Friday, July 24th 2020

Intel 7nm CPUs Delayed by a Year, Alder Lake in 2H-2021, Other Commentary from Intel Management

Intel's silicon fabrication woes refuse to torment the company's product roadmaps, with the company disclosing in its Q2-2020 financial results release that the company's first CPUs built on the 7 nanometer silicon fabrication node are delayed by a year due to a further 6-month delay from prior expectations. The company will focus on getting its 10 nm node up to scale in the meantime.

The company mentioned that the 10 nm "Tiger Lake" mobile processor and "Ice Lake-SP" enterprise processor remains on-track for 2020. The company's 12th Generation Core "Alder Lake-S" desktop processors won't arrive before the second half of 2021. In the meantime, Intel will launch its 11th Gen Core "Rocket Lake" processor on the 14 nm node, but with increased IPC from the new "Cypress Cove" CPU cores. Also in 2H-2021, the company will launch its "Sapphire Rapids" enterprise processors that come with next-gen connectivity and updated CPU cores.
Intel 7 nanometer delay
It's interesting to note that Intel was specific about "CPU" when talking about 7 nm, meaning that Intel's foundry woes only affect its CPU product stack, and not a word was mentioned in the release about the company's discrete GPU and scalar compute processors that are being prototyped and validated. This is probably the biggest hint we'll ever get from Intel that the company's dGPUs are being designed for third-party foundries (such as Samsung or TSMC), and that the Xe dGPU product roadmap is disconnected from that of Intel's fabs.
Intel is accelerating its transition to 10 nm products this year with increasing volumes and strong demand for an expanding line up. This includes a growing portfolio of 10 nm-based Intel Core processors with "Tiger Lake" launching soon, and the first 10 nm-based server CPU "Ice Lake," which remains planned for the end of this year. In the second half of 2021, Intel expects to deliver a new line of client CPU's (code-named "Alder Lake"), which will include its first 10 nm-based desktop CPU, and a new 10 nm-based server CPU (code-named "Sapphire Rapids"). The company's 7 nm-based CPU product timing is shifting approximately six months relative to prior expectations. The primary driver is the yield of Intel's 7 nm process, which based on recent data, is now trending approximately twelve months behind the company's internal target.
Intel's post results call also revealed a handful interesting tentative dates. For starters, "Tiger Lake" is shipping in "a matter of weeks," indicating an imminent launch ahead of the "Back to School" shopping season. Next up, the company's high-performance scalar compute processor, codenamed "Ponte Vecchio" remains slated for 2021-22, and given that it's reportedly being designed for 7 nm, we have our next big hint confirmation that these dGPUs will be built on third-party 7 nm fabs. Intel did mention that the Foveros packaging technology could be further developed over the years, and its upcoming discrete GPUs could combine dies (tiles) from multiple sources, which could include its own fabs.

Given the delays in Intel's 7 nm foundry node, the first Intel client-segment processors based on the node won't arrive before late-2022 or 2023, which means refinements of the current 10 nm silicon fabrication node should support Intel's client-segment product stack for the foreseeable future. The first enterprise 7 nm processors will arrive by the first half of 2023. Intel also mentioned that they expect to see "one full node improvement" from a refined 10 nanometer process, which isn't surprising, given how much experience they have improving their 14 nanometer process.
Add your own comment

175 Comments on Intel 7nm CPUs Delayed by a Year, Alder Lake in 2H-2021, Other Commentary from Intel Management

#151
londiste
I just tried that out of curiosity. After ~15 seconds the CPU usage goes down to 2-3% which is negligible. Playing 9 different videos on Youtube peaks higher but again after ~15 seconds CPU usage goes down to ~17% for a quite unrealistic use case.
Posted on Reply
#152
Vayra86
EarthDog
Yeah, MS Office is a bitch to run, eh? Same with default services. THANK GOD AMD is here with all their cores and threads. o_O

Jokes aside, in any kind of enterprise or office environment updates are rolled out after hours 99% of the time.... not while users are sitting in front of their PC.

That said, it really depends on use. CAD designers, content creators etc, likely run a higher load than most general office workers (accounting, HR, CSRs, etc). But typical office functionality isn't much on a PC. There will always be exceptions.
The bottleneck is the intranet services really, not so much the workstations themselves. Although if you do anything more serious than basic administrative jobs, you'll definitely get load fom Office alongside all of its cloud based stuff. I can choke my lappy with Word and Excel, no problem whatsoever. Add a few tables and databases to Excel and it gets real fun, too. There are word docs we do edits on that are so vast Word has trouble processing the thing. Its not so much absolute performance all the time, as it is simultaneous processes running.

I've managed to destroy one laptop fan so far from all the heavy duty, took less than a year, but maybe I was lucky.
Posted on Reply
#153
EarthDog
ARF
This is what my CPU does when I open YouTube in 9 new tabs.
Neat! Not one core even spiked to 100% when you did that and that (unrealistic) use scenario. So, what's your point?
londiste
I just tried that out of curiosity. After ~15 seconds the CPU usage goes down to 2-3% which is negligible. Playing 9 different videos on Youtube peaks higher but again after ~15 seconds CPU usage goes down to ~17% for a quite unrealistic use case.
Similar story... I really don't know what he's on about. :(
Vayra86
The bottleneck is the intranet services really, not so much the workstations themselves. Although if you do anything more serious than basic administrative jobs, you'll definitely get load fom Office alongside all of its cloud based stuff. I can choke my lappy with Word and Excel, no problem whatsoever. Add a few tables and databases to Excel and it gets real fun, too. There are word docs we do edits on that are so vast Word has trouble processing the thing. Its not so much absolute performance all the time, as it is simultaneous processes running.

I've managed to destroy one laptop fan so far from all the heavy duty, took less than a year, but maybe I was lucky.
Yeah... I run some pivot tables/data bases, etc... but nothing too heavy. All I am saying is that, for the most part, office PC's are not high utilization machines... in particular if the company keeps up on product life cycle. Running a dually with HT in 2020 even for an office box is a travesty. :p
Posted on Reply
#154
ARF
EarthDog
Neat! Not one core even spiked to 100% when you did that and that (unrealistic) use scenario. So, what's your point?
People sometimes open dozens of tabs and keep them running. Imagine if some of them keep running flash videos and other stuff in the background.
Your CPU goes to 100% with ease.
Not to mention that your GPU also participates and is heavily loaded.

Don't underestimate the browsing. I was quite mild with only 9 open tabs. Do you know what happens with 29 open tabs ?

You need a powerful CPU to cope with them.
Posted on Reply
#155
EarthDog
ARF
People sometimes open dozens of tabs and keep them running. Imagine if some of them keep running flash videos and other stuff in the background.
Your CPU goes to 100% with ease.
Not to mention that your GPU also participates and is heavily loaded.

Don't underestimate the browsing. I was quite mild with only 9 open tabs. Do you know what happens with 29 open tabs ?

You need a powerful CPU to cope with them.
lol, ARF... come on man... let's not let one offs and poor PC management dictate the majority of office use cases. Don't be a dolt (not you) and forget to close your shiza on an office PC. lol

I had over 30+ tabs open at once (said that earlier...), I know exactly what happens (described that earlier). I'm also not a brick and don't leave 29 flash/yt vids running at the same time... o_O

We'll agree to disagree. I'm moving on. :)

I'll leave this here.
EarthDog
That said, it really depends on use. CAD designers, content creators etc, likely run a higher load than most general office workers (accounting, HR, CSRs, etc). But typical office functionality isn't much on a PC. There will always be exceptions.
Posted on Reply
#156
Chrispy_
EarthDog
That said, it really depends on use. CAD designers, content creators etc, likely run a higher load than most general office workers (accounting, HR, CSRs, etc). But typical office functionality isn't much on a PC. There will always be exceptions.
That's my industry.

An alarming amount of that software is single-threaded, poorly-optimised legacy junk. Outside of Premiere and software rendering (which is rapidly being replaced by GPU rendering) you can work just fine on an old Core2 Duo, as long as you have enough RAM.

There are some legacy 10+ year-old machines hooked up to CNC mills and laser-cutting beds with either Core2 or we've replaced some of them with low-end Bay-Trail or Apollo-Lake NUCs and with 16GB RAM to hold the models in memory they are perfectly-snappy, 100% capable CAD modelling platforms that can multitask Office suites and Web browsers in the background.

CAD software is old and simple. Parametric real-time simulations are where the CPU demands in my CAD/AEC industry are and once again, it's cores, cores, and more cores to feed those which is why anyone without a 3900X at their desk just offloads that to the Threadripper/3950X farm (or the old Broadwell-E farm if they've ave angered the compute department).
Posted on Reply
#157
Vayra86
ARF
People sometimes open dozens of tabs and keep them running. Imagine if some of them keep running flash videos and other stuff in the background.
Your CPU goes to 100% with ease.
Not to mention that your GPU also participates and is heavily loaded.

Don't underestimate the browsing. I was quite mild with only 9 open tabs. Do you know what happens with 29 open tabs ?

You need a powerful CPU to cope with them.
29 Tabs is a sign of madness, that is what that is. I know them, I've seen it. But even so, the PC will not choke on CPU load, it will choke on RAM.
Posted on Reply
#158
cat1092
Intel has been 'delaying' 10nm for years, while the AMD FX series CPU's & a wide variety of 970 chipset MB's were still being sold on Newegg. :D

Then AMD comes in & goes to work & now most all of their mainstream chips are 7nm. Be it AMD or NVIDIA, they've both came in & claimed Moore's Law as their own. It would seem, with all of Intel's massive resources, plus the cash saved by swapping cheap thermal paste in place of quality solder (which most consumers would pay for), they'd have 7nm chips to compete.

Intel became too laid back for their own good when they thought AMD was dead. Now they've lost Apple, haven't met their goals for years (they've has 14nm since Broadwell) & are nowhere close to putting a chip on the market which boots at 5.0 GHz. By now, Intel should be there, were the first to break the 4.0 GHz barrier with the i7-4790K, and other than adding more cores/threads, have done little more. Why I've not seen the need to upgrade my Z97 system on that mentioned i7-4790K, although have done part of their work in delidding & replacement with liquid metal. AMD has steadily soldered their mainstream chips for as far back as I can remember, this decreases heat & in most instances, eliminates the need for an expensive cooler.

Don't count on Intel having 7nm chips across the board for another 5 years at least!:clap:
Posted on Reply
#159
RandallFlagg
cat1092
Then AMD comes in & goes to work & now most all of their mainstream chips are 7nm.
You mean TSMC, right? AMD does not fab chips.
cat1092
Be it AMD or NVIDIA, they've both came in & claimed Moore's Law as their own. It would seem, with all of Intel's massive resources, plus the cash saved by swapping cheap thermal paste in place of quality solder (which most consumers would pay for), they'd have 7nm chips to compete.
Intel's 10nm is equivalent to TSMCs 2nd/3rd gen 7nm and better than TSMC 1st gen 7nm. It is also better than Samsung 7nm.
cat1092
Don't count on Intel having 7nm chips across the board for another 5 years at least!:clap:
As noted, Intel 10nm is already equivalent to TSMC 7nm.

We will not have AMD 5nm chips until Zen 4, which means maybe late 2021. Zen 3 will be 7nm+, which as noted is about the same as Intel 10nm.

Don't be a sucker for the marketing gimmickry of process node names which are not comparable with each other.
Posted on Reply
#160
Vayra86
RandallFlagg
You mean TSMC, right? AMD does not fab chips.



Intel's 10nm is equivalent to TSMCs 2nd/3rd gen 7nm and better than TSMC 1st gen 7nm. It is also better than Samsung 7nm.




As noted, Intel 10nm is already equivalent to TSMC 7nm.

We will not have AMD 5nm chips until Zen 4, which means maybe late 2021. Zen 3 will be 7nm+, which as noted is about the same as Intel 10nm.

Don't be a sucker for the marketing gimmickry of process node names which are not comparable with each other.
What are you even talking about, all we saw of Intel's 10nm is low performance chip material, so even if they might have efficiency and density perks, they still won't scale to a high performance setup and 7nm clearly does. Why do you think that Intel's entire high performance segment (in fact anything faster than a quadcore laptop CPU, basically) is still stuck on 14nm?

TSMC doesn't need gimmicking, Intel does, and they're going full steam on it. This 10nm fairy tale is one of the examples. It was true once upon a time when nobody had anything on the market. But these things have changed faster than Intel could blink. Intel's proposed higher density on 10nm is exactly the thing that killed their roadmaps and they could have seen it coming as they jumped from one heatspreader solution to the next to keep things in check. TSMC has a much better node in their hands.
Posted on Reply
#161
RandallFlagg
Vayra86
What are you even talking about, all we saw of Intel's 10nm is low performance chip material, so even if they might have efficiency and density perks, they still won't scale to a high performance setup and 7nm clearly does. Why do you think that Intel's entire high performance segment (in fact anything faster than a quadcore laptop CPU, basically) is still stuck on 14nm?
...
Read the post I was responding to and enlighten yourself.

As far as performance, don't be too cocky.

Here's a picture of a 4C/8T 2.8Ghz Tiger Lake 10nm CPU, and circled is an 8C/8T 4.1Ghz AMD Zen 2 getting 7% lower in multi-threaded Time Spy CPU benchmark.

i.e. Tiger Lake 4C/8T beats 8C/8T AMD.

Even the desktop 3300X 4C/8T 3.8-4.3 Ghz part is only 9% faster, despite 27% higher base clock and 35% higher turbo, not to mention being a 65W part compared to the Intel's 15W part.

The moment this type of increased IPC hits the laptop space in anything 6C or higher, AMD is done in that space. The laptop market is much bigger than desktops. I expect this to happen late Q3 and Q4 this year.

It's entirely possible that Intel's high end laptops will be faster than all but the highest end desktop parts - AMD or Intel.

Posted on Reply
#162
mtcn77
RandallFlagg
Intel's 10nm is equivalent to TSMCs 2nd/3rd gen 7nm and better than TSMC 1st gen 7nm. It is also better than Samsung 7nm.
This inflection point is critical because it totally distinguishes the fundamental differences of Intel and Samsung.
Samsung builds processes on previous foundational IPs. Intel aimed a new process both at foundry and microarchitecture leaps without validating one on the other, first.
Posted on Reply
#163
RandallFlagg
mtcn77
This inflection point is critical because it totally distinguishes the fundamental differences of Intel and Samsung.
Samsung builds processes on previous foundational IPs. Intel aimed a new process both at foundry and microarchitecture leaps without validating one on the other, first.
Don't know about that, but that isn't how the naming conventions between Intel in the foundries got out of sync.

They got out of sync back when the foundries started going from 20nm to 12/14/16nm. You see, when Intel went from 22nm to 14nm, they also added FinFet. They called it one node though, 22nm->14nm.

When the foundries (TSMC, Samsung, GloFlo) took their 20nm nodes and added FinFet, they started calling it 16nm/14nm/12nm even though they did not really shrink the node.

This is why Intel's 14nm process node is actually roughly equivalent to the foundries 10/12/14nm nodes.

And their 10nm is roughly equivalent to foundries 7nm nodes.
Posted on Reply
#164
mtcn77
That's all nice and dandy, but doesn't play crysis when push meets the shove.
Posted on Reply
#165
Vayra86
RandallFlagg
Read the post I was responding to and enlighten yourself.

As far as performance, don't be too cocky.

Here's a picture of a 4C/8T 2.8Ghz Tiger Lake 10nm CPU, and circled is an 8C/8T 4.1Ghz AMD Zen 2 getting 7% lower in multi-threaded Time Spy CPU benchmark.

i.e. Tiger Lake 4C/8T beats 8C/8T AMD.

Even the desktop 3300X 4C/8T 3.8-4.3 Ghz part is only 9% faster, despite 27% higher base clock and 35% higher turbo, not to mention being a 65W part compared to the Intel's 15W part.

The moment this type of increased IPC hits the laptop space in anything 6C or higher, AMD is done in that space. The laptop market is much bigger than desktops. I expect this to happen late Q3 and Q4 this year.

It's entirely possible that Intel's high end laptops will be faster than all but the highest end desktop parts - AMD or Intel.


Yeah, if you think laptops are somehow not built with limited TDPs, Tiger Lake looks amazing. When you let that chip fully stretch its legs... it can do with 2.8 Ghz what Skylake could do with 4.0. But it still caps at that eternal 4c8t Skylake perf level. That is the point. Now look at the TDP of the 202% faster (!!) 10900K. Change the desktop? The proof is right there in the chart that it won't.

Now, let's get back to reality. I get your point, but this won't change a thing in anything that is not power limited. Its easy to get lost in the woods of Intel benchmarking logic.
RandallFlagg
And their 10nm is roughly equivalent to foundries 7nm nodes.
Its not wise to keep repeating that because it underlines you missed the point entirely about power budgets and what high performance entails. Its like it went straight past you, but its not a lie or anything...

Even investors and Intel itself have lost faith in that node, and here you are ;)

One last point, IPC is an architecture trait, not a node trait. They could obtain equal IPC on 14nm with the same core on a different node, but the power usage would change. In the same vein we saw planned 10nm architectural updates rolled back to 14nm. Most notably anything that had to have high clocks and ditto performance.
Posted on Reply
#166
RandallFlagg
Vayra86
Yeah, if you think laptops are somehow not built with limited TDPs, Tiger Lake looks amazing. When you let that chip fully stretch its legs... it can do with 2.8 Ghz what Skylake could do with 4.0. But it still caps at that eternal 4c8t Skylake perf level. That is the point. Now look at the TDP of the 202% faster (!!) 10900K. Change the desktop? The proof is right there in the chart that it won't.

Now, let's get back to reality. I get your point, but this won't change a thing in anything that is not power limited. Its easy to get lost in the woods of Intel benchmarking logic.
Ridiculous. That is a 4c8t Tiger Lake beating an AMD 8c8t part, not Skylake. I didn't mention Skylake, you did that.

And I said it may challenge all "but the highest" performing AMD/Intel chips. You went and ran with that to the 10900k, their top chip. You got a reading comprehension problem or a social issue? Just wondering.

And yes it is entirely possible that a 35W version of this with 6C/12T will challenge all but the highest AMD/Intel desktops. What do you suppose Tiger Lake looks like at 4Ghz with 6 C and 12 T? 33% faster clock and 50% more cores would put this on par with a 3800X or a 10700K.

Maybe you should try thinking before posting?

And no I didn't miss anything about the limitations on current Ice Lake low power chips. You missed the basic math part (see above). You also missed the part where I noted laptops were larger market than desktops. And on and on.
Posted on Reply
#167
Vayra86
RandallFlagg
Ridiculous. That is a 4c8t Tiger Lake beating an AMD 8c8t part, not Skylake. I didn't mention Skylake, you did that.

And I said it may challenge all "but the highest" performing AMD/Intel chips. You went and ran with that to the 10900k, their top chip. You got a reading comprehension problem or a social issue? Just wondering.

And yes it is entirely possible that a 35W version of this with 6C/12T will challenge all but the highest AMD/Intel desktops. What do you suppose Tiger Lake looks like at 4Ghz with 6 C and 12 T? 33% faster clock and 50% more cores would put this on par with a 3800X or a 10700K.

Maybe you should try thinking before posting?

And no I didn't miss anything about the limitations on current Ice Lake low power chips. You missed the basic math part (see above). You also missed the part where I noted laptops were larger market than desktops. And on and on.
Okay buddy, keep believing whatever you think is true. You're just grasping specs out of thin air thinking you can simply multiply some numbers and boom you've predicted what a node will do under higher power delivery. If you want to base yourself on that, hey, who am I to stop you. At least you got your 'basic math' down, eh?

Try some knowledge on semiconductors in the mix, it might help a little bit. If semicon fabbing was basic math, surely we'd see lots of basic mathematicians over there wouldn't you think?

About Skylake... I pulled that out because it is the perf level Intel has capped on for a decade, and with Tiger Lake, they have once again created and optimized a product around that very same performance level. Look at where the 6700 lands compared to Tiger Lake on full steam. I'm not here to pester you or anything, I'm trying to provide insight you have obviously missed when you say 10nm equals TSMC's new nodes.

Either learn something or don't, but you don't need to convince me. This is not an AMD - Intel pissing contest for me either, contrary to what you might think. AMD's laptop segment for CPUs is severely lacking even today, I'll concede that right away.
Posted on Reply
#168
RandallFlagg
Vayra86
Okay buddy, keep believing whatever you think is true. You're just grasping specs out of thin air thinking you can simply multiply some numbers and boom you've predicted what a node will do under higher power delivery.
I never said high power delivery, you said that. And that's the third time you've attributed your statements to me.

I said their laptop segment (READ: LOW POWER) with Tiger Lake may be as powerful as all but the the highest end AMD/Intel desktop variants.

Again, that reading comprehension / social issue of yours.
Posted on Reply
#169
Vayra86
RandallFlagg
I never said high power delivery, you said that. And that's the third time you've attributed your statements to me.

I said their laptop segment (READ: LOW POWER) with Tiger Lake may be as powerful as all but the the highest end AMD/Intel desktop variants.

Again, that reading comprehension / social issue of yours.
I've already said that I agree on your laptop segment dominance for Intel... you can point me to the post where I said otherwise? Reading comprehension goes both ways, perhaps. On top of that, your bench showed us a Tiger Lake chip running without power constraints, and those dó matter for laptops, in fact throttling is a major issue on almost all of them.

You said, and that is what I responded to because that is where you missed the mark: that Intel's 10nm is better than TSMC's 7nm. I specifically pointed to the density issue in play that makes the difference here. With higher density, higher power in Intel's node will quickly lead to disaster. So no, it won't rival faster desktop parts, because it won't compete with them. TSMC has a node that can scale, Intel has one that is relegated to bottom end performance. Their 10nm is a dead end, basically.
Posted on Reply
#170
RandallFlagg
And where did you get the impression the thermal limits were removed? The source for that data was Notebookcheck, and they stated the Tiger Lake part was clock-locked at 2.8Ghz. That's not thermal unlocked. That's crippled.

But I do believe I know where you got that from (not me). That would be a different comparison at wccftech, between the 8C 4700U vs the i7-1165G7, where they were both set to 25-28W. They also used a different bench, not Time Spy.

That comparison was even worse for AMD, as the 4C Tiger Lake beat the 8C AMD by 34% on 3dMark overall, 35% in graphics, and the little tiger lake matched the AMD chip in multi-threaded CPU performance almost exactly despite having half the cores.

wccftech.com/intel-tiger-lake-core-i7-1165g7-on-par-with-amd-renoir-ryzen-7-4700u-cpu-3dmark-time-spy/
Posted on Reply
#171
Vayra86
RandallFlagg
And where did you get the impression the thermal limits were removed? The source for that data was Notebookcheck, and they stated the Tiger Lake part was clock-locked at 2.8Ghz. That's not thermal unlocked. That's crippled.

But I do believe I know where you got that from (not me). That would be a different comparison at wccftech, between the 8C 4700U vs the i7-1165G7, where they were both set to 25-28W. They also used a different bench, not Time Spy.

That comparison was even worse for AMD, as the 4C Tiger Lake beat the 8C AMD by 34% on 3dMark overall, 35% in graphics, and the little tiger lake matched the AMD chip in multi-threaded CPU performance almost exactly despite having half the cores.

wccftech.com/intel-tiger-lake-core-i7-1165g7-on-par-with-amd-renoir-ryzen-7-4700u-cpu-3dmark-time-spy/
From the wccftech article:



Note the CPU test. Tiger Lake has a good IGP. Great. But that is still circling the laptop point which I already agreed on - three times now... CPU is like you said, a match at best - but we have already got our hands on 7nm AMD desktop parts while there are no 10nm high performance Intel parts. That underlines why TSMC has a better node at this point. They can serve the entire market.
Posted on Reply
#172
RandallFlagg
I was looking specifically at the 8C/8T part. The 4800U has 8C/16T. I usually consider the T part of that to be like half a real CPU, so an 8C/8T vs a 4C/8T would be like comparing 8C vs 6C w/o hyperthreading. Basically the numbers come out the same as stated in the article - Tiger Lake is ~ 30 to 35% faster per core vs Renoir.

This means even Zen 3 laptop chips from AMD would have a hard time competing with Willow Cove (Tiger Lake) cores.

4C/8T Tiger Lake (right) vs 8C/8T Renoir (left). It is besting Zen 2 APU on every single metric.

Said this before, if Intel can get this out in a 6C/12T part and up to 35W+ AMD won't have anything in the laptop space to compete.

Doubly so when you consider what the Xe graphics are doing here - I haven't mentioned nor cared about that, but for 90% of laptop owners the iGPU is very important.

Renoir will wind up being one of those things that came out and was really great for 3 months, then got obsolete.


Posted on Reply
#173
Vayra86
RandallFlagg
I was looking specifically at the 8C/8T part. The 4800U has 8C/16T. I usually consider the T part of that to be like half a real CPU, so an 8C/8T vs a 4C/8T would be like comparing 8C vs 6C w/o hyperthreading. Basically the numbers come out the same as stated in the article - Tiger Lake is ~ 30 to 35% faster per core vs Renoir.

This means even Zen 3 laptop chips from AMD would have a hard time competing with Willow Cove (Tiger Lake) cores.

4C/8T Tiger Lake (right) vs 8C/8T Renoir (left). It is besting Zen 2 APU on every single metric.

Said this before, if Intel can get this out in a 6C/12T part and up to 35W+ AMD won't have anything in the laptop space to compete.

Doubly so when you consider what the Xe graphics are doing here - I haven't mentioned nor cared about that, but for 90% of laptop owners the iGPU is very important.

Renoir will wind up being one of those things that came out and was really great for 3 months, then got obsolete.



Note the clock speeds. Ryzen at 2000mhz and Tiger Lake at 2800mhz. Its a perfect CPU score match. 0.01 FPS gap and 2 points. Tiger Lake needs higher clocks to get the same performance.

Higher IPC? That is not what this bench shows you - it shows the opposite but you've mixed core count into that equation as well, haven't you ;) That is maybe where you're missing the mark in your comparisons - Zen's strength is exactly its chiplet design which means AMD can cram more cores in with no issue whatsoever, and this translates to a 'wider' CPU that can remain in an efficiency curve more easily, not bursty but steady in clock behaviour. In addition, Intel has a higher peak clock as well, which is just another echo of the bursty laptop CPUs we already know. I think you're overestimating what you see in a big, big way. Its no coincidence we're looking at 4C8T Intel chips here. IF Intel can get this out as a 6c12t? That G7 was announced, released and then not seen in the wild for over a half year. And even now, they're rare unicorns. I don't believe in whatever Zen 3 might offer or what Intel might do with bigger chips that don't exist. I only care about actual releases, and this is what we have.

And the 'Xe' graphics are just another rebrand of Iris. Please... you said something about happily believing marketing about nodes... you're falling for the Intel branding madhouse, bigtime. There isn't that much here that is going to radically change anything. If there even is anything. You think Raja came to Intel, farted three times and poof new IGP? They work from the top down, not the bottom up. Whatever Xe is supposed to be in the lower end, it cannot be more than a few tweaks on whatever Intel already had.

All you've got is 'Tiger Lake in a different situation with higher core count and much higher TDP limits will destroy them all'... yeah. Cool story. Intel dominates laptops because they have supplied that market for over a decade and they've been proven reliable. Tiger Lake merely continues that spree, nothing more. Its not a game changer - it just manages to get by against nodes that are slowly but surely going to trump it anyway. If Intel does not move to 7nm and smaller within now and the next two years, they've got 14nm woes all over again. Keep in mind that recent history was not all about performance either, but also about security issues, heat/throttling issues (new heatspreader designs every gen, does not instill faith in any long term strategy) and lots of confusion and delays in supply chains. Intel is losing investors and rapidly losing share price - and it has already held on to the old glory for quite a while if you ask me, despite countless reports and bad omens.

For AMD, Renoir is just another stepping stone. For Intel, Tiger Lake is a desperate attempt to keep up.
Posted on Reply
#174
londiste
IPC isn't higher but it isn't lower either. It seems to match up pretty well to what is expected from 4c8t at 2800MH vs 8c8t at 2000MHz. This is actually a good sign.

G7 was announced, released and has been available pretty reliably from what I can see. There are about 2 real SKUs of Ice Lake that seem to be steadily available - 1065G7 and 1035G1 - the rest could as well not exist.

Chiplets have nothing to do with Ryzen 4000 series - this is a monolithic CPU. At least the current chiplet design is not well suitable for mobile usage, primarily due to high idle power.

Xe is as much a rebrand of Iris as RDNA2 is a rebrand of GCN. It is in part and isn't in the other.
Posted on Reply
#175
RandallFlagg
You are both comparing base clock to base clock. During relatively short periods where these benchmarks are run, turbo counts. And in turbo, it's 4.2Ghz (ryzen) vs 4.7 Ghz (intel). In the end these semantics matter not, as it is just the raw performance that matters. Ignore that and you just get lost in the weeds of minutia.
Posted on Reply
Add your own comment