Intel Rocket Lake CPUs Will Bring up to 10% IPC Improvement and 5 GHz Clocks

[AleksandarK]

Intel is struggling with its node development and it looks like next-generation consumer systems are going to be stuck on 14 nm for a bit more. Preparing for that, Intel will finally break free from Skylake-based architectures and launch something new. The replacement for the current Comet Lake generation is set to be called Rocket Lake and today we have obtained some more information about it. Thanks to popular hardware leaker rogame (_rogame), we know a few stuff about Rocket Lake. Starting off, it is known that Rocket Lake features the backport of 10 nm Willow Cove core, called Cypress Cove. That Cypress Cove is supposed to bring only 10% IPC improvements, according to the latest rumors.

With 10% IPC improvement the company will at least offer some more competitive product than it currently does, however, that should be much slower than 10 nm Tiger Lake processors which feature the original Willow Cove design. It shows that backporting of the design doesn't just bring loses of the node benefits like smaller design and less heat, but rather means that only a fraction of the performance can be extracted. Another point that rogame made is that Rocket Lake will run up to 5 GHz in boost, and it will run hot, which is expected.


View at TechPowerUp Main Site

 

[Mamya3084]

Intel will also commit to a 10% order fulfillment

 

[watzupken]

If this is 10% IPC uplift from Sky Lake, then it looks really poor. Even the Sunny Cove used in Ice Lake U was supposed to be up to 18% IPC uplift from Sky Lake based processors. If AMD's Zen 3 introduces up to 15% IPC improvement, then its all over for Intel when it comes to performance and efficiency. To hit the 5Ghz, I am sure they are feeding it with the same insane amount of power just like the current Comet Lake.

 

[efikkan]

I have no idea whether the IPC gains are 10%, 20% or even 25%, but 10% would certainly be disappointing.

But approximating IPC on early engineering samples is pointless, since we don't have context and control over the chip's clocks. I suspect this "10% estimate" is derived from that Geekbench score from the other day, a benchmark which is useless even on a good day.

 

[Kohl Baas]

If this is 10% IPC uplift from Sky Lake, then it looks really poor. Even the Sunny Cove used in Ice Lake U was supposed to be up to 18% IPC uplift from Sky Lake based processors. If AMD's Zen 3 introduces up to 15% IPC improvement, then its all over for Intel when it comes to performance and efficiency. To hit the 5Ghz, I am sure they are feeding it with the same insane amount of power just like the current Comet Lake.

For now.

Never forget we're speaking about a 10 times bigger company having some hard times because 5-10 years of lacking competition, emerging security issues and a 5 year saga of strugling 10nm R&D. They will make a comeback. It's inevitable. They just need time to do that. They've past the point of salvability with 10nm, focusing on the next step which may be later than expected, but anoder point worth remembering: Intel's manufacturing processes were always better. That means the just dropped 10nm would be on par with TSMC/Samsung 7nm. Which is the cause the industry is urging to work out a new method of measurement for manufacturing techniques because the "nm" count is no longer a valid measurement of comparison.

 

[Caring1]

5GHz boost clocks, who cares.
Increased TDP makes these pointless.

 

[bug]

At this point, I think Intel should never move off 14nm. They should just keep it around as a "vintage edition" or something like that.

 

[watzupken]

For now.

Never forget we're speaking about a 10 times bigger company having some hard times because 5-10 years of lacking competition, emerging security issues and a 5 year saga of strugling 10nm R&D. They will make a comeback. It's inevitable. They just need time to do that. They've past the point of salvability with 10nm, focusing on the next step which may be later than expected, but anoder point worth remembering: Intel's manufacturing processes were always better. That means the just dropped 10nm would be on par with TSMC/Samsung 7nm. Which is the cause the industry is urging to work out a new method of measurement for manufacturing techniques because the "nm" count is no longer a valid measurement of comparison.

I don't deny they are a much bigger company, but you need to remember that AMD Is not Intel's only competitor when you are looking at just CPU business. Objectively speaking, Intel have had multiple missteps for a number of years now. If not because of their financial might and market share, they would have been in bigger trouble.

I am no expert in this area, but what I understand is that Intel's fab tend to be superior to competition in terms of density. So their 14nm for example, could equal that of a 10nm from their competitor for example. But starting with 10nm, I am starting to wonder if that is still the case. What Intel originally planned for 10nm, it was well and good (and aggressive). But because they failed to deliver it for many years, I am pretty sure they had to make compromises to deliver it and appease their investors. My take is that if Intel's 10nm is that great, why is AMD able to deliver and 8 core Renoir with a respectable iGPU, while TIger Lake can only cram in 4 cores with Xe graphics? This is just my speculation. In addition, 10nm yields may have improved, but still unlikely to be as good as their matured 14nm. I feel Intel is certainly not out of the woods with their 10nm issues, thus, we are seeing them dragging their feet when it comes to delivery of 10nm products.

As to 7nm, I will not hold my breathe for it. Its about 2+ to 3 years away and competitors are not sitting there idle. Whether it will be better or not, time will tell. But 1 thing I feel is certain, and that is Intel has lost the fab advantage and likely not going to get it back that soon. Just hope that they don't fall behind on their 7nm further because that will be devastating news for them.

5GHz boost clocks, who cares.
Increased TDP makes these pointless.

Objectively, I feel some people won't be put off by the high TDP. I am just waiting to see if Zen 3 is going to give AMD the boost required in games. Currently, high clockspeed from Intel processors are helping them to retain the gaming crown. But with a lackluster 10% IPC improvement it may not be a big step up from the current highest clocked i7 10700K Coment Lake CPU.

 

[bug]

Objectively, I feel some people won't be put off by the high TDP. I am just waiting to see if Zen 3 is going to give AMD the boost required in games. Currently, high clockspeed from Intel processors are helping them to retain the gaming crown. But with a lackluster 10% IPC improvement it may not be a big step up from the current highest clocked i7 10700K Coment Lake CPU.

TDP is rather hard to understand these days. Basically you get a base TDP and base frequencies that the CPU will sustain indefinitely. You get a boost TDP which is a max value the CPU will push, thermals allowing. And in practice you get the real boost TDP which is determined by your cooling solution and airflow (because most people will not have the blowers required to reach the max boost TDP).

Personally, I feel that while Intel could have done a better job explaining all of that better to the layman, I feel what all that does in the end is let a CPU stretch as much as it has legroom to do so. But with that in mind, it's pretty hard to predict what people will find acceptable and what not.

 

[kiriakost]

Objectively speaking, Intel have had multiple missteps for a number of years now. If not because of their financial might and market share, they would have been in bigger trouble.

I am no expert in this area, but what I understand is that Intel's fab tend to be superior to competition in terms of density. So their 14nm for example, could equal that of a 10nm from their competitor for example.

Since 1998 and later and for every release of a New CPU series, INTEL this has to generate enough funds ( from us consumers) so to pay the R&D work and also to deliver profits to their shareholders.
From the other hand there is a financial crisis in the markets in the past ten years, consumers spending this gradually minimized.

In simple English, the Industry move forwards when past investment at R&D and shareholders has be paid, only then a new cycle will begin.
From the era of quad core Q6600 and later, the vast majority of CPU intense and productive applications has be covered for home use.

For example the people and industry they had the expectation of getting a faster CPU for video editing and or 3D product design software.
New CPU series development and production just for gaming this is not how INTEL thinks or act.

Statistically the spending capacity of a PC gamer this is zero ... the father is the hero and the one paying the bills.
When all households are in crisis, INTEL economists they have awareness that there is no much to expect as growth from them.

And as last, lets not leave out the Microsoft Windows deception (operating system), since Win7 and later the operating system become full of processes running simultaneously with out user control, supposedly these are a new form of software based intelligence, this collecting and recording all day long what the user does in his PC.
Such a poor made concept it does keep the CPU and hard-drives busy even when the PC hypothetically this is idle, since Win7 and later there is no CPU Idle time.
If you keep quad cores busy just for the operating system, then you need another quad cores solely for intense gaming.

In conclusion now that your CPU this has eight cores , you do still consider it as slow.
I have the feeling that game consoles will do better with less required hardware simply because they will not run Windows as operating system.

 

[Bruno Vieira]

I have no idea whether the IPC gains are 10%, 20% or even 25%, but 10% would certainly be disappointing.

But approximating IPC on early engineering samples is pointless, since we don't have context and control over the chip's clocks. I suspect this "10% estimate" is derived from that Geekbench score from the other day, a benchmark which is useless even on a good day.

These chips are to be released this year, this is not early, probably semi-final silicon and final clocks. With 10% IPC and only 8 cores Intel will not even win gaming benchmarks against zen3.

 

[JustAnEngineer]

Intel has made TDP a joke. Their "125W TDP" CPUs are pulling 280-300 watts continuously if you have enough cooling to handle it.

 

[efikkan]

Intel has made TDP a joke. Their "125W TDP" CPUs are pulling 280-300 watts continuously if you have enough cooling to handle it.

That's incorrect.
At stock, Intel's CPUs will throttle to the TDP after 28-56 seconds, unless you disable the power limit. While I'm not a fan of this burst speed, if you're disabling the power limit to do it continuously you're no longer running it at stock, and that's on you. Unfortunately, many reviews do this, which makes these CPUs look terrible in terms of power consumption.

For AMD on the other hand, TDP is only a "guidance", where you can draw a bit more with sufficient cooling.

 

[Hexa]

im really hoping they will be good, Intel has showed us what happens when there is no competition in the market and if intel wont be able to offer good cpu´s at reasonable prices, zen3 prices will be high aswell

 

[bug]

im really hoping they will be good,

Based on?

Intel has showed us what happens when there is no competition in the market

If by that you mean we get progress in other areas (i.e. laptops/ultrabooks), then you're spot-on.

 

[AnarchoPrimitiv]

Intel will also commit to a 10% order fulfillment

Hahaha, that really made me laugh.


Anyway, hasn't it been reported that Zen3 has 17% IPC uplift at worst? Also, isn't AMD doubling the cores per CCX? I'm not sure how that translates to performance, but I do know from the Techpowerup review of the 3100 and the 3300x that TPU reported a 12% performance advantage on average for the 3300x because of the 3300x have a single CCX versus two like with the 3100. Additionally, I've also read that we should expect a 200-300Mhz boost in frequencies moving to the upgraded 7nm node. I'm not sure how the math works in this situation (as in I'm not sure if it's as easy as adding those percentages together) , but wouldn't that mean that core for core, we can expect a 25% total performance increase for Zen3 at least?

 

[Houd.ini]

[...]we can expect a 25% total performance increase for Zen3 at least?

At least? That is really hoping for a lot, I would not set such high expectations.

 

[Renald]

If true, it's around a 3 generation upgrade in Intel world

They might get a bit close to Zen 2, but at which price and power usage ?
And I bet that there will be a black market because of shortage...

 

[Steevo]

5Ghz for how long and how many cores? 10% IPC with more cache, compared to....... Cache makes heat, so the two articles should have been merged, cause it shows a lower muticore performance, caused by high TDP and temps....

 

[R0H1T]

If by that you mean we get progress in other areas (i.e. laptops/ultrabooks),

I'd give Apple as much credit if not more than Intel for that category & of course going for better battery life.

 

[techguymaxc]

10% IPC improvement after at least 5 years is nowhere near enough. Intel promised 18% average IPC uplift when Sunny Cove was detailed back in 2018.

 

[EarthDog]

10% IPC improvement after at least 5 years is nowhere near enough. Intel promised 18% average IPC uplift when Sunny Cove was detailed back in 2018.

What does enough mean? This takes the ipc crown back from amd (until zen3) and they are clocked a lot higher.

We all want more, and expect more... but this isn't a die shrink which brings with it even mkre native improvements. Understand what they are working with (regardless of how they got here).

 

[Tomgang]

10 % IPC improvement. Hmm is it not around that mark we have been the last decade or so. Since after Hassell. The IPC gains has been 5-10 % and not really more than that.

Sure Intel's reaching high core clock, but at what cost. High temperature and power consumption = expensive cooling solution needed as well.

No so far I am not impressed by Intel's offerings. So far I have my eyes set on AMD's Zen 3.

 

[londiste]

10% IPC improvement after at least 5 years is nowhere near enough. Intel promised 18% average IPC uplift when Sunny Cove was detailed back in 2018.

Sunny Cove did bring that ~18% increase. Which makes the tweet this snippet is based on strange - Willow Cove only +10% over Skylake? How? Willow Cove is worse than Sunny Cove which was previous new core architecture?

 

[bug]

I'd give Apple as much credit if not more than Intel for that category & of course going for better battery life.

Why? You think Apple designed those chips for Intel?

My point was, while the desktop didn't see much action because of lack of competition, Intel hasn't simply been sitting on their asses cashing in from Core.