks in the layout to achieve better thermals and lower voltage, which is utilized for marginal clock gains. This is very much possible, but boring.
2) Some architectural tweaks (like Cascade Lake-SP) to achieve small IPC gains.
3) A "backport" of Ice Lake/Sunny Cove to 14nm.
(or a combination of these)
The only indicator I've seen so far is the support in Core Boot and Linux indicating it may be a relative of Skylake.

Please no more Skylake. We will have to start calling the process 14##nm to save space.

 

[notb]

Its also a low performance product, so it does not justify the jump to 10nm.

Why not? Why do you think 10nm is for high performance? Because AMD told you so?

The point was this CPU doesnt tell us much about the state of 10nm.

Again: why not?
It tells us 10nm is good enough for one of the most important CPU lines Intel makes. The next step (and final confirmation that all is well) will be when (if) Intel puts it in Xeon Platinum.

And rest assured this CPU will be scarce and cost a pretty penny because of it. Not directly the greatest perks for a mainstream product...

-U SoCs are expensive anyway. Even a mainstream i5-8250U (4C/8T) costs $297. Top of the range i7-8665U costs $409.
But you buy a laptop, not a SoC. You never really know how it impacts the final price.

Now, the supply of these CPUs is still a mystery, so lets just wait for that.

 

[qcmadness]

Why not? Why do you think 10nm is for high performance? Because AMD told you so?

Again: why not?
It tells us 10nm is good enough for one of the most important CPU lines Intel makes. The next step (and final confirmation that all is well) will be when (if) Intel puts it in Xeon Platinum.

Advanced lithography is now driven by performance jump rather than economical reasons as per-transistor cost (not per die size cost) is now going up starting from 14nm.
If you cannot deliver higher performance, you are just better to stay on 14nm. That's why Intel's 10nm is a failure judging from present data (lower frequency, worse thermals).

That's why Globalfoundries just abandoned the plans to upgrade to 7nm. No one will ask GF to manufacturer chips if GF could not bring up a competitive node.

The continued problems for 10nm is a sad tale, since Intel is sitting on a new faster architecture they can't use (yet).

But in all the Zen(2) hype, it's easy to forget that despite all these troubles, Intel still have the fastest core performance, and is quite likely going to do well against Zen 2 for the mainstream platform. The only "hole" will be the lack of a direct competitor to AMD's expected 12-core CPU, but that shouldn't matter a lot. Coffee Lake (v2) will not be inferior when Zen 2 launches, the only sad part is that Intel will be stagnant and seemingly have nothing new in this segment this year.

I'm more curious in what lies ahead. We still don't really know much about Comet Lake-S, and as far as I can see there are three possibilities;
1) Just another iteration of Skylake with no real IPC changes, but minor tweaks in the layout to achieve better thermals and lower voltage, which is utilized for marginal clock gains. This is very much possible, but boring.
2) Some architectural tweaks (like Cascade Lake-SP) to achieve small IPC gains.
3) A "backport" of Ice Lake/Sunny Cove to 14nm.
(or a combination of these)
The only indicator I've seen so far is the support in Core Boot and Linux indicating it may be a relative of Skylake.

From what Intel presented recently, they don't have a new faster architecture. The so-called Sunny Cove is just a tweaked Skylake with minor updates to architecture. The IPC gain is likely less than 10%.
Unless Intel can magically clock their 10nm parts to 5.5GHz (very unlikely), I don't think their lead on single-thread performance is large enough to warrant the disadvantages on thermals and multi-core performance.

 

[John Naylor]

At this point I don't even care about Intel anymore, because with SO friggin many architectures/designs/CPU names/whatever I don't even know what is what and I wouldn't be able to choose anything even if I wanted to.
Screw it, anything AMD will be a great upgrade from 3770K.

Ok most of here are nerds or geeks so we read about this 'nerd [porn" stuff ... But really no need to pay attention other than if you into "the science" .... If I wanna race cars, I don't care how many cylinders or how many cu.in. If Motor A goes faster than Motor B, all other important things (reliability, fuel conservation, etc) being relatively equal, my only concern need be "what one has the best chance of me crossing the line first" ? Look at what you do application wise and simply the one that completes those tasks faster. In the end, nothing else really matters.

AMD said HBM was gonna change everything how many years ago ? .... and here we are in 2019 and it's and GDDR is still crushing it ... it's been a non-ffactor. Why care if a CPU has 12, 18, 24, 32 cores, if the competition's 6 core completes what ya wanna do faster, it don't matter. Why care if it's 7, 10, 12 nm, if the competition's unlucky 13 nm CPU completes what ya wanna do faster, it don't matter. Focusing on the technology before it's advantages have actually been demonstarted is fun perhaps, but little else.

 

[voltage]

I wont hold my breath, but I do hope ice lake becomes available in 2019, even if its just mobile for now.

 

[R0H1T]

Ok most of here are nerds or geeks so we read about this 'nerd [porn" stuff ... But really no need to pay attention other than if you into "the science" .... If I wanna race cars, I don't care how many cylinders or how many cu.in. If Motor A goes faster than Motor B, all other important things (reliability, fuel conservation, etc) being relatively equal, my only concern need be "what one has the best chance of me crossing the line first" ? Look at what you do application wise and simply the one that completes those tasks faster. In the end, nothing else really matters.

AMD said HBM was gonna change everything how many years ago ? .... and here we are in 2019 and it's and GDDR is still crushing it ... it's been a non-ffactor. Why care if a CPU has 12, 18, 24, 32 cores, if the competition's 6 core completes what ya wanna do faster, it don't matter. Why care if it's 7, 10, 12 nm, if the competition's unlucky 13 nm CPU completes what ya wanna do faster, it don't matter. Focusing on the technology before it's advantages have actually been demonstarted is fun perhaps, but little else.

Except you must've missed the many high performance chips where HBM not only crushes anything GDDRxx but also killed HMC, Intel's answer to HBM. As of this moment Intel, Google, Nvidia, AMD, Xilinx & a few other major players employ HBM for various solutions ranging from GPU, FPGA to dedicated AI accelerators. Now the price & availability is a different topic but so far as crushing (the competition) is concerned, it's HBM that's doing it not the other way around.

 

[efikkan]

Advanced lithography is now driven by performance jump rather than economical reasons as per-transistor cost (not per die size cost) is now going up starting from 14nm.

Cost is absolutely a consideration. And yields are much more complex than just how many chips are working or not. And even with generally low yields there are probably a few good chips capable of decent clocks. So if they chose to make Ice Lake-S, it would probably result in very few salable chips since most of these need relatively high clocks, but if used for Ice Lake-SP, most of which will operate at lower clocks, they can actually manage to produce a decent number of chips, and get good benefits from the new expensive node.

From what Intel presented recently, they don't have a new faster architecture. The so-called Sunny Cove is just a tweaked Skylake with minor updates to architecture. The IPC gain is likely less than 10%.

That's where you're very wrong.
Sunny Cove is the largest overhaul in many years, and is larger in scope than both Skylake and Haswell. Performance gains remains to be seen, but it does change/improve almost everything across the design, including major overhauls of cache (capacity and bandwidth), more than doubling of int mul/div performance, improved memory address calculation, doubling of load/store bandwidth and more.

Unless Intel can magically clock their 10nm parts to 5.5GHz (very unlikely), I don't think their lead on single-thread performance is large enough to warrant the disadvantages on thermals and multi-core performance.

AMD recently stated that they expect clock speeds to decrease over the next years. We are probably at or near the peak of clock speed on the current type of semiconductor materials, so don't expect to see much in the 5-6-7 GHz range. Both Coffee Lake and Zen+ are pushing their respective chips into throttle territory, and it should be obvious that there are not huge gains to be expected in the future. I'm more interested in achieving good base clocks across many cores rather than max boost.

The good thing about IPC changes, is that even a modest IPC boost with a modest increase in die size, is likely to outperform the previews architecture while operating at lower clocks and achieve better thermals. And when it comes to multicore scaling, single core performance is actually very important, as most non-server workloads are synchronous, fewer faster cores is always going to outperform many slower cores. And going into the next decade, IPC is only going to get more important.

 

[Vayra86]

Advanced lithography is now driven by performance jump rather than economical reasons as per-transistor cost (not per die size cost) is now going up starting from 14nm.
If you cannot deliver higher performance, you are just better to stay on 14nm. That's why Intel's 10nm is a failure judging from present data (lower frequency, worse thermals).

That's why Globalfoundries just abandoned the plans to upgrade to 7nm. No one will ask GF to manufacturer chips if GF could not bring up a competitive node.


From what Intel presented recently, they don't have a new faster architecture. The so-called Sunny Cove is just a tweaked Skylake with minor updates to architecture. The IPC gain is likely less than 10%.
Unless Intel can magically clock their 10nm parts to 5.5GHz (very unlikely), I don't think their lead on single-thread performance is large enough to warrant the disadvantages on thermals and multi-core performance.

That's where you're very wrong.
Sunny Cove is the largest overhaul in many years, and is larger in scope than both Skylake and Haswell. Performance gains remains to be seen, but it does change/improve almost everything across the design, including major overhauls of cache (capacity and bandwidth), more than doubling of int mul/div performance, improved memory address calculation, doubling of load/store bandwidth and more.

He's not wrong. He's cautious, we haven't seen numbers so Sunny Cove might just as well be Intel's Bulldozer. The fact remains that Intel is not pushing a higher performance product first to 'show off' 10nm, which is telling. It means they haven't got a real message for high performance to deliver. If they can only gain IPC and perf/watt by cutting back on clocks and the end result is a slower CPU, Sunny Cove might very well be a faster architecture, but the end result is still not satisfactory. Keep in mind this is the same company that felt a 5 Ghz water chiller demo was an important thing to do (performance!).

So far what we have is a crapload of powerpoint slides and vague marketing terminology, alongside the occasional stab at the competition. What we don't have is Intel showing to be proud of what's on the horizon. We're seeing damage control here.

So far Intel hasn't managed a 'major' IPC boost since Haswell, really. They've been lots of baby steps and Skylake wasn't all that earth shattering either besides the move to DDR4. The biggest gain in Core since Sandy Bridge has been that we got our high clocks 'back' after all those IPC bumps and being 'stuck' at 4.2~4.6 Ghz for years (Best-case). That is also the only thing giving Intel the lead for many a use case.

 

[efikkan]

The fact remains that Intel is not pushing a higher performance product first to 'show off' 10nm, which is telling.

Well, they are, it's called Ice Lake-SP. (unless they change their mind)
With limited production capacity, especially for higher clocked chips, it's no point in wasting that on mainstream mid-range and low-end CPUs.

It means they haven't got a real message for high performance to deliver. If they can only gain IPC and perf/watt by cutting back on clocks and the end result is a slower CPU

With a hypothetical 5% IPC gain across the board, they could easily cut 200 MHz max boost and still have higher performance, just as an example.

 

[SL2]

You could define lazy as being content with providing just quad cores for well over a decade or in fact overcharging for HEDT.

Yeah, but like I said, the struggle with 10 nm is not a sign of laziness.

 

[qcmadness]

That's where you're very wrong.
Sunny Cove is the largest overhaul in many years, and is larger in scope than both Skylake and Haswell. Performance gains remains to be seen, but it does change/improve almost everything across the design, including major overhauls of cache (capacity and bandwidth), more than doubling of int mul/div performance, improved memory address calculation, doubling of load/store bandwidth and more.

https://images.anandtech.com/doci/13699/Ronak26.jpg
https://images.anandtech.com/doci/13699/Ronak28.jpg

To summarize,
1. Increase buffer sizes
2. 50% increase in L1 data cache
3. Larger L2 cache size
4. Larger microops cache
5. Larger 2nd level TLB
6. 2 more execution ports (from 8 to 10)
7. Double store data width
8. Execution port more capable

Haswell improvements
Exists: 1, 5, 6 (from 6 to 8), 8, double cache bandwidth, double load / store bandwidth

Skylake improvements
Exists: 1

Is Haswell much better than IvyBridge? I don't think so.
Is Sunny Cove introducing more improvements than Haswell? I don't think so too.

 

[efikkan]

Haswell improvements
Exists: 1, 5, 6 (from 6 to 8), 8, double cache bandwidth, double load / store bandwidth

Skylake improvements
Exists: 1

Is Haswell much better than IvyBridge? I don't think so.
Is Sunny Cove introducing more improvements than Haswell? I don't think so too.

I don't know what you mean by this?
The improvements I talked of was of Sunny Cove vs. Skylake.
Here is another image to compare with Skylake:
https://images.anandtech.com/doci/13405/Ronak20_575px.jpg
As you can see, it replaces one Mul unit with two Mul units and a dedicated Div unit which will more than double integer mul/div performance.
Sunny Cove expands/extends pretty much every aspect except int ALUs (which they increased to 4 in Haswell), and vector units is only marginally improved over Skylake-X/SP, but significantly improved over Skylake-S of course.

 

[qcmadness]

I don't know what you mean by this?
The improvements I talked of was of Sunny Cove vs. Skylake.
Here is another image to compare with Skylake:
https://images.anandtech.com/doci/13405/Ronak20_575px.jpg
As you can see, it replaces one Mul unit with two Mul units and a dedicated Div unit which will more than double integer mul/div performance.
Sunny Cove expands/extends pretty much every aspect except int ALUs (which they increased to 4 in Haswell), and vector units is only marginally improved over Skylake-X/SP, but significantly improved over Skylake-S of course.

Doubling the number of MUL units = double integer mul/div resource =/= double integer mul/div performance =/= double integer performance
If you know what the ports means.

 

[Scrizz]

Ok most of here are nerds or geeks so we read about this 'nerd [porn" stuff ... But really no need to pay attention other than if you into "the science" .... If I wanna race cars, I don't care how many cylinders or how many cu.in. If Motor A goes faster than Motor B, all other important things (reliability, fuel conservation, etc) being relatively equal, my only concern need be "what one has the best chance of me crossing the line first" ? Look at what you do application wise and simply the one that completes those tasks faster. In the end, nothing else really matters.

AMD said HBM was gonna change everything how many years ago ? .... and here we are in 2019 and it's and GDDR is still crushing it ... it's been a non-ffactor. Why care if a CPU has 12, 18, 24, 32 cores, if the competition's 6 core completes what ya wanna do faster, it don't matter. Why care if it's 7, 10, 12 nm, if the competition's unlucky 13 nm CPU completes what ya wanna do faster, it don't matter. Focusing on the technology before it's advantages have actually been demonstarted is fun perhaps, but little else.

This

 

[R-T-B]

One Intel

As cool as a Intel civil war complete with rebel devteam would be for competition, I kinda think that's NOT happening. Unneccesary propaganda is unnecessary. You work for the same company, bro.

 

[londiste]

For perspective, based on what both manufacturers have revealed - changes in Sunny Cove are much more extensive than Zen/Zen+ > Zen2.

The fact remains that Intel is not pushing a higher performance product first to 'show off' 10nm, which is telling. It means they haven't got a real message for high performance to deliver. If they can only gain IPC and perf/watt by cutting back on clocks and the end result is a slower CPU, Sunny Cove might very well be a faster architecture, but the end result is still not satisfactory.

This is a fairly likely scenario for Intel's early 10nm. Not cutting back clocks but the process simply not clocking as far compared to 14nm++.

 

[efikkan]

https://www.techpowerup.com/forums/attachments/d755ed66-f0ea-4787-b9e4-5ce974c8a263-png.121784/
https://www.techpowerup.com/forums/attachments/219a9611-ac77-4695-bace-5de45021ef2f-png.121785/

These roadmaps, if genuine, are internal ones for Dell, which probably explains why it shows Coffee Lake-S Refresh starting in the middle of Q1 2019, even though some of the models launched earlier than that. One of the puzzling things in the roadmap is the Comet Lake Xeon-E launching Q1 2020 while Comet Lake-S (consumer) launching Q2 2020. Even if this is accurate, it doesn't preclude Comet Lake-S models that Dell doesn't use from launching ahead of that.
There are also at few other typos or misplacements in here, so I wold take some of these details with a grain of salt.

 

[Vayra86]

These roadmaps, if genuine, are internal ones for Dell, which probably explains why it shows Coffee Lake-S Refresh starting in the middle of Q1 2019, even though some of the models launched earlier than that. One of the puzzling things in the roadmap is the Comet Lake Xeon-E launching Q1 2020 while Comet Lake-S (consumer) launching Q2 2020. Even if this is accurate, it doesn't preclude Comet Lake-S models that Dell doesn't use from launching ahead of that.
There are also at few other typos or misplacements in here, so I wold take some of these details with a grain of salt.

Absolutely. Still though, this does tell us we won't be swimming in 10nm chips anytime soon, especially not for desktop, and for mobile we're talking about sub-top performing chips. Bottom line, for Intel's Core we're really going to be looking at that 14nm performance plateau for quite a while. AMD's got quite a good position for at least that long.

 

[londiste]

Still though, this does tell us we won't be swimming in 10nm chips anytime soon, especially not for desktop, and for mobile we're talking about sub-top performing chips. Bottom line, for Intel's Core we're really going to be looking at that 14nm performance plateau for quite a while. AMD's got quite a good position for at least that long.

Intel would not release something in any considerable volume if it did not make sense. Their primary motivation right now is to keep as much mobile market as they possibly can. Mobile CPUs are not limited by performance but by power - if they can get the 10nm to be efficient (which they should be able to, eventually) the clear intent is to shove off AMD from mobile. Intel rules mobile market today and mobile really has enormous inertia - AMD is struggling to get APUs in big manufacturers' laptops, the replacement cycle is longer and Intel still has upper hand there technically as well.

I suppose they are just giving up on straight wins on desktop. They are basically following the same strategy AMD has in number of occasions - screw the efficiency as long as performance is there. Whether this will work in this day and age remains to be seen.

 

[Vayra86]

Intel would not release something in any considerable volume if it did not make sense. Their primary motivation right now is to keep as much mobile market as they possibly can. Mobile CPUs are not limited by performance but by power - if they can get the 10nm to be efficient (which they should be able to, eventually) the clear intent is to shove off AMD from mobile. Intel rules mobile market today and mobile really has enormous inertia - AMD is struggling to get APUs in big manufacturers' laptops, the replacement cycle is longer and Intel still has upper hand there technically as well.

I suppose they are just giving up on straight wins on desktop. They are basically following the same strategy AMD has in number of occasions - screw the efficiency as long as performance is there. Whether this will work in this day and age remains to be seen.

100% the same view on this, yes. Its funny to see Intel being forced to follow the exact route AMD has in terms of that performance crown. I'm not even remotely worried about Intel surviving or the business sense behind this launch, either. And I also think AMD has some real work ahead of itself when it comes to mobile. Its just funny to see how the tables have turned, and how that happens so rapidly.

 

[lexluthermiester]

Steve at GN has some interesting input;