Intel Launches First 10th Gen Core Processors: Redefining the Next Era of Laptop Experiences

[notb]

Well, huge is a gross overstatement. Its a step forward, but if it doesn't really scale up that well (4c8t in 2020... come on)

It makes absolutely no sense to give it more cores at this time. They're limited by the power draw. They would have to drastically limit single-core performance and this CPU wouldn't make any sense (it would be an expensive i7 but perform like an Atom).

and if clocks are stuck at 4.1 Ghz or so we might as well rock our 7/8/9th gen chips a while longer. Because they'll be faster. A whole lot - and yes that goes for current day laptops too.

The early results show 10th gen being faster than 8th gen Whiskey Lake (let alone earlier). I don't understand what you mean.

Consider that Intel has not been capable of producing a single fart that even reeks of anything more than 4c8t and this meagre boost clock. The only upside is the higher base, but then for a laptop CPU, how much does thát tell us.

At this moment neither side makes a high-end 15/25W CPU with more than 4C, so I really don't get your problem. That's the state of technology today.

I highly doubt that 4k editing on IRIS graphics.

It's possible even on 8th gen mobile SoCs. I mean: it is happening. Really.

 

[TheLostSwede]

the perf/wat is probably good,better than anything out now.
I just don't understand "brings high performance to AI". What? A 12W CPU ?

"Intel Deep Learning Boost, a new, dedicated instruction set" as per the press release.
Some new Intel only CPU instructions that may or may not take off...

 

[Crackong]

1065G7: 15W/25W 4C8T 1.3 - 3.9GHz
8665U : 15W/25W 4C8T 1.9 - 4.8GHz

So Intel moving from 14+++ to 10nm causes 31.5% reduction in base freq. and 18.75% reduction in Max freq.

But AMD moving from 12nm to 7nm gains frequency.

How?

 

[1d10t]

That base clock though
Well if this for convertible than it'll make sense, but if they targeting casual audiences I find it really hard to combat (already) high base clock from Stoney Ridge.

 

[R-T-B]

Sorry, but this is becoming a pattern here at TPU..

Uh, yeah. It's been their policy since day 1 to post press releases.

I've never understood why they aren't in a separate feed though...

 

[FordGT90Concept]

I was literally looking at these yesterday trying to figure out if Intel had the capacity yet to fabricate high-transistor count 10nm processors yet. I was sorely disappointed to discover that they couldn't. Just lower power, nothing to get excited about, ridiculously low core count laptop and tablet processor offerings. Can't even manage six cores. At least these have GPUs where Cannon Lake (first iteration of 10nm) didn't. It's an improvement but still underwhelming.

2.3 GHz at best. Psssh!

 

[Space Lynx]

I was literally looking at these yesterday trying to figure out if Intel had the capacity yet to fabricate high-transistor count 10nm processors yet. I was sorely disappointed to discover that they couldn't. Just lower power, nothing to get excited about, ridiculously low core count laptop and tablet processor offerings. Can't even manage six cores. At least these have GPUs where Cannon Lake (first iteration of 10nm) didn't. It's an improvement but still underwhelming.

yeah I noticed the no 6 cores yet as well... and its almost 2020... lol it's so sad I almost feel bad for how bad Intel has been managed (for that specific market)

 

[FordGT90Concept]

Intel didn't just fall from their high horse, they got bucked off it and then squarely kicked in the head.

Yeah, I do feel bad for Intel just as I felt bad for AMD when they had terrible luck with die shrinks around 2006-2007 resulting in the formation of Global Foundries. Physics are relentless sometimes and it can't really be helped.

 

[THANATOS]

i7-8569U
CPU: 4C8T Base: 2.8Ghz (Turbo: 4.7Ghz) ; IGP: 48EU 1.2Ghz ; TDP: 28W

i7-8557U
CPU: 4C8T Base: 1.7Ghz (Turbo: 4.5Ghz) ; IGP: 48EU 1.15Ghz ; TDP: 15W

i7-1068G7
CPU: 4C8T Base: 2.3Ghz (Turbo: 4.1Ghz) ; IGP: 64EU 1.1Ghz ; TDP: 28W

i7-1065G7
CPU: 4C8T Base: 1.3Ghz (Turbo: 3.9Ghz) ; IGP: 64EU 1.1Ghz ; TDP: 15W

i7-1065G7 is in Cinebench R20 on par with i7-8559U both in single and multicore score. In Cinebench R11.5 and R15 is faster by ~11% in multi. Review
Ice Lake has higher IPC which is good news, but also lower clocks than It's predecessor and that's bad news. I don't think in desktop the situation would be any different.
I am pleased with the IGP's performance. It's true that the CPU and 3733Mhz memory helps the performance a lot, but I consider this as Intel's advantage over AMD, which only offers Zen+ cores paired with 10CU Vega and only 2400mhz memory instead of Zen 2 paired with 8-10CU Navi and faster memory support.
As an "APU" Intel wins in this round.

 

[aQi]

Instead of this PR crap, can we have please a proper presentation article, something like the other quality sites are doing? Reading that PR shit just makes the whole article not worth reading anymore.
Sorry, but this is becoming a pattern here at TPU...

Back to topic, looks like intel is doing his shit again providing a Core i7 with the same amount of Cores and Threads as an i5. This pos trend for laptops needs to stop!

Its pure marketing tactics rather then performance oriented brand modifier. Personally i am not impressed except the fact about connectivity and hardware level AI instruction. What i along side other are waiting is what intel has to do with desktop and HEDT parts. As laptop these days dont have much to do rather then essential tasks.

 

[W1zzard]

Instead of this PR crap, can we have please a proper presentation article, something like the other quality sites are doing? Reading that PR shit just makes the whole article not worth reading anymore.
Sorry, but this is becoming a pattern here at TPU...

Back to topic, looks like intel is doing his shit again providing a Core i7 with the same amount of Cores and Threads as an i5. This pos trend for laptops needs to stop!

I didn't hear anything from Intel at all on this launch, so that's all we can do. I suggest you reach out to Intel, tell them how awesome TPU is and that they should send us more samples.

 

[HTC]

1065G7: 15W/25W 4C8T 1.3 - 3.9GHz
8665U : 15W/25W 4C8T 1.9 - 4.8GHz

So Intel moving from 14+++ to 10nm causes 31.5% reduction in base freq. and 18.75% reduction in Max freq.

But AMD moving from 12nm to 7nm gains frequency.

How?

That's because moving from a super refined 14nm older process that can get very high clock speeds to a newer process that is nearing the limits of what is possible on a physical level gets hit ... hard ... with the realization that going with smaller nodes will cause, to some extent, reduced clocks and you'll need REAL improvements in the IPC department in order to counteract the clocks reduction under penalty of the newer chips not being an upgrade and being a downgrade instead.
Since Intel have been "lazy" in the improved IPC department, they were "caught off guard" by how much they were forced to reduce clocks and that had bad consequences ... for the 10nm node ...

AMD managed to not lose frequency due to their chiplet approach: AMD's Forrest Norrod explains it in this video.

 

[bug]

I didn't hear anything from Intel at all on this launch, so that's all we can do. I suggest you reach out to Intel, tell them how awesome TPU is and that they should send us more samples.

Except Intel didn't send samples at all this time. It was an invitation to test some reference design laptops within 8 hours at location of Intel's choosing.

 

[W1zzard]

Except Intel didn't send samples at all this time. It was an invitation to test some reference design laptops within 8 hours at location of Intel's choosing.

Ah ok, doesn't sound like the kind of thorough testing I'd usually do. And you can bet they hand picked the one sample that got reviewed

 

[bug]

Ah ok, doesn't sound like the kind of thorough testing I'd usually do.

They were rather flexible about it. For example AT's Ian was able to install his own benchmarks and run whatever he wanted. Of course, he did all that at his own expense and as such only got two-two and a half hours of actual testing. He also said that he wished for two-days testing so that he could sit on the results of the first day for a while before deciding on how to follow up.

And you can bet they hand picked the one sample that got reviewed

The sample didn't seem to be hand picked. It was just a development SKU (kind of what Qualcomm does for their Snapdragons). The only unrealistic part about it was the fan stuck at 100%. Plus, there wasn't much to hand pick, it's not like anyone tried to overclock that thing

Bottom line, that thing was as good or better than a similar Whiskey Lake part, while running at 500MHz less base and 700MHz less turbo. Not too shabby, eh?
Of course, the question remains about the attainability of higher clocks. Because while those clocks may be fine on a laptop, they won't fly on a desktop. At all.

 

[efikkan]

Bottom line, that thing was as good or better than a similar Whiskey Lake part, while running at 500MHz less base and 700MHz less turbo. Not too shabby, eh?
Of course, the question remains about the attainability of higher clocks. Because while those clocks may be fine on a laptop, they won't fly on a desktop. At all.

This should give some hints about the potential of this architecture.
There have been several cases in the past where Intel or AMD have "regressed" in clock speed, but each time they have advanced in performance.

We shouldn't dismiss Ice Lake/Sunny Cove as an architecture based only on seeing the least interesting core configurations. While these lighter laptops certainly will benefit from any performance gain, the Y/U series will continue to perform poorly for the foreseeable future. Chips in this class perform very unreliably, and can only run "good clocks" in short bursts. It will get more interesting when we see server/workstation chips.

 

[bug]

This should give some hints about the potential of this architecture.
There have been several cases in the past where Intel or AMD have "regressed" in clock speed, but each time they have advanced in performance.

We shouldn't dismiss Ice Lake/Sunny Cove as an architecture based only on seeing the least interesting core configurations. While these lighter laptops certainly will benefit from any performance gain, the Y/U series will continue to perform poorly for the foreseeable future. Chips in this class perform very unreliably, and can only run "good clocks" in short bursts. It will get more interesting when we see server/workstation chips.

Worst case scenario, 10nm goes bust and Intel builds these on 7nm in a few more years. So yeah, the numbers are still interesting.

 

[john_]

Do people really believe that OEMs will put 3733MHz RAM in their laptops and not something much slower and of course also cheaper?

 

[bug]

Do people really believe that OEMs will put 3733MHz RAM in their laptops and not something much slower and of course also cheaper?

Usually RAM speed is not a problem. The problem is laptops that come with only one RAM slot populated.
But nice red herring

 

[john_]

Usually RAM speed is not a problem. The problem is laptops that come with only one RAM slot populated.
But nice red herring

In fact if you have only one dimm, the frequency becomes even more important.

 

[notb]

1065G7: 15W/25W 4C8T 1.3 - 3.9GHz
8665U : 15W/25W 4C8T 1.9 - 4.8GHz

So Intel moving from 14+++ to 10nm causes 31.5% reduction in base freq. and 18.75% reduction in Max freq.

But AMD moving from 12nm to 7nm gains frequency.

How?

Intel's starting point was much more polished than AMD's. That's about it.

5650U - the first Intel's 14nm mobile flagship - was a 2.2/3.2 GHz. In 3 years they've managed to squeeze 1.9/4.8 GHz (doubling the core count as well).

Remember this is a 15W mobile CPU. Low base clocks and high boost is what you want - not the other way around.

And yes... it's also a new arch with new instructions. Big jump, potentially great performance and battery life. And another 3-4 years of mobile dominance while AMD finds new ways to squeeze more cores in desktop CPUs.

In fact if you have only one dimm, the frequency becomes even more important.

It's LPDDR4X. AFAIK the standard is 3200+, but focus is on the top-end 4266.
You say RAM frequency is important and - instead of cheering very good RAM support on these CPUs - you're bickering that laptop makers we'll surely use something bad. What's your problem?

 

[Crackong]

Intel's starting point was much more polished than AMD's. That's about it.
5650U - the first Intel's 14nm mobile flagship - was a 2.2/3.2 GHz. In 3 years they've managed to squeeze 1.9/4.8 GHz (doubling the core count as well).
Remember this is a 15W mobile CPU. Low base clocks and high boost is what you want - not the other way around.
And yes... it's also a new arch with new instructions. Big jump, potentially great performance and battery life. And another 3-4 years of mobile dominance while AMD finds new ways to squeeze more cores in desktop CPUs.

My question was simple.
Intel managed to have a 14nm+++++ 15W chip doing 1.9 GHz base freq but now the new 10nm 15W chip needs to start at 1.3 GHz.
The IPC boost is 18% claimed by Intel, but with this 31.5% reduction in base freq I am not so sure............
Isn't the smaller node supposed to save energy, so they should squeeze more freq out with the same TDP, not the other way around?

 

[efikkan]

My question was simple.
Intel managed to have a 14nm+++++ 15W chip doing 1.9 GHz base freq but now the new 10nm 15W chip needs to start at 1.3 GHz.
The IPC boost is 18% claimed by Intel, but with this 31.5% reduction in base freq I am not so sure............
Isn't the smaller node supposed to save energy, so they should squeeze more freq out with the same TDP, not the other way around?

A lot of this is spent on a larger integrated GPU and to some extent AVX512.

 

[notb]

My question was simple.
Intel managed to have a 14nm+++++ 15W chip doing 1.9 GHz base freq but now the new 10nm 15W chip needs to start at 1.3 GHz.

And the answer is simple as well. It's a mobile CPU. Lower frequency, if followed by lower idle power consumption, is an advantage.
CPU boosts when it has something to do. And early tests show this one is faster than earlier generations.

I also gave you the example of the 5th gen 5650U for a reason. 8th gen mobile CPUs have lower base clocks - despite being built on a more modern 14nm. And they are faster (when speed is needed) and a lot more frugal (when it isn't).

The IPC boost is 18% claimed by Intel, but with this 31.5% reduction in base freq I am not so sure............

Looking at base frequencies makes little sense. CPUs boost differently. You should only look at resulting performance.
Base clocks on these CPUs are enough for supporting the OS and showing 2D graphics - like when you code or edit a document or read something on a website. That's why laptops work for 15h today - not because they use less power under heavy load, but because they're more frugal in idle/light usage. But even basic tasks, like opening a new page, will make them shift to a higher state.

 

[Vayra86]

Looking at base frequencies makes little sense. CPUs boost differently. You should only look at resulting performance.
Base clocks on these CPUs are enough for supporting the OS and showing 2D graphics - like when you code or edit a document or read something on a website. That's why laptops work for 15h today - not because they use less power under heavy load, but because they're more frugal in idle/light usage. But even basic tasks, like opening a new page, will make them shift to a higher state.

That really only flies for a selection of notebooks and often only the better/best/priciest ones. In many cases, what a high turbo means is that it will throttle like nobody's business in any half-serious use case and in all others, its practically idling and you lose any and all performance to do anything. The real question here is whether a baby step (because really, given all those factors that influence performance in a regular use case, that is what this is even with the constant shifting of base/turbo clocks) like this, with respect to the lowered base clocks, is really even worth mentioning.

I mean yes, in their very narrow use case, these 'U' parts shine. But their use case is doing as little as possible. If you're even half serious about even a little productivity, you avoid this line.

So TL DR I don't believe Intel has a very competitive part here because they managed to tweak things a bit. This 'new arch' is only worth a damn if it can scale to high performance parts. And I'm entirely with @bug in the thought they may scrap it altogether and feel forced to go 7nm after all for anything more than this PoC we've seen right now.

Looking at base frequencies makes little sense. CPUs boost differently. You should only look at resulting performance.

Correct, and the resulting performance really is just more of the same, in the end. So your battery may last 20 minutes longer, wooptiedoo That's not going to change a thing.