Desktops are irrelevant. Corporate issuances(biggest use of laptops) are almost completely laptop. Personal use is moving more towards mobiles and ipads(and may be chromebooks). Both Intel and AMD are irrelevant on the mobile side.
Depends on the company. I work for a company with 100k+ employees, and due to the significantly higher likelihood of device theft, have begun migrating from laptops-for-most back to desktops-for-most.
But that also means your company still provides desks for a good chunk of 100k+ employees, which is bucking the trend of making everyone a virtual / work-from-home employee to cut costs.
The work from home trend is actually reversing. More and more companies are starting to require a "work in office, but take your laptop home in case of disaster" type scenario.
"Work in the office where we know you're devoting your full attention to us, then take your laptop home and work from home too, so we know you are devoting your full life to us."
Indeed. The company I work for is shooting themselves in the foot - they are taking away free phones to save $700k in carrier fees annually but didn’t bother to calculate all the free hours they get out of their employees because of them. Thank god I am a contractor but regardless I don’t work a second I’m not paid for.
WFH is proven to be less effective if used more than 1 day a week. Communication overhead was much higher. And yes I agree that I see it often that people have desktop (windows/linux) + mac.
The worst of both worlds. Not only do you have to be in the office, but you have to lug the laptop both ways with you every day as well with no reward.
Let's be honest they wont be motivated at work either. I think you will find working from home just makes it easier to figure out who are the slackers you need to toss out.
The truth is that there has never been a whole lot of that. Also, it hasn’t worked out as well as hoped. It turns out that true face to face interactions are better.
This is the most ridiculous statement I've ever heard. You're telling me there is a corporate trend toward less autonomy and portability because of...device theft?
Enterprise packaging and volume sales numbers don't back that statement up. Desktop PC's have been declining (5% loss YoY) for a decade, while mobile PC's have been basically neutral ( less than 10% loss of growth over a NINE year period, mostly offset by tablet sales) while projections show desktop sales to continue falling and mobile sales to begin rising as has been the case lately since the introduction of Coffee Lake - a milestone for laptop performance improvements (twice as many cores, over 30% productivity performance improvement, for the same amount of power)
AT won't let me publish the Statista link so you will just have to Google it for reference.
Device theft is one of the excuses companies are using in some places. "Synergy" or team environment is also another term being used to tell people they can't work from home or remotely. IBM is using that as an excuse to get people to quit so they don't have to pay severance or report layoffs.
Honestly it's just going to be poor upper management at this company t hat is moving towards desktops. The worse the management is the closer they want their employees and the more micromanaging you will see.
"Honestly it's just going to be poor upper management at this company t hat is moving towards desktops. The worse the management is the closer they want their employees and the more micromanaging you will see. "
For the life of me I cannot fathom why people come out with this guff and expect not to be challenged on it.
I think what's happened, is there are so many people talking absolute BS on the Internet these days, that people don't get pulled up, and then carry on believing they're correct.
What's actually happened is that YOU have had a run in with someone in YOUR upper management, and so because people are generally stupid, you extrapolate YOUR personal experience to being factual accross the entire board.
Don't worry though, you're far from being alone in this. It's the Facebook Generation, they are all guily of it somewhat. From "Journalists" at (Once) "Respected Publishers", to glass collectors, to Police Officers, and so on.
I call it the "Pub" test. Could you say what you did in a pub and get away with it. The difference being that in a Pub it would be YOU who has to walk out in shame for talking absolute BS after people publically correct you.
Whilst online, you can surround yourselves with other idiots who are prepared to agree with whatever BS you come out with so long as you do the same in return (AKA the Echo Chamber).
Still, if you were to correct every last soul on the Internet whose chatting BS, and saying their personal experience as absolute facts, you'd need more time than the universe is able to provide.
So, I leave in it the hands of whatever religion is in fashion this year, pray to those volcano Gods...
Totally irrelevant when the PC Gaming Market makes more money than the Hollywood Movie. (Not counting consoles) yep totally irrelevant. But please do tell me more.
I couldn't figure out how to get that site to show history back to the 80s, since that's when they lost to compaq and intel. They had the entire market. Can you imagine if there was a microsoft of hardware today? Some stock figures from their business machines tucked away in a warehouse say nothing of what they lost due to hubris.
Whoops just found it. They are worth more now than they were in the 80s, but that doesn't say what they could be had they not dropped the entire PC market.
If you want to play what-if like that, imagine if Intel had gotten the CPU contract for the iPhone and some what buckled to Apple's low margin, high volume demand for that market.
Another fun what-if is AMD buying nVidia instead of ATI.
Intel still had several Netburst design waiting in the wings before there right turn at Core 2, so what if they continued down the Netburst path?
Last fun what-if: Intel was actually on time with Itanium and did manage to get vendors to push out 64 bit workstations/highend PC's and AMD still released their x86 extensions.
IBM didn't have the entire market in the 80s. Not minicomputers, not personal computers. There was very robust competition from an abundance of architectures. Even more if you include home computers instead of just business machines.
That said, they did far better than many of their fellow old-guard computer companies because they insulated their personal computer division so the minicomputer division couldn't sabotage it. There's a world where we all use machines derived from DEC's PDP-11, TI's TI-990, or Data General Nova. To name but a few of the contenders that self-sabotaged. I think IBM is the only company with a minicomputer division that successfully made the transition to personal computers. Most that tried never shipped a micro, and those that did shipped horribly neutered machines at the demands of the minicomp divisions.
Which is why AMD will release devices sporting Zen 2 based APUs by the year's holiday season as well. On the CPU core side Ice Lake-U/Y are going to be *barely* faster than the Whiskey/Amber Lake CPUs they are to replace. While they have a higher IPC (not 18% higher, that's over the original Skylake of 2015 - *not* 2016 btw) most of the IPC gains will be eaten away by their much lower clocks. The much lower clocks, in turn, are due to their 10nm+ node's atrocious yields.
Extrapolating from the graph Intel released (they only compared Ice Lake-U 15W to ... Broadlake, in just one graph, in an obvious attempt to obfuscate the numbers) they are going to have a 3 or 4% higher single core performance, 15W for 15W. At best, and if the Sun always shines on them, a +5% performance tops. Unless they trimmed a lot of their "multi-core overhead", which I doubt, the multi-core performance gain will also be comparable, in the 3 - 5% range. If this was a mere refresh the performance gain would be fine, but since we are talking about both a new μarch and a new process node it is truly pitiful, which why Intel did their best to mask Ice Lake's CPU performance deficit in their Computex presentation (they masked it with iGPU, neural network and ..... wi-fi performance numbers, "forgetting" to talk about CPU numbers at all in their main presentation; they released them later to the press in an obfuscated form, almost as a footnote).
Nevertheless, despite their CPU deficiency, Ice Lake-U/Y will also sport a new iGPU, a new media engine with more codecs and higher resolution support, support for Optane memory, and native Thunderbolt 3. The new iGPU, in particular, is probably the most dangerous for AMD since until now they had a clear edge on that era. Intel will need to be quite aggressive in their iGPU deployment for that to pose any threat though, and not retain the 48 core and 64 core iGPUs just for the high end SKUs. I do not expect that to happen, with Intel still being Intel here. The 64-core iGPUs in particular will surely be super rare, just for 28W parts with eDRAM.
In any case AMD need fast and capable Zen 2 based APUs in order to compete. They will probably not exceed 15W TDPs either, so they will have a clear gain in power efficiency as well. Intel's "abnormal" TDPs are also due to their yield issues. Fanless devices with 9W Ice Lake-Y parts are probably still possible, but they will require quite sophisticated -and more expensive- passive cooling solutions.
Using Sky Lake as baseline is fine as they are still effectively on that micro-architecture. The biggest changes in over the past couple of years have been from an increase in shared L3 cache size (a side effect of higher core count) and the patching of security vulnerabilities. The real comparison is what the base clocks are vs. turbos.
Ice Lake does have lower turbos from the information that has leaked thus far but I have yet to hear much about base clock speeds, especially when under AVX or AVX-512 modes. This may not be as dire if Intel can keep up average clock speeds similar or higher than Coffee Lake and ride the IPC increase for the generational performance gain.
I would expect Optane DIMM support in the memory controller but I highly doubt that Intel would genuinely enable that in consumer systems. (Xeon E is a different question.)
The parts with eDRAM will likely find a home inside of MacBooks, like what happened with similar Sky Lake, Kaby Lake and Coffee Lake designs. The real radical was Kaby Lake-G with its Vega graphics as I am really shocked that that hasn't been picked up by Apple.
For now they have their present Ryzen Mobile 3000 line. There isn't news about APUs on 7nm right now but we can argue that the producty cycle is minimun one year, so we will not see 7nm APUs since Q1/Q2 next year. Remember that Intel has Tiger Lake for 2020 with the far more efficient Xe ark on GPU, so AMD has to react fast to not lose the small market share in Laptops.
And Gamers Nexus made an amazing discovery: Ryzen CPU's undervolt to a stable level as low as 1 volt -- they maintain 3Ghz at 1v, yes, but they are stable at super low voltages... meaning AMD doesn't even need to spin new silicon for mobile chips, they just need to work with board partners to make low profile sockets.
So true. I haven't seen a desktop PC in a workplace in ages. Most desktops seem like either rendering or heavy lifting workstations or are delusional slap-together machines people think makes them better at vidya games.
It comes down to ergonomics. A full keyboard, mouse, and large monitor is much cozier than a tablet or laptop for 8 hours of work a day.
You could dock your laptop and be able to move it around or you could get a higher performing desktop for the same price. There is still value to desktops to businesses.
Laptops are a serious financial risk for companies, because they have zero ergonomy. A monitor has to be in front, and same level as the eyes. No Laptop can do that, so they all cause neck injuries in long term. Any serious company should ban the use of laptops. Even personal ones.
Some companies have moved to thin clients. We use Dell Wyse DX0D thin clients for like 90% of our work force. Roughly 2k people. Which still use the low tier Intel or AMD procs. Like Atom or something.
I also don't see how both Intel and AMD are irrelevant on the mobile side, cause they're both still in your laptops, ultra books, Chromebooks, etc. Android and iPads is where they aren't. Sure, everyone has some kind of Android or iOS phone, but not so much Android tablets or iPads.
Most still rock some kind of cheap PC laptop. Especially when you can get them for a lower price than many Android tablets or iPads.
You can always benefit from more cores, especially if clock speed isn't sacrificed. Right now just browsing the internet and having a few background apps open I have 229 processes and over 3000 threads. I get occasional small spikes on every core + smt from those processes/threads. The system is smooth and responsive. Usually the guys saying that 16 core CPUs (or 8 core for that matter) the the same guys running quad core CPUs, SATA drives, and low end GPUs.
Maybe not and 16 might be overkill for browsing but websites get heavier and browsers are building multithreaded browser engines - Firefox is halfway there and the rest will follow.
it depends on what you are browsing. saying browser does nt need 16 cores is like saying .net programs dont need 16 cores. what does javascript have to do with core count
those are going to be a big step up from previous mainstream chips for mutlithreaded work. but its disappointing that a chip as old as a 8700k or even a 7700k has the same single thread performance when either chip is clocked to its max. ill declare victory for amd when an intel and amd chip, both clocked at their max, and the amd chip wins in both. but right now they really are pulling ahead. but you have to realize its not just because they are doing so well but mostly because intel are sidetracked. amd fully expected to launch these chips and be just behind intel. and if intel had gotten sunnycove and 10nm out they would have been.
We haven't seen benchmarks from the 3950X yet, and AMD has AGESA issues to sort out. You will also note that coffee lake was Intel's last big hurrah. The 9900k simply features 2 additional cores and a higher single core boost. No IPC increase. Intel has hit a wall. Note that I'm not a fanboy at all, simply acknowledging the truth. I also take issue with Intel (and now AMD to a certain extent) are lying about TDP.
You just read an article where intel claims a 18% IPC increase in their latest chips. They are not yet for desktop and don’t clock well, granted, but see at what speeds the first 14nm chips hit the market and see what they do now. Then you know where 10nm is going.
Amd and TSMC will have to ramp up clock speeds on their 7nm quick. I expect a refresh next year with better clocks, I hope at least 500mhz on top of what the current chips promise-but-do-not-deliver... otherwise intel might come back quick.
AMD roadmap has Zen 3 with TSMC N7+ node slated for next year. AMD could be focused on getting higher clocks, but I'm more inclined to believe they will improve IPC instead.
AMD will do both. AMD is working on core design improvements while TSMC improves its fab process. It's not one or the other, unless you do what Intel did, and link design improvements to the fab process, which has severely hurt the company. Why Intel couldn't have put a new design on 14nm is a very good question that I have not seen the answer to.
When you see benchmark results, you see them done on systems that are very clean, with a new install of Windows, and then the test suite. As a result, these benchmarks are NOT being run on the primary systems that are being used for office work. If you have Steam, Origin, Antivirus, plus other things, you are already putting a greater load on the system than what is on systems that are posted for benchmarks.
The result, you see less of an advantage to additional CPU cores than people do in the real world.
the clock are always lower in the first wave of an intel process. the first process is always about power and density. then the second wave, the plus, (14nm+,10nm+, etc) is about higher clocks but sacrifices is a bit of density to do it (for instance 14nm++ is 35 million transistors per mm2 while 14nm was 45, incidently amd 7nm is rumored to only be about 40-55 by the time amd actually uses its own libraries and adapts the 7nm tsmc process to its own needs).
so yes, just like broadwell was the first 14nm and sacrificed a lot of clocks so will these. and just like broadwell these chips are mobile focused so clocks dont matter as much, only power.
once intel finally rolls out 10nm for desktops we may see lower clocks but not like these chips. i wouldnt be surprised if a max oc is 4.6-4.8ghz ish. could go as low as 4.4ghz. but then again they could still be reaching 5hgz. maybe thats why they are putting off 10nm desktop cpus for so long. that and cost
i would expect that the next move for intel is to either put some form of backported 14nm sunnycove out which is what i think rocketlake (the 11th gen is). if they could get the same clocks they are getting now and combine it with 18% ipc gains then that would be a competitive chip. even if they only make core counts similar to what they have now they may lose in multithread performance still but for most average consumers lightly threaded defines performance still.
but the sunnycove cores are quite a bit bigger so how do they fix that on 14nm? well they could leave off the igpu and use that space. the igpu on current chips already takes up the exact amount of space as 4 cores. and coincidently rocket lake is listed as having seperate igpu and cores on intel slides. so if they used that space to make these larger cores in the same amount of space they wouldnt really cost any more unless you bought a version equipped with an igpu. current kf cpus with no igpu actually still have an igpu but its just turned off. so no space savings there.
intels 14nm is actually quite cheap by now. their 14nm cpus cost so much now because a) they still demand a premium price because they are still good chips b) the most important part is that they just have a shortage. they cant produce enough.
if intel move igpu and chipset production to samsung, plus move some of their line over to 10nm for mobile and maybe expand some 14nm production where they can then they could afford to make an 8 core 16 thread sunnycove based cpu and it would cost no more to make than an i9. actually with no production shortfall they could make it the next i7 and sell it for well under 400 dollars. that would be a fairly competitve cpu except in production workloads where amd would win by a good margin.
this would hold them over till they could get 7nm EUV going and then skip straight to the less dense, higher clocking, + version of 7nm for desktop.
This is actually the second 'wave' of 10 nm parts. Cannon Lake was the first which actually did ship but in chronically small volumes due to manufacturing issues.
It should be noted that for desktop, Intel has not committed to bringing 10 nm parts to it in 2019 or 2020 with the possible exception of HEDT. Thus far Ice Lake is for mobile and server which carries them until 2021.
Rocket Lake could leverage some of Intel's new packaging technologies but that too is currently targeted at mobile where physical space constraints can justify the price premium and server for scalability. Much like the lack of 10 nm desktop chips on their roadmap for 2019/2020, these advanced packaging technologies will likely come to desktop last.
Intel is pretty much treading water until 2021 but their line up for that year on their road maps does look very interesting.
It would benefit Intel a lot more to port 'willow cove' to 14nm. It is their 2020 architecture and any backports are unlikely to arrive before 2021 (Comet lake in 2020). Intel will have a hard time defending their single core performance advantage in 2020. They will have to make a big advancement in 2021, how they'll do it is speculation at this point.
If you buy a new, "top of the line" product, and it isn't better than the previous generation products, you will be disappointed. TSMC 7nm is a new fab process, and it's better for AMD than the 12nm parts in clock speed, so claiming that the new fab process should be forgiven for not being as good if there is zero benefit to the user is a bit of a joke. Intel has been claiming that 10nm is on track since 2015, and still isn't better than 14nm.
Eventually they'll get things ramped up. The big issue with node shrinks are increased power leakage, which means lower clocks. They've been on 14nm for many years so they've had time to optimize their designs. Once they get a few more generations underway, you'll see some clock speed improvements. As to the base clocks, I can't remember exactly, but they were much lower. That is mainly due to the fact that these are laptop chips.
“Once they get a few more generations underway” So they’ll have competitive chips in, I don’t know, 5 yrs from now? (A few more generations...)
Good thing Apple and AMD plan to take a break over the next five years...
(BTW if Apple manages to get just 20% more single-threaded performance out of A13X — which is below there annual trend ever since the A6 — they’ll already match the BEST Intel has to offer that’s not overclocked...)
So Intel have finally got a 10nm chip - however it is a small chip with a maximum of 4 cores. As Intel do not use the equivalent of the AMD chiplet design, the acid test will be how long it takes Intel to produce a 10nm equivalent of its current high end designs (16-28 cores). I personally do not expect volume deliveries of 10nm high end CPUs from Intel before 2021 (possibly even 2022). (Intel high end 10nm chips may also have thermal problems for Intel to overcome as the chip would end up with about half the surface area but the same power dissipation as current generations - this is not a problem for AMD as the chiplets are physically separated so no single hot spot exists.)
According to everything I've heard, Intel's 10nm process has the following issues:
- The new metals they're using (Cobalt etc.) have different thermal expansion coefficients, putting more thermal strain on the chip.
- Cobalt has a much lower thermal conductivity than Copper, basically acting as an insulator, so hot spots inside the chip become even hotter. This puts even more strain on the chip and doesn't allow high clock speeds.
- Cobalt is brittle and doesn't handle strain as well as copper.
This means the 10nm chips are much more likely to die from thermal fatigue than their 14nm models. Which is probably why they are having yield issues - the chips are working when they come off the production line, but most of them would die too early in the hands of customers.
I'll stick with AMD while other people are playing guinea pigs for Intel's 10nm.
first.. they are using cobalt interconnects instead of tungsten. secondly, cobalt is the standard the entire semi-conductor industry will eventually switch to, in order to lower the contact resistance and reduce thermals.
Umm yes you are absolutely right. Officially Samsung too is going with Cobalto to increase performance, same applies to TSMC. IMO the 10nm "issue" is only related to yields not to absolute clock speed.
If you consider actual TSMC painful situation, you can realize that right now they are unable to manufacture in volume a large die in 7nm, they can give only a very low volume output with some overheat around. If you look at the 7nm process it is limited on clock speeds, virtually unable to overclock and pretty hot at the silicon level under load.
The real test on TSMC 7nm is Nvidia, but all journalits are silent about this bashing instead Intel all around. In a recent statement Nvidia said that 14/12nm is a better choice for the Company. Nvidia said that 7nm is too expensive to sustain and for now they will stay on 12nm more like is doing Intel (this is a July 2019 thing) I'm asking to myself what is the real AMD net profit over Ryzen 3000 line, expecially over the large footprint 12 and 16 cores SKUs. Considering the likely less than 50% yields level of these high turbo dies, they have to pay two to have one or four to have two good enough. Not to mention the very high packaging costs to assemble three dies with fast data links on the same substrate. The situation is far different from the stellar 90/95% yields level of 14nm or 12nm.
I fully understand you. The resurgence of AMD have badly cancelled any ARM hopes to penetrate the very lucrative Laptop segment and absolutely destroyed the so much hyped server adventure of this RISC ISA. Funny enough AMD is best Intel Friend. Who is ARM outside phones.....nothing.
There are multiple issues here. First is the difference between those of us looking forward and those people looking backward or wanting to buy something today.
Second (and this is SO important) is the issue of what people like you would count as an "ARM laptop". Apple will (probably) be shipping these soon, (probably) by end of 2020. But would you count that as an "ARM laptop"? For anyone apart from Apple, there are basic issues: - an OS is needed. So who provides that? +MS, the company that has NEVER been able to figure out its OS strategy since at least .NET, still has no clue exactly what it wants. Does it want x86 to be "real" Windows and ARM to be "protected Windows"? Or is ARM supposed to be "consumer Windows"? Or is ARM supposed to be "Windows that never pisses of Intel"?
+ how about Linux? Oh, you mean we'll FINALLY see the year of the Linux desktop? There's a reason no-one plans to get rich selling Linux desktops, and that ain't gonna change.
+ Android? Well Android or ChromeOS? Just like MS, WTF knows what Google's OS strategy is long-term?
And of course in all these cases, what do YOU, the complainers, ACTUALLY want. Half the time you seem to want a super-cheap chromebook like ARM laptop. Then, when those are available, you complain that they don't have the nice screen, touchpad, battery, flash, etc of an Apple device, suggesting you want a ~$1000_ ultrabook...
I don't think ANYONE apart from Apple can provide anything useful in this space. Which means we wait for Apple to ship, then see how excited the market is about the results compared to current MacBooks. (Noticeably faster? Noticeably longer battery life? Noticeably innovative features?) Then we wait for MS and/or Google to slowly pivot to copy all that.
The other dark horse would be a Chinese company creating a very cheap ChromeBook like thing. But once again there is the OS problem which doesn't just go away because China. Sure, they can slap some horrible Android-kludge on it which never gets updated again, and the result might even be desirable and useful for the poorest parts of the world, better than nothing. But that doesn't provide what people like you mean when you ask for an ARM laptop.
So yeah, for the next five years, it's basically Apple (hopefully in a year!) or nothing...
U and Y series have an on-package PCH, not on-die (the CPU and PCH dies are discrete but part of a multi-chip package). And H series is still mobile but the PCH is a separate package. So not the same at all.
The point being that Intel did not integrate the Thunderbolt controller into the PCH die, but into the CPU die itself.
You're correct. Unfortunately Intel's own diagrams from this info (which Anton was working from) are themselves incorrect in this case, compared to the ones from Intel's one SoC deep dive event.
I understand the concept of managing inventory, but if Intel needs to start shipping CPUs in Q1 in order to stock up for a holiday Q4 launch, something seems off. Are yields that low or production capacity that limited that such a long lead time is needed ?
No, chips have to be shipped several months before the launch so that people can buy the actual products (fully assembled laptops ready for sale) on the official launch day.
No, there is nothing normal about this product ramp / launch from Intel.
They considered ICL “announced” and “shipping to OEMs” as of May 27, 2019 in order to book 10nm revenue and convince investors they were hitting their targets. And despite the devices “on shelves for the 2019 holiday season” messaging which implies the sales embargo being lifted as late as Q4, leaked Lenovo and Dell roadmaps showed ICL products as early as June and August respectively.
But Lenovo already shipped those products with Whiskey Lake, and Apple just refreshed the MacBook Air and 13-inch MacBook Pro with Amber Lake 2+2 (same renamed Kaby Lake as before) and off-roadmap Coffee Lake-U 4+3e CPUs. We also know that Ice Lake-U/Y will be sold alongside Whisky Lake-U, Amber Lake-Y, and Comet Lake-U/Y. It is looking increasingly like HP and Dell might be the only major OEMs to feature Ice Lake in flagship products this year. In other words, Intel was only able to supply 2 of the top 4 OEMs with sufficient volumes of chips for them to ship key products in 2019.
Intel has dilated the launch window massively here by claiming they were "shipping" during what would normally be considered the qualification sampling period. And the thing is, they probably did get the OEMs to pony up the cash already, so now the OEMs are sitting on Intel's entire output of 10nm chips, bought and paid for, while they work out any problems and wait months for the sales embargo to lift.
Good! Now, let's see what AMD has to offer in the low and lower power mobile space, and see who's on first! Whoever gives me the best bang for my buck gets my orders.
I’m afraid amd has no plans short term bringing their new cpu and gpu architectures to mobile... the upcoming app is still last gen cpu and gpu. Too bad, given the size of the mobile market, but I guess amd expects to use most of their manufacturing capacity for the enterprise market...
@Jospoor - Navi is their attempt to scale up and down - and the Samsung partnership is huge when you think in terms of mobile and scale for easy royalty payments.
It is wierd that AMD is a generation behind on mobile parts and Intel on desktop parts, but that's how it is. AMD will not have 7nm mobile ready for holiday season, while Intel will not have 10nm desktop ready.
That’s not the point. The point, I think, is that it suggests that the existing AMD chips provide a fairly easy path to laptops as soon as AMD wants to make that move. At least that was my interpretation of why the issue was being raised.
Let me just pull up my graph of IPC improvements going from Skylake to Kaby Lake, Kaby Lake Refresh, Coffee Lake, Coffee Lake Refresh, Amber Lake, Whiskey Lake, and Comet Lake:
o—o—o—o—o—o—o—o
Doesn’t matter which “gen” you choose from, 7th, 8th, 9th, or even 10th, Skylake cores all have the same IPC. Cannon Lake did show some modest IPC gains, but seeing as only 10K or so units ever shipped, it’s not really a meaningful comparison.
No touted xtreem battery improvement this spin around? No Doubt battery's will still shrink by something like 5% to insure we are tossing these in the trash and crawling back for more in two years time.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
115 Comments
Back to Article
Machinus - Friday, July 26, 2019 - link
Meanwhile, we're about to order 16-core 7nm AMD chips...for the DESKTOP!imaheadcase - Friday, July 26, 2019 - link
Yah that like %1 of people will actually be able to use.... for the DESKTOP!trivik12 - Friday, July 26, 2019 - link
Desktops are irrelevant. Corporate issuances(biggest use of laptops) are almost completely laptop. Personal use is moving more towards mobiles and ipads(and may be chromebooks). Both Intel and AMD are irrelevant on the mobile side.catavalon21 - Friday, July 26, 2019 - link
Depends on the company. I work for a company with 100k+ employees, and due to the significantly higher likelihood of device theft, have begun migrating from laptops-for-most back to desktops-for-most.repoman27 - Friday, July 26, 2019 - link
But that also means your company still provides desks for a good chunk of 100k+ employees, which is bucking the trend of making everyone a virtual / work-from-home employee to cut costs.eek2121 - Saturday, July 27, 2019 - link
The work from home trend is actually reversing. More and more companies are starting to require a "work in office, but take your laptop home in case of disaster" type scenario.Lord of the Bored - Saturday, July 27, 2019 - link
"Work in the office where we know you're devoting your full attention to us, then take your laptop home and work from home too, so we know you are devoting your full life to us."ballsystemlord - Saturday, July 27, 2019 - link
Ah, the refreshing cynical viewpoint.More seriously, I've read several articles where Amazon does just that.
29a - Saturday, July 27, 2019 - link
This.Icehawk - Sunday, July 28, 2019 - link
Indeed. The company I work for is shooting themselves in the foot - they are taking away free phones to save $700k in carrier fees annually but didn’t bother to calculate all the free hours they get out of their employees because of them. Thank god I am a contractor but regardless I don’t work a second I’m not paid for.deil - Monday, July 29, 2019 - link
WFH is proven to be less effective if used more than 1 day a week. Communication overhead was much higher. And yes I agree that I see it often that people have desktop (windows/linux) + mac.psychobriggsy - Tuesday, July 30, 2019 - link
The worst of both worlds. Not only do you have to be in the office, but you have to lug the laptop both ways with you every day as well with no reward.willis936 - Saturday, July 27, 2019 - link
I too can make up facts to support my narrative.Flunk - Sunday, July 28, 2019 - link
Employers are realizing that for less motivated employees "work from home" means "do half the work".FreckledTrout - Monday, July 29, 2019 - link
Let's be honest they wont be motivated at work either. I think you will find working from home just makes it easier to figure out who are the slackers you need to toss out.melgross - Monday, July 29, 2019 - link
The truth is that there has never been a whole lot of that. Also, it hasn’t worked out as well as hoped. It turns out that true face to face interactions are better.Samus - Saturday, July 27, 2019 - link
This is the most ridiculous statement I've ever heard. You're telling me there is a corporate trend toward less autonomy and portability because of...device theft?Enterprise packaging and volume sales numbers don't back that statement up. Desktop PC's have been declining (5% loss YoY) for a decade, while mobile PC's have been basically neutral ( less than 10% loss of growth over a NINE year period, mostly offset by tablet sales) while projections show desktop sales to continue falling and mobile sales to begin rising as has been the case lately since the introduction of Coffee Lake - a milestone for laptop performance improvements (twice as many cores, over 30% productivity performance improvement, for the same amount of power)
AT won't let me publish the Statista link so you will just have to Google it for reference.
Marlin1975 - Saturday, July 27, 2019 - link
Device theft is one of the excuses companies are using in some places. "Synergy" or team environment is also another term being used to tell people they can't work from home or remotely. IBM is using that as an excuse to get people to quit so they don't have to pay severance or report layoffs.FreckledTrout - Monday, July 29, 2019 - link
Honestly it's just going to be poor upper management at this company t hat is moving towards desktops. The worse the management is the closer they want their employees and the more micromanaging you will see.Oliseo - Tuesday, July 30, 2019 - link
"Honestly it's just going to be poor upper management at this company t hat is moving towards desktops. The worse the management is the closer they want their employees and the more micromanaging you will see. "For the life of me I cannot fathom why people come out with this guff and expect not to be challenged on it.
I think what's happened, is there are so many people talking absolute BS on the Internet these days, that people don't get pulled up, and then carry on believing they're correct.
What's actually happened is that YOU have had a run in with someone in YOUR upper management, and so because people are generally stupid, you extrapolate YOUR personal experience to being factual accross the entire board.
Don't worry though, you're far from being alone in this. It's the Facebook Generation, they are all guily of it somewhat. From "Journalists" at (Once) "Respected Publishers", to glass collectors, to Police Officers, and so on.
I call it the "Pub" test. Could you say what you did in a pub and get away with it. The difference being that in a Pub it would be YOU who has to walk out in shame for talking absolute BS after people publically correct you.
Whilst online, you can surround yourselves with other idiots who are prepared to agree with whatever BS you come out with so long as you do the same in return (AKA the Echo Chamber).
Still, if you were to correct every last soul on the Internet whose chatting BS, and saying their personal experience as absolute facts, you'd need more time than the universe is able to provide.
So, I leave in it the hands of whatever religion is in fashion this year, pray to those volcano Gods...
AlyxSharkBite - Friday, July 26, 2019 - link
Totally irrelevant when the PC Gaming Market makes more money than the Hollywood Movie. (Not counting consoles) yep totally irrelevant. But please do tell me more.azfacea - Friday, July 26, 2019 - link
the last guy that thought desktop was irrelevant was IBM wonder what happened to themJorgp2 - Saturday, July 27, 2019 - link
https://www.marketwatch.com/investing/stock/ibmI don't know, you tell me
willis936 - Saturday, July 27, 2019 - link
I couldn't figure out how to get that site to show history back to the 80s, since that's when they lost to compaq and intel. They had the entire market. Can you imagine if there was a microsoft of hardware today? Some stock figures from their business machines tucked away in a warehouse say nothing of what they lost due to hubris.willis936 - Saturday, July 27, 2019 - link
Whoops just found it. They are worth more now than they were in the 80s, but that doesn't say what they could be had they not dropped the entire PC market.Kevin G - Sunday, July 28, 2019 - link
If you want to play what-if like that, imagine if Intel had gotten the CPU contract for the iPhone and some what buckled to Apple's low margin, high volume demand for that market.Another fun what-if is AMD buying nVidia instead of ATI.
Intel still had several Netburst design waiting in the wings before there right turn at Core 2, so what if they continued down the Netburst path?
Last fun what-if: Intel was actually on time with Itanium and did manage to get vendors to push out 64 bit workstations/highend PC's and AMD still released their x86 extensions.
Lord of the Bored - Monday, July 29, 2019 - link
IBM didn't have the entire market in the 80s. Not minicomputers, not personal computers. There was very robust competition from an abundance of architectures. Even more if you include home computers instead of just business machines.That said, they did far better than many of their fellow old-guard computer companies because they insulated their personal computer division so the minicomputer division couldn't sabotage it. There's a world where we all use machines derived from DEC's PDP-11, TI's TI-990, or Data General Nova. To name but a few of the contenders that self-sabotaged. I think IBM is the only company with a minicomputer division that successfully made the transition to personal computers. Most that tried never shipped a micro, and those that did shipped horribly neutered machines at the demands of the minicomp divisions.
Santoval - Saturday, July 27, 2019 - link
Which is why AMD will release devices sporting Zen 2 based APUs by the year's holiday season as well. On the CPU core side Ice Lake-U/Y are going to be *barely* faster than the Whiskey/Amber Lake CPUs they are to replace. While they have a higher IPC (not 18% higher, that's over the original Skylake of 2015 - *not* 2016 btw) most of the IPC gains will be eaten away by their much lower clocks. The much lower clocks, in turn, are due to their 10nm+ node's atrocious yields.Extrapolating from the graph Intel released (they only compared Ice Lake-U 15W to ... Broadlake, in just one graph, in an obvious attempt to obfuscate the numbers) they are going to have a 3 or 4% higher single core performance, 15W for 15W. At best, and if the Sun always shines on them, a +5% performance tops. Unless they trimmed a lot of their "multi-core overhead", which I doubt, the multi-core performance gain will also be comparable, in the 3 - 5% range. If this was a mere refresh the performance gain would be fine, but since we are talking about both a new μarch and a new process node it is truly pitiful, which why Intel did their best to mask Ice Lake's CPU performance deficit in their Computex presentation (they masked it with iGPU, neural network and ..... wi-fi performance numbers, "forgetting" to talk about CPU numbers at all in their main presentation; they released them later to the press in an obfuscated form, almost as a footnote).
Nevertheless, despite their CPU deficiency, Ice Lake-U/Y will also sport a new iGPU, a new media engine with more codecs and higher resolution support, support for Optane memory, and native Thunderbolt 3. The new iGPU, in particular, is probably the most dangerous for AMD since until now they had a clear edge on that era. Intel will need to be quite aggressive in their iGPU deployment for that to pose any threat though, and not retain the 48 core and 64 core iGPUs just for the high end SKUs. I do not expect that to happen, with Intel still being Intel here. The 64-core iGPUs in particular will surely be super rare, just for 28W parts with eDRAM.
In any case AMD need fast and capable Zen 2 based APUs in order to compete. They will probably not exceed 15W TDPs either, so they will have a clear gain in power efficiency as well. Intel's "abnormal" TDPs are also due to their yield issues. Fanless devices with 9W Ice Lake-Y parts are probably still possible, but they will require quite sophisticated -and more expensive- passive cooling solutions.
Kevin G - Sunday, July 28, 2019 - link
Using Sky Lake as baseline is fine as they are still effectively on that micro-architecture. The biggest changes in over the past couple of years have been from an increase in shared L3 cache size (a side effect of higher core count) and the patching of security vulnerabilities.The real comparison is what the base clocks are vs. turbos.
Ice Lake does have lower turbos from the information that has leaked thus far but I have yet to hear much about base clock speeds, especially when under AVX or AVX-512 modes. This may not be as dire if Intel can keep up average clock speeds similar or higher than Coffee Lake and ride the IPC increase for the generational performance gain.
I would expect Optane DIMM support in the memory controller but I highly doubt that Intel would genuinely enable that in consumer systems. (Xeon E is a different question.)
The parts with eDRAM will likely find a home inside of MacBooks, like what happened with similar Sky Lake, Kaby Lake and Coffee Lake designs. The real radical was Kaby Lake-G with its Vega graphics as I am really shocked that that hasn't been picked up by Apple.
Gondalf - Sunday, July 28, 2019 - link
For now they have their present Ryzen Mobile 3000 line.There isn't news about APUs on 7nm right now but we can argue that the producty cycle is minimun one year, so we will not see 7nm APUs since Q1/Q2 next year.
Remember that Intel has Tiger Lake for 2020 with the far more efficient Xe ark on GPU, so AMD has to react fast to not lose the small market share in Laptops.
XiroMisho - Saturday, July 27, 2019 - link
And Gamers Nexus made an amazing discovery:Ryzen CPU's undervolt to a stable level as low as 1 volt -- they maintain 3Ghz at 1v, yes, but they are stable at super low voltages... meaning AMD doesn't even need to spin new silicon for mobile chips, they just need to work with board partners to make low profile sockets.
Jorgp2 - Saturday, July 27, 2019 - link
So you're saying they're worse than Intels chips, which can go below 1v?regsEx - Sunday, July 28, 2019 - link
But with massive performance degradation.PeachNCream - Saturday, July 27, 2019 - link
So true. I haven't seen a desktop PC in a workplace in ages. Most desktops seem like either rendering or heavy lifting workstations or are delusional slap-together machines people think makes them better at vidya games.lmcd - Sunday, July 28, 2019 - link
my silverstone case is beautiful and u can stfu m9!!Refurb Dell Haswell i7s run for under $300 so idk, seems fine to me?
Arnulf - Saturday, July 27, 2019 - link
2014 is calling and wants its tablet fad back.Noone considers mobiles, iPads or Chromebooks for serious work.
name99 - Saturday, July 27, 2019 - link
Seriously? You want to make this comment in a thread where most of the arguments boil down to “which one plays games better”?OK...
willis936 - Saturday, July 27, 2019 - link
It comes down to ergonomics. A full keyboard, mouse, and large monitor is much cozier than a tablet or laptop for 8 hours of work a day.You could dock your laptop and be able to move it around or you could get a higher performing desktop for the same price. There is still value to desktops to businesses.
29a - Saturday, July 27, 2019 - link
I'm not sure irrelevant means what you think it means.pavag - Sunday, July 28, 2019 - link
Laptops are a serious financial risk for companies, because they have zero ergonomy.A monitor has to be in front, and same level as the eyes. No Laptop can do that, so they all cause neck injuries in long term.
Any serious company should ban the use of laptops. Even personal ones.
lazarpandar - Sunday, July 28, 2019 - link
Erm.. most laptops are used docked.Oliseo - Tuesday, July 30, 2019 - link
"Desktops are irrelevant."Proof please, otherwise I can safely assume your some kid in mummies bedroom pretending to be a grown up.
khanikun - Tuesday, July 30, 2019 - link
Some companies have moved to thin clients. We use Dell Wyse DX0D thin clients for like 90% of our work force. Roughly 2k people. Which still use the low tier Intel or AMD procs. Like Atom or something.I also don't see how both Intel and AMD are irrelevant on the mobile side, cause they're both still in your laptops, ultra books, Chromebooks, etc. Android and iPads is where they aren't. Sure, everyone has some kind of Android or iOS phone, but not so much Android tablets or iPads.
Most still rock some kind of cheap PC laptop. Especially when you can get them for a lower price than many Android tablets or iPads.
eek2121 - Saturday, July 27, 2019 - link
You can always benefit from more cores, especially if clock speed isn't sacrificed. Right now just browsing the internet and having a few background apps open I have 229 processes and over 3000 threads. I get occasional small spikes on every core + smt from those processes/threads. The system is smooth and responsive. Usually the guys saying that 16 core CPUs (or 8 core for that matter) the the same guys running quad core CPUs, SATA drives, and low end GPUs.LogitechFan - Saturday, July 27, 2019 - link
Yeah, because we all know that browsing on something with fewer than 16 cores is a fucking nightmare!jospoortvliet - Saturday, July 27, 2019 - link
Maybe not and 16 might be overkill for browsing but websites get heavier and browsers are building multithreaded browser engines - Firefox is halfway there and the rest will follow.azfacea - Saturday, July 27, 2019 - link
it depends on what you are browsing. saying browser does nt need 16 cores is like saying .net programs dont need 16 cores. what does javascript have to do with core countbobhumplick - Friday, July 26, 2019 - link
those are going to be a big step up from previous mainstream chips for mutlithreaded work. but its disappointing that a chip as old as a 8700k or even a 7700k has the same single thread performance when either chip is clocked to its max. ill declare victory for amd when an intel and amd chip, both clocked at their max, and the amd chip wins in both. but right now they really are pulling ahead. but you have to realize its not just because they are doing so well but mostly because intel are sidetracked. amd fully expected to launch these chips and be just behind intel. and if intel had gotten sunnycove and 10nm out they would have been.eek2121 - Saturday, July 27, 2019 - link
We haven't seen benchmarks from the 3950X yet, and AMD has AGESA issues to sort out. You will also note that coffee lake was Intel's last big hurrah. The 9900k simply features 2 additional cores and a higher single core boost. No IPC increase. Intel has hit a wall. Note that I'm not a fanboy at all, simply acknowledging the truth. I also take issue with Intel (and now AMD to a certain extent) are lying about TDP.jospoortvliet - Saturday, July 27, 2019 - link
You just read an article where intel claims a 18% IPC increase in their latest chips. They are not yet for desktop and don’t clock well, granted, but see at what speeds the first 14nm chips hit the market and see what they do now. Then you know where 10nm is going.Amd and TSMC will have to ramp up clock speeds on their 7nm quick. I expect a refresh next year with better clocks, I hope at least 500mhz on top of what the current chips promise-but-do-not-deliver... otherwise intel might come back quick.
Rudde - Saturday, July 27, 2019 - link
AMD roadmap has Zen 3 with TSMC N7+ node slated for next year. AMD could be focused on getting higher clocks, but I'm more inclined to believe they will improve IPC instead.Targon - Monday, July 29, 2019 - link
AMD will do both. AMD is working on core design improvements while TSMC improves its fab process. It's not one or the other, unless you do what Intel did, and link design improvements to the fab process, which has severely hurt the company. Why Intel couldn't have put a new design on 14nm is a very good question that I have not seen the answer to.Icehawk - Sunday, July 28, 2019 - link
8th gen is barely a year oldTEAMSWITCHER - Saturday, July 27, 2019 - link
And most people will be running Cinebench R20 benchmark on those cores and ... not ... much ... else.Targon - Monday, July 29, 2019 - link
When you see benchmark results, you see them done on systems that are very clean, with a new install of Windows, and then the test suite. As a result, these benchmarks are NOT being run on the primary systems that are being used for office work. If you have Steam, Origin, Antivirus, plus other things, you are already putting a greater load on the system than what is on systems that are posted for benchmarks.The result, you see less of an advantage to additional CPU cores than people do in the real world.
twtech - Monday, July 29, 2019 - link
I'm hoping the 32C TR3 will not be gimped memory-access wise like the previous generation. I hope it also can hit around 4GHz turbo.Dark42 - Friday, July 26, 2019 - link
So the Max Turbo of 4,1 GHz is indeed much lower than previous generation, how about the base clocks?bobhumplick - Friday, July 26, 2019 - link
the clock are always lower in the first wave of an intel process. the first process is always about power and density. then the second wave, the plus, (14nm+,10nm+, etc) is about higher clocks but sacrifices is a bit of density to do it (for instance 14nm++ is 35 million transistors per mm2 while 14nm was 45, incidently amd 7nm is rumored to only be about 40-55 by the time amd actually uses its own libraries and adapts the 7nm tsmc process to its own needs).so yes, just like broadwell was the first 14nm and sacrificed a lot of clocks so will these. and just like broadwell these chips are mobile focused so clocks dont matter as much, only power.
once intel finally rolls out 10nm for desktops we may see lower clocks but not like these chips. i wouldnt be surprised if a max oc is 4.6-4.8ghz ish. could go as low as 4.4ghz. but then again they could still be reaching 5hgz. maybe thats why they are putting off 10nm desktop cpus for so long. that and cost
i would expect that the next move for intel is to either put some form of backported 14nm sunnycove out which is what i think rocketlake (the 11th gen is). if they could get the same clocks they are getting now and combine it with 18% ipc gains then that would be a competitive chip. even if they only make core counts similar to what they have now they may lose in multithread performance still but for most average consumers lightly threaded defines performance still.
but the sunnycove cores are quite a bit bigger so how do they fix that on 14nm? well they could leave off the igpu and use that space. the igpu on current chips already takes up the exact amount of space as 4 cores. and coincidently rocket lake is listed as having seperate igpu and cores on intel slides. so if they used that space to make these larger cores in the same amount of space they wouldnt really cost any more unless you bought a version equipped with an igpu. current kf cpus with no igpu actually still have an igpu but its just turned off. so no space savings there.
intels 14nm is actually quite cheap by now. their 14nm cpus cost so much now because a) they still demand a premium price because they are still good chips b) the most important part is that they just have a shortage. they cant produce enough.
if intel move igpu and chipset production to samsung, plus move some of their line over to 10nm for mobile and maybe expand some 14nm production where they can then they could afford to make an 8 core 16 thread sunnycove based cpu and it would cost no more to make than an i9. actually with no production shortfall they could make it the next i7 and sell it for well under 400 dollars. that would be a fairly competitve cpu except in production workloads where amd would win by a good margin.
this would hold them over till they could get 7nm EUV going and then skip straight to the less dense, higher clocking, + version of 7nm for desktop.
Kevin G - Friday, July 26, 2019 - link
This is actually the second 'wave' of 10 nm parts. Cannon Lake was the first which actually did ship but in chronically small volumes due to manufacturing issues.It should be noted that for desktop, Intel has not committed to bringing 10 nm parts to it in 2019 or 2020 with the possible exception of HEDT. Thus far Ice Lake is for mobile and server which carries them until 2021.
Rocket Lake could leverage some of Intel's new packaging technologies but that too is currently targeted at mobile where physical space constraints can justify the price premium and server for scalability. Much like the lack of 10 nm desktop chips on their roadmap for 2019/2020, these advanced packaging technologies will likely come to desktop last.
Intel is pretty much treading water until 2021 but their line up for that year on their road maps does look very interesting.
YoloPascual - Saturday, July 27, 2019 - link
I bet 500 bucks we will not see desktop 10nm at all.Rudde - Saturday, July 27, 2019 - link
It would benefit Intel a lot more to port 'willow cove' to 14nm. It is their 2020 architecture and any backports are unlikely to arrive before 2021 (Comet lake in 2020). Intel will have a hard time defending their single core performance advantage in 2020. They will have to make a big advancement in 2021, how they'll do it is speculation at this point.damianrobertjones - Monday, July 29, 2019 - link
I do like the information but, of course, the lack of capitals confuses me?Targon - Monday, July 29, 2019 - link
If you buy a new, "top of the line" product, and it isn't better than the previous generation products, you will be disappointed. TSMC 7nm is a new fab process, and it's better for AMD than the 12nm parts in clock speed, so claiming that the new fab process should be forgiven for not being as good if there is zero benefit to the user is a bit of a joke. Intel has been claiming that 10nm is on track since 2015, and still isn't better than 14nm.SaturnusDK - Friday, July 26, 2019 - link
Base clock is 1.3GHz for the i7.eek2121 - Saturday, July 27, 2019 - link
Eventually they'll get things ramped up. The big issue with node shrinks are increased power leakage, which means lower clocks. They've been on 14nm for many years so they've had time to optimize their designs. Once they get a few more generations underway, you'll see some clock speed improvements. As to the base clocks, I can't remember exactly, but they were much lower. That is mainly due to the fact that these are laptop chips.name99 - Saturday, July 27, 2019 - link
“Once they get a few more generations underway”So they’ll have competitive chips in, I don’t know, 5 yrs from now? (A few more generations...)
Good thing Apple and AMD plan to take a break over the next five years...
(BTW if Apple manages to get just 20% more single-threaded performance out of A13X — which is below there annual trend ever since the A6 — they’ll already match the BEST Intel has to offer that’s not overclocked...)
ikjadoon - Friday, July 26, 2019 - link
Is it too early to complain about the model names? Maybe, but if you want to prime youreslf before Intel sets their silly narrative.To: Intel
From: the rest of us
Subject: what the hell
These laptop SKUs all release in six months:
i7-10710U (Comet Lake)
i7-1065 G7 (Ice Lake)
i7-10510U (Comet Lake)
i5-1035 G1 (Ice Lake)
i5-10210U (Comet Lake)
i3-10110U (Comet Lake)
i3-1005 G1 (Ice Lake)
Somehow, you kept i3-i5-i7 but decided the rest should be a soup of letters. Who are you, HP?
Farfolomew - Friday, July 26, 2019 - link
As much as I tire of the ‘x’ moniker, in this instance it actually makes sense. Why not call them ‘i7-X710’, ‘i5-X510’, ‘i3-X210’ etc0iron - Sunday, July 28, 2019 - link
And after X, XI or 11? Maybe another dash? i7-10-710Duncan Macdonald - Friday, July 26, 2019 - link
So Intel have finally got a 10nm chip - however it is a small chip with a maximum of 4 cores. As Intel do not use the equivalent of the AMD chiplet design, the acid test will be how long it takes Intel to produce a 10nm equivalent of its current high end designs (16-28 cores). I personally do not expect volume deliveries of 10nm high end CPUs from Intel before 2021 (possibly even 2022).(Intel high end 10nm chips may also have thermal problems for Intel to overcome as the chip would end up with about half the surface area but the same power dissipation as current generations - this is not a problem for AMD as the chiplets are physically separated so no single hot spot exists.)
Rudde - Saturday, July 27, 2019 - link
Ice lake server is set for 2020 on Intel roadmaps. I don't know about core counts though, as they also have Cooper lake for 2020.johannesburgel - Friday, July 26, 2019 - link
According to everything I've heard, Intel's 10nm process has the following issues:- The new metals they're using (Cobalt etc.) have different thermal expansion coefficients, putting more thermal strain on the chip.
- Cobalt has a much lower thermal conductivity than Copper, basically acting as an insulator, so hot spots inside the chip become even hotter. This puts even more strain on the chip and doesn't allow high clock speeds.
- Cobalt is brittle and doesn't handle strain as well as copper.
This means the 10nm chips are much more likely to die from thermal fatigue than their 14nm models. Which is probably why they are having yield issues - the chips are working when they come off the production line, but most of them would die too early in the hands of customers.
I'll stick with AMD while other people are playing guinea pigs for Intel's 10nm.
NirXY - Sunday, July 28, 2019 - link
first.. they are using cobalt interconnects instead of tungsten.secondly, cobalt is the standard the entire semi-conductor industry will eventually switch to, in order to lower the contact resistance and reduce thermals.
Gondalf - Monday, July 29, 2019 - link
Umm yes you are absolutely right. Officially Samsung too is going with Cobalto to increase performance, same applies to TSMC.IMO the 10nm "issue" is only related to yields not to absolute clock speed.
If you consider actual TSMC painful situation, you can realize that right now they are unable to
manufacture in volume a large die in 7nm, they can give only a very low volume output with some overheat around. If you look at the 7nm process it is limited on clock speeds, virtually unable to overclock and pretty hot at the silicon level under load.
The real test on TSMC 7nm is Nvidia, but all journalits are silent about this bashing instead Intel all around. In a recent statement Nvidia said that 14/12nm is a better choice for the Company.
Nvidia said that 7nm is too expensive to sustain and for now they will stay on 12nm more like is doing Intel (this is a July 2019 thing)
I'm asking to myself what is the real AMD net profit over Ryzen 3000 line, expecially over the large footprint 12 and 16 cores SKUs. Considering the likely less than 50% yields level of these high turbo dies, they have to pay two to have one or four to have two good enough. Not to mention the very high packaging costs to assemble three dies with fast data links on the same substrate.
The situation is far different from the stellar 90/95% yields level of 14nm or 12nm.
joconor - Friday, July 26, 2019 - link
Still no LPDDR4?!?!smilingcrow - Friday, July 26, 2019 - link
The slide suggests it is supported as do other reports and up to 3,733.repoman27 - Friday, July 26, 2019 - link
Yeah, LPDDR4/LPDDR4X-3733 are supported. The slides are a little wonky in how they list that, but it’s pretty clear if you read the product brief: https://www.intel.com/content/www/us/en/products/d...name99 - Saturday, July 27, 2019 - link
Of course DDR5 is now shipping, and the 2019H2 high-end phones will probably be using LPDDR5.So, yeah, once again awesome forward planning Intel! Proudly shipping yesterday’s specs tomorrow.
Jorgp2 - Saturday, July 27, 2019 - link
As opposed to AMD supporting Ddr4 2400?name99 - Saturday, July 27, 2019 - link
They’re both dinosaurs.Gondalf - Sunday, July 28, 2019 - link
I fully understand you. The resurgence of AMD have badly cancelled any ARM hopes to penetrate the very lucrative Laptop segment and absolutely destroyed the so much hyped server adventure of this RISC ISA.Funny enough AMD is best Intel Friend. Who is ARM outside phones.....nothing.
Meteor2 - Sunday, July 28, 2019 - link
Yet.stockolicious - Monday, July 29, 2019 - link
@Gondalf i think your right and IBM Power also takes a hitname99 - Wednesday, July 31, 2019 - link
There are multiple issues here.First is the difference between those of us looking forward and those people looking backward or wanting to buy something today.
Second (and this is SO important) is the issue of what people like you would count as an "ARM laptop". Apple will (probably) be shipping these soon, (probably) by end of 2020. But would you count that as an "ARM laptop"?
For anyone apart from Apple, there are basic issues:
- an OS is needed. So who provides that?
+MS, the company that has NEVER been able to figure out its OS strategy since at least .NET, still has no clue exactly what it wants. Does it want x86 to be "real" Windows and ARM to be "protected Windows"? Or is ARM supposed to be "consumer Windows"? Or is ARM supposed to be "Windows that never pisses of Intel"?
+ how about Linux? Oh, you mean we'll FINALLY see the year of the Linux desktop? There's a reason no-one plans to get rich selling Linux desktops, and that ain't gonna change.
+ Android? Well Android or ChromeOS? Just like MS, WTF knows what Google's OS strategy is long-term?
And of course in all these cases, what do YOU, the complainers, ACTUALLY want. Half the time you seem to want a super-cheap chromebook like ARM laptop. Then, when those are available, you complain that they don't have the nice screen, touchpad, battery, flash, etc of an Apple device, suggesting you want a ~$1000_ ultrabook...
I don't think ANYONE apart from Apple can provide anything useful in this space.
Which means we wait for Apple to ship, then see how excited the market is about the results compared to current MacBooks. (Noticeably faster? Noticeably longer battery life? Noticeably innovative features?) Then we wait for MS and/or Google to slowly pivot to copy all that.
The other dark horse would be a Chinese company creating a very cheap ChromeBook like thing. But once again there is the OS problem which doesn't just go away because China. Sure, they can slap some horrible Android-kludge on it which never gets updated again, and the result might even be desirable and useful for the poorest parts of the world, better than nothing. But that doesn't provide what people like you mean when you ask for an ARM laptop.
So yeah, for the next five years, it's basically Apple (hopefully in a year!) or nothing...
repoman27 - Friday, July 26, 2019 - link
“The new 300-series chipsets for ICL will natively support Thunderbolt 3...”Actually, Thunderbolt 3 is integrated directly into the ICL CPU die, not the PCH, so it’s not really a chipset feature per se.
nevcairiel - Saturday, July 27, 2019 - link
Mobile chips have a on-die chipset, so in this case its really quite the same.... :)repoman27 - Saturday, July 27, 2019 - link
U and Y series have an on-package PCH, not on-die (the CPU and PCH dies are discrete but part of a multi-chip package). And H series is still mobile but the PCH is a separate package. So not the same at all.The point being that Intel did not integrate the Thunderbolt controller into the PCH die, but into the CPU die itself.
Ian Cutress - Saturday, July 27, 2019 - link
You're correct. Unfortunately Intel's own diagrams from this info (which Anton was working from) are themselves incorrect in this case, compared to the ones from Intel's one SoC deep dive event.Khilos - Friday, July 26, 2019 - link
I understand the concept of managing inventory, but if Intel needs to start shipping CPUs in Q1 in order to stock up for a holiday Q4 launch, something seems off. Are yields that low or production capacity that limited that such a long lead time is needed ?eastcoast_pete - Friday, July 26, 2019 - link
No, chips have to be shipped several months before the launch so that people can buy the actual products (fully assembled laptops ready for sale) on the official launch day.repoman27 - Saturday, July 27, 2019 - link
No, there is nothing normal about this product ramp / launch from Intel.They considered ICL “announced” and “shipping to OEMs” as of May 27, 2019 in order to book 10nm revenue and convince investors they were hitting their targets. And despite the devices “on shelves for the 2019 holiday season” messaging which implies the sales embargo being lifted as late as Q4, leaked Lenovo and Dell roadmaps showed ICL products as early as June and August respectively.
But Lenovo already shipped those products with Whiskey Lake, and Apple just refreshed the MacBook Air and 13-inch MacBook Pro with Amber Lake 2+2 (same renamed Kaby Lake as before) and off-roadmap Coffee Lake-U 4+3e CPUs. We also know that Ice Lake-U/Y will be sold alongside Whisky Lake-U, Amber Lake-Y, and Comet Lake-U/Y. It is looking increasingly like HP and Dell might be the only major OEMs to feature Ice Lake in flagship products this year. In other words, Intel was only able to supply 2 of the top 4 OEMs with sufficient volumes of chips for them to ship key products in 2019.
Intel has dilated the launch window massively here by claiming they were "shipping" during what would normally be considered the qualification sampling period. And the thing is, they probably did get the OEMs to pony up the cash already, so now the OEMs are sitting on Intel's entire output of 10nm chips, bought and paid for, while they work out any problems and wait months for the sales embargo to lift.
eastcoast_pete - Friday, July 26, 2019 - link
Good! Now, let's see what AMD has to offer in the low and lower power mobile space, and see who's on first! Whoever gives me the best bang for my buck gets my orders.jospoortvliet - Saturday, July 27, 2019 - link
I’m afraid amd has no plans short term bringing their new cpu and gpu architectures to mobile... the upcoming app is still last gen cpu and gpu. Too bad, given the size of the mobile market, but I guess amd expects to use most of their manufacturing capacity for the enterprise market...stockolicious - Monday, July 29, 2019 - link
@Jospoor - Navi is their attempt to scale up and down - and the Samsung partnership is huge when you think in terms of mobile and scale for easy royalty payments.Rudde - Saturday, July 27, 2019 - link
It is wierd that AMD is a generation behind on mobile parts and Intel on desktop parts, but that's how it is. AMD will not have 7nm mobile ready for holiday season, while Intel will not have 10nm desktop ready.eva02langley - Saturday, July 27, 2019 - link
AMD mobiles chip will arrive in the next 6 months.Teckk - Saturday, July 27, 2019 - link
Did AMD announce this at Computex?urbanman2004 - Friday, July 26, 2019 - link
A/b freaking time Intel, took you long enough 😂wr3zzz - Saturday, July 27, 2019 - link
Y-series 10th Gen is now 9W? Can you do fanless at 9W?Xyler94 - Monday, July 29, 2019 - link
Technically, yes. But Intel uses base clock as TDP, not nominal boost. So to get best performance, a better cooling system is bestazfacea - Saturday, July 27, 2019 - link
intel is lying again. 10nm has launched before. i'll believe it when i see iteva02langley - Saturday, July 27, 2019 - link
4 cores laptop chips...eva02langley - Saturday, July 27, 2019 - link
And 4.1 GHz...azfacea - Saturday, July 27, 2019 - link
actually only 3.9 GHz in laptop SKUsyannigr2 - Saturday, July 27, 2019 - link
Oh, look, 4 cores. We are back at 2009 again.Ironchef3500 - Monday, July 29, 2019 - link
+11_rick - Saturday, July 27, 2019 - link
If I remember correctly, those Ryzen chips drastically dropped performance when undervolted so much.name99 - Saturday, July 27, 2019 - link
That’s not the point. The point, I think, is that it suggests that the existing AMD chips provide a fairly easy path to laptops as soon as AMD wants to make that move. At least that was my interpretation of why the issue was being raised.1_rick - Saturday, July 27, 2019 - link
"You just read an article where intel claims a 18% IPC increase in their latest chips."That's 18% *over Skylake*, not over the 9th gen.
repoman27 - Saturday, July 27, 2019 - link
Let me just pull up my graph of IPC improvements going from Skylake to Kaby Lake, Kaby Lake Refresh, Coffee Lake, Coffee Lake Refresh, Amber Lake, Whiskey Lake, and Comet Lake:o—o—o—o—o—o—o—o
Doesn’t matter which “gen” you choose from, 7th, 8th, 9th, or even 10th, Skylake cores all have the same IPC. Cannon Lake did show some modest IPC gains, but seeing as only 10K or so units ever shipped, it’s not really a meaningful comparison.
konbala - Saturday, July 27, 2019 - link
Want to buy a laptop and I want Ice Lake but it's almost half a year away. Painstaking.zepi - Sunday, July 28, 2019 - link
Where the hell is the LPDDR4 support?MASSAMKULABOX - Monday, July 29, 2019 - link
ALl these lakes .. and Intel still treading water . Rocket lake ..emibs ..multi-chip pacage?tipoo - Monday, July 29, 2019 - link
In Ice Lake, shipping now, as stated by this article?Gunbuster - Monday, July 29, 2019 - link
No touted xtreem battery improvement this spin around? No Doubt battery's will still shrink by something like 5% to insure we are tossing these in the trash and crawling back for more in two years time.