if Ice Lake basically traded blows with Amber Lake (or was it Whiskey Lake? - seriously Intel, use a new second noun in your codenames, we're tired of lakes!), how is it going to compete with Comet Lake?
Just like the article mentioned, it will have a weaker CPU performance watt for watt (and *much* weaker if we include the 6-core part in the comparison) but stronger iGPU performance. Ice Lake also has a better video encoding engine, faster and more efficient wireless (which you actually require multiple devices to take advantage of, i.e. have a WLAN with 3+ routers all supporting Wi-Fi 6) and better AI inference performance.
In other words if you need both the best 6-core 15W CPU with the best iGPU you need to wait a few months for the upcoming Zen 2 based APUs by AMD, which will also have faster Navi iGPUs. Let's just hope AMD upgrades their memory controller to LPDDR4x-3733 as well. While it has not yet been officially confirmed AMD will almost certainly move to 6-core APUs as well as the max option. TSMC's 7nm node allows them to do that with a smaller die than Intel's 6-core Comet Lake-U.
I'd say quite realistic though. Ryzen 4*** mobile should use Zen2 architecture and 7nm. 4C/8T should be the entry level, although for ultra-low power they may go to 2C/4T, but I doubt it. I hope to see an 8C/16T for 15W: seems doable, seeing that intel gets to 6C/12T with a 14nm and a design not optimized for high core count.
7nm is going to operate in mobile much better because you don't need frequencies that high. It is going to be easy to compete with Intel's "6 cores" operating @ 1.1GHz each. 7nm is going to 3GHz in freaking PHONES.
Current Ryzen Mobile CPUs have trash battery life. Hopefully they can fix that with Zen2 mobile. I'm eyeing a Dell 13 XPS aio with the Icelake for the insane battery life.
Most of those issues are on the OEMs. The processors themselves have reasonable draw. Intel is far better at working with OEMs to optimize the whole system for battery life.
Put down the crack pipe, old man. *Existing* APUs are 5th gen, Polaris is 4th gen. So at worst they would CONTINUE to use 5th gen "Vega" graphics, albeit with a new process and likely faster memory.
"In other words if you need both the best 6-core 15W CPU with the best iGPU you need to wait a few months for the upcoming Zen 2 based APUs by AMD, which will also have faster Navi iGPUs."
Well said, 7nm Zen2 APU likely in Q1 2020 (demo at CES 2020), a simple common sense from what happened in CES 2019.
It's 3.9 All Core Turbo, that's the frequency you will most of the time, and is in fact quite impressive. Most OEMs will have decent cooling and it can stay at those frequencies for long time.
The base frequency is what heavy LINPACK will run at if you run it for hours and restrict power consumption at no more than 15W, but non-LINPACK will run at higher freqs even for long term. So, no sham.
Intel can go Jump-in-a Lake... they're obfuscating to make up for their F-ups and to ensure no one has any idea what processor is what and just think they're getting the latest and greatest
How can you forget it, though. For a lot of people who need a laptop this year, Comet will be the best option.
If you can wait, hopefully AMD made some big bets about their next APUs. Currently, Ryzen Mobile is more competitive, but it not the aggression that they have on desktop and in server.
I've been looking at some of the laptops being offered with AMD processors.. It's nice to see Asus bringing out some interesting models with 120HZ panels and all the little goodies one would expect in a higher performing laptop. I've been thinking of buying one since the prices are pretty good overall.
I wholeheartedly disagree. Being still on 14nm it means that they are only anecdotally better that the previous generation (at least in terms of performance per watt). The Ryzen 3*** mobile are one generation behind the desktop, still on the 12nm (which I think we can safely compare to Intel's 14nm+++). The complexity of Intel's offer is mind boggling, and completely useless.
I can get a 4C/8T Ryzen 3700U (15W), with Vega 10 (https://www.anandtech.com/show/14784/asus-launches... The CPU should be on-par, or close to Intel's top of the line (except for the 6C/12T, which is interesting) and the GPU will run circles around the 620. The Ryzen 4*** mobile should be based on Zen2 and 7nm, and it should trash any 14nm offering from Intel in terms of price/performance.
Theoretically, yes, but so far Intel has been ahead of the game in the mobile platform.
They're getting excellent speeds and power usage out of 14nm and on that front I'm not sure the 12nm process can compete. On desktops, absolutely. Super low power usage, not sure.
There is a glaring lack of HQ processors from Intel and the highest I've seen from AMD was 35W.
I've still got my 45W Skylake i5, which to be fair almost never goes above 30W, so it could just be they dialed back the "standard" or whatever you want to call it.
I agree with you: Intel's offering today is better than AMD's on 12nm, for sure. That said, I have a Core i7 5500U (2C/4T) and I cannot wait to move to - at least - a 4C/8T. I am holding off to see the Ryzen 4***, because the feeling is that it will wipe out anything 14nm.
Intel will keep on outputting 14nm based parts for the entire 2020, at least, and they'll probably keep on doing it for the first half of 2021 as well. Ice Lake and Tiger Lake are both expected to be dual release nodes (10nm + 14nm). After Tiger Lake (in 2H2021) Intel *might* have fixed their mess. Or they might have not and 10nm and its variants will be dual release nodes until their end of life.
forget 1.1 ... man, Intel's marketing is asleep! ... people are looking at "1.1" and think it's relevant ... it's 3.9 GHz All Core Turbo - "Turbo" can run for minutes on end on a laptop OEM did not screw up with ... and Intel makes sure, for the most part, they don't ...
As per ark, neither i7-10710U (6-core) nor i7-10510Y support Thermal Velocity Boost while i7-10510U (4-core) does. Could you comment whether ark is wrong on that one? This is likely some more hundreds MHz of difference.
Ryan, based on your experience (and any other AT editor that might care to weigh in) is there any number claimed on those slides that users should take with a grain of salt, or require some clarification? The question was triggered by the +41% in Office 365 scenarios which lacks any kind of disclaimer (from Intel and AT) and suggests it will be generally the case for *any* O365 workloads.
All of the Intel performance numbers should be taken with a disclaimer - or at the very least, you'll want to read the footnotes. Intel is doing 25W vs. 15W here, so it's not an apples-for-apples comparison in terms of power limits.
As for the O365 scenario in particular, it's measuring "the time it takes Microsoft Office to perform 3 tasks in a multitasking scenario (export PowerPoint file to video, export word doc to pdf and excel calculation).
Does intel make mistel on the document they released? On their product web site, Comet Lake-U does not support LPDDR4x. Anandtech said "i7-10510Y is Comet Lake-Y". intel product web site said it is "Amber Lake-Y"
Come on man, you have done far better than this in the past. Instead of using marketing slides, say what you have to say efficiently. This particular Intel announcement does not merit a wall of text. I'm not giving you hard time to roll. (I genuinely enjoy your u-arch deep dive articles.)
Have to agree with Ryan, there is nothing wrong with these marketing slides and some add info. Actually i use it a lot to gather data and information in prof life. Actual testing will be available afterwards.
Ryan is correct. By publishing those slides, AnandTech is providing all currently available information. Doing anything else would be to provide less information, and most people come to AnandTech to get as much information as possible.
They should publish those, if only for posterity and history. The article doesn’t read like an ad. It takes what Intel claims and responds in kind. Until they have a sample to review, it’s all that can be done. These soft launches tell us all we need to know!
This is like so confusing , regular people like me just goign to see GHz / cores , so they would see comet lake-u as the better option but then they see Ice Lake with the what looks on paper , waaaaaaay better GPU and memory speed.
So if you buy a laptop without any dedicated graphics , you would never go for Comet Lake?
If all you do is light office work, browse the web and watch YouTube, the GPU really doesn't matter, either.
Everyone that buys a laptop should evaluate the whole system performance (through reviews of individual laptops), and not get hung up on individual components like you would on a desktop, because more often then not the entire package has a far bigger influence, be it thermal limitations, or the overall system configuration/tuning done by the manufacturer.
GPU should matter, e.g. FF now uses it to accelerate web rendering. If both offer the same performance, then you would want the smaller node as less power draw/longer battery life, but manufacturer should usually make a claim to those ends.
I agree. Also, one aspect I have been disappointed by with Ice Lake on Intel's side is that the significant compute power of the more capable iGPU is not put to use for GPU-assisted computing. Ok, it's not like having two Titans in a desktop workstation, but couldn't those 64 EU be put to use for occasional number crunching?
This is so false that "you don't need good good for regular web browsing". Having half decent integrated GFX has a huge impact on overall usability. Especially on higher res screens. Intel's current integrated graphics can't even draw windows 10 UI at a consistent 60fps
Say what? My work laptop is an I7-6600U, that's a 520 iGPU. here is no discrete GPU in this machine. It runs dual monitor Windows 10 just fine. It runs my EDA software just fine, both schematic drawing and PCB layout and rendering. It is more than perfectly usable. My newest machine at home has an I3-6100, stepping me up to 530 graphics - now we're cooking. Runs Windows 10, same EDA, and other stuff in a perfectly usable fashion at 2560x1440. No, you do NOT need more just to get acceptable usability, especially for an everyday use machine where the hardest thing it will ever do is play Candy Crush on Facebook. John Q Public who is not a gamer, and just emails the grandkids and looks at their Facebook walls does not need some ultra performacne computer. COmputers have been fast enough for that for probably 10 years now (my main desktop at home - that's got an E3-1230 V2, it's 7 years old, though it DOES have a GTX970 from 4 years ago - but it plays everything I actually play, which isn't much, WoT and Kerbal mostly - and it certainly handles all other tasks with ease. Granted, dual monitor 1080, but at on the same size screen I can't even see higher resolutions without using scaling, so I see no point. The I3 machine has a bigger display so the 2560 is fine. ANd video playback on all 3 machines is perfectly fine - pretty much every TV show or movie I watch is watched on my computer, and I watch plenty of YouTube content at max resolution - even picking 2160p on a 1080 display improves the image quality.
Anyone who doesn't need to dig into the details of processor specs would be best served by whatever is least expensive at Best Buy or Amazon. Either one of these will offer blazing fast web browsing and movie watching experiences.
Is segmentation really the reason for the lack of 6 core ICL parts? Considering the 10nm+ node is supposed to be more power efficient, I was expecting to see a higher core count, despite the better GPU. Is intel still struggling with yield and/or power consumption issues even with the reworked 10nm+?
Also, has intel canceled ICL-H and S parts? I haven't seen any mention of them, or perhaps I haven't looked hard enough?
lol, funny that I end up "defending" Intel ... but should not we all give it a break, it made one mistake with the Si process in 40 years (and then only because they over pushed it, went for >>2X area scaling), they will fix it, their 10nm+ is already better by all characteristics than TSMC 7nm ... just a hiccup, Intel is a GOAT of Si design ... few games does not make a damn difference ...
For laptops, I am expecting 7nm APUs in November-February(November would get AMD more competitive in laptops sooner, rather than waiting for a CES release event). It doesn't even need to be EUV, just the current 7nm node.
Thanks Ryan! Two thoughts: 1. I have the strong suspicion that the key reason for keeping Comet Lake-U's memory speed limited to 2933 Mhz on LPDDR4 is to avoid the otherwise likely situation of the cheaper, older node CPU kicking Ice Lake's butt. Ice Lake-U's 4C/8T vs. the cheaper Comet Lake-U's 6C/12T might otherwise just make the old 14 nm+++(+++++) Comet Lake come out on top. So, better slow it down by limiting memory speed. 2. The Comet Lake-Y situation shows what happens when one company has an entire market (low and very low power CPUs) for itself - incremental progress at best, all at high prices. With AMD essentially out of this market for the foreseeable future, I find myself rooting for Qualcomm's 8cx chip, if only to light a fire under Intel's rear end.
You ever thought that Skylake's memory controller can not handle 2933 and only IceLake's can handle it U.
Also I never really care for the Windows for Qualcomm computer with dog slow emulation. Also it looks like from Ice Lake Y specs the only difference is lower power and frequence to keep the power down. Cores and architexture look the same.
Comet Lake is an updated design, not just a renamed SkyLake. Thus, it was Intel's decision what memory controller to implement. From their perspective, it makes sense to not give the cheaper chips the same fast memory as Ice Lake. Just sucks for those of us who get stuck with the lower-priced Comet Lake machines. I agree that emulated Windows on Qualcomm sucks/sucked. However, the 8cx notebooks will run a native version of Windows courtesy of MS, as well as several productivity programs also adapted to run native on 8cx and likely other A76 SoCs, so that should be interesting.
Emulators don't need to be slow. Remember when VirtualPC7 with G5 was the fastest windows computer. MSFT solved it fast by buying VirtuaPC and killing G5 support for a couple of years. Also, remember that Universal apps for Windows are natively compiled for ARM. And here is the real reason why non X86 processors are slower today in windows: Intel has a habit of adding more and more instructions like AVX256/AVX512. Killing non X86 on Windows is one reason for it. If you compare real CPU vs CPU strength RISC have always been faster than X86. Adding more instructions is a way for Intel to try to keep its monopoly including stopping AMD. Remember that a 10 core Intel with AVX512 is faster in media tasks than AMDs 16 core with 128x2 AVX. Intel could off course compete with AMD if they wanted. Taking of the iGPU that takes 50-80% of the die and add more cores. Removing AVX512 that is at least 5% of die area. Add eDRAM cache that would boost performance 10-20%. Intel don't care. Its more fun selling Xeons for 10K/each. Laptops is over 50% of PC market, and all high end is Intel. PCs over 1K dollars is 90% intel (because 90% of computers sold over 1K is Apple). Just a reality check. The best stuff never wins. X86 will never be real 64bit, something that RISC was in 1990. Same with Windows. They could make a working OS using a Unix/Linux core, but they need to ship crap for people to upgrade. The famous case of Novel 3.12. It worked = they did never upgrade = death of a company.
innovating is hard, you can blame Intel for many things ... but it is hard to blame them for furthering the tech ... and I hope you are not referring to ARM as "RISC" :) ... they are the king of CISC, have you seen Neon? ... and then it all does not matter as all are "RISC" inside ... New instructions do help performance/efficiency (yes AMD would rather you not use them as they optimize of legacy, non-AVX, non-FMA is where they kill Intel now), ARM has been changing the whole ISA (current one is ARM v8 mind you) and is adding new instructions too it still, they learned the lesson
Yes this is what we're all wanting to know. I'm assuming since they're both 10th gen, they'd be performing on par. The only reason they're doing Comet Lake is because they simply can't produce enough 10nm CPUs to meet demand.
and the frequency - these 10nm+ Ice Lakes still have a problem going much above >4GHz - Intel's so much refined their 14nm++++++++ it's amazing, it beats TSMC's 7nm on frequency (5GHz 9900K, and overclocks like mad, 3x00 Ryzen is stuck at ~4.5-ish and not a half a bin to overclock above the label) ... my bet is that next 14++++++++ desktop, Rocket Lake, is using Sunny Cove, but then may Intel fixes should-have-been-called-7nm++ by then ... they got hurt bad ... there will be a come back from Intel, it will be a treat to see ...
Which socket or motherboard connection does CometLake-Y utilize?
Is it the BGA layout the same as KabyLake-Y which would allow vendors to re-use existing board designs with existing BGA or is this a new BGA layout? Did Intel provide any information on backwards compatibility between designs?
In other words, perhaps an older board design could be upgraded to newer CometLake-Y if BGA pinout was very, very close.
Assuming Ark is correct here (and that's a big if at the moment), it looks like CML-Y is using a new layout. ABL-Y used FCBGA1515, which is a 20mm X 16.5mm package. However CML-Y, while not listing a specific BGA interface name, is listed as being a 26.5mm x 18.5mm package; significantly larger than the old one.
I have to assume that, owing to the larger CPU die, Intel needed to move to a larger package (they were already really packing it in there with ABL-Y). Coincidentally, this is the same package dimensions for Ice Lake-Y.
It would also be better to have a generational comparison chart, to show what has changed (along with any additional security vulnerabilities that have yet to be fixed — without resorting to big performance hits via software/BIOS).
This would be a whole lot less confusing if Intel just said: "We can't produce enough 10nm Ice Lake CPUs for everyone, so Comet Lake (14nm+++) is our way of mitigating that." Simple as that.
They both perform roughly the same hence why they're both 10th generation. The average consumer won't even care, techies are the only ones confused.
"Even then, it's the 4 core i7-10510U that gets the highest turbo clock 4.9GHz for a single core, which is incredibly high for a 15W part and only 100MHz behind what Intel's top desktop parts turbo to." The table has that processor at 4.5GHz.
Impressive for 12MB cache and 15w, but I'm willing to bet now once the reviews are in, the 4C i7 will actually be faster in most mobile tasks (productivity) due to its much higher clocks for base, turbo and all-core turbo.
That said, I think the Y series is the real star of the show here. At 7W, the i7-10510Y could be a monster at those clock speeds.
I like 4C too, damn AMD with their cores war pushing Intel in a wrong direction and using Cinebench-marketing (which for CPUs testing is like testing Buicks in Formula 1 run - it's not for real life) - I want 1.5X bigger core, and same 4-cores, but please boost my ST by 25%+ ... people lost their mind ... who again runs ray tracing on 15W laptops??
Is anyone else sick of these soldered, low power, low performance, heavily throttled parts? I miss fully socketed, user upgradeable 45w CPUs like the i7-3740QM. We have the 8750H which now surpasses it in performance but is soldered and frequently subject to subpart cooling which cripples its performance.
In 2027 will release the brand new SkyLake+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ on a most advanced 14++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ process!
Just like the article mentioned, it will have a weaker CPU performance watt for watt (and *much* weaker if we include the 6-core part in the comparison) but stronger iGPU performance. Ice Lake also has a better video encoding engine, faster and more efficient wireless (which you actually require multiple devices to take advantage of, i.e. have a WLAN with 3+ routers all supporting Wi-Fi 6) and better AI inference performance.
Comet Lake is just a rebadged Whiskey Lake with slightly faster clock speeds. They made the highest model a 6 core part, but that's it. THERE IS NOTHING NEW HERE. There are already 2 sites calling Intel out on this, why isn't Anandtech? When Nvidia and AMD rebadge their GPUs, we call them out. Why not Intel? They're just not innovating anymore, that's fine but now they're trying to deceive consumers by saying last year's Whiskey Lake CPU is suddenly 10th Gen because of a slight speed bump. It's unethical!
forget the "base" freqs - Intel CPUs almost never runs at those (unless you run matrix muls for hours) - it's 3.9 All Core Turbo, which is what you will actually see most of the time, and it's quite impressive
To avail subsidy under Pradhan Mantri Awas Yojana CLSS scheme, the eligibility criteria are defined for Lower Income Group/Economically Weaker Section ... https://pradhanmantriyojana.net
I am tired of all the Intel "lakes" and I do not keep track of them. It appears that Intel has finally managed to squeeze somewhat decent yields out if its 10nm process at least for the U-series and this is good news. The frequencies of these U series chips appear to be already comparable to the 14nm U series. So given the higher IPC these should hopefully be faster too.
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.
130 Comments
Back to Article
drexnx - Wednesday, August 21, 2019 - link
if Ice Lake basically traded blows with Amber Lake (or was it Whiskey Lake? - seriously Intel, use a new second noun in your codenames, we're tired of lakes!), how is it going to compete with Comet Lake?Santoval - Wednesday, August 21, 2019 - link
Just like the article mentioned, it will have a weaker CPU performance watt for watt (and *much* weaker if we include the 6-core part in the comparison) but stronger iGPU performance. Ice Lake also has a better video encoding engine, faster and more efficient wireless (which you actually require multiple devices to take advantage of, i.e. have a WLAN with 3+ routers all supporting Wi-Fi 6) and better AI inference performance.In other words if you need both the best 6-core 15W CPU with the best iGPU you need to wait a few months for the upcoming Zen 2 based APUs by AMD, which will also have faster Navi iGPUs. Let's just hope AMD upgrades their memory controller to LPDDR4x-3733 as well. While it has not yet been officially confirmed AMD will almost certainly move to 6-core APUs as well as the max option. TSMC's 7nm node allows them to do that with a smaller die than Intel's 6-core Comet Lake-U.
movax2 - Wednesday, August 21, 2019 - link
Was there any information about Zen 2 APU with more than 4 cores?Fulljack - Wednesday, August 21, 2019 - link
currently no. it's all speculation for now.yankeeDDL - Thursday, August 22, 2019 - link
I'd say quite realistic though. Ryzen 4*** mobile should use Zen2 architecture and 7nm. 4C/8T should be the entry level, although for ultra-low power they may go to 2C/4T, but I doubt it.I hope to see an 8C/16T for 15W: seems doable, seeing that intel gets to 6C/12T with a 14nm and a design not optimized for high core count.
0ldman79 - Thursday, August 22, 2019 - link
AMD is bumping their core voltage to get those yields though.Not sure 7nm is going to perform as well in mobile platforms as it does in desktop.
More than a little concerned about that actually. Maybe a later batch/revision of the CPU will have better yields and lower core voltages.
Fulljack - Friday, August 23, 2019 - link
those voltage are safe. because in desktop, those voltage aren't sustained, but for pbo which usually just a few seconds.it's safe. it's uncommon to see voltage as high as 1.5v today, but it won't fry your cpu or motherboard. don't worry.
peevee - Thursday, August 29, 2019 - link
7nm is going to operate in mobile much better because you don't need frequencies that high. It is going to be easy to compete with Intel's "6 cores" operating @ 1.1GHz each. 7nm is going to 3GHz in freaking PHONES.Byte - Thursday, August 22, 2019 - link
Current Ryzen Mobile CPUs have trash battery life. Hopefully they can fix that with Zen2 mobile. I'm eyeing a Dell 13 XPS aio with the Icelake for the insane battery life.0ldman79 - Friday, August 23, 2019 - link
Rumor has it the 3000 series has fixed most of that, but I have yet to see it.Even my i5 6300HQ has better battery life than the Zen+ AMD laptops.
Alexvrb - Friday, August 23, 2019 - link
Most of those issues are on the OEMs. The processors themselves have reasonable draw. Intel is far better at working with OEMs to optimize the whole system for battery life.mazicato - Friday, August 23, 2019 - link
I am also looking at a dell 13 xps for my son post your reply if you get one sorry can you be my guinney pig LOL0ldman79 - Thursday, August 22, 2019 - link
Last I heard the AMD APU will still be using a Polaris GPU.Navi would be preferred, obviously, but they may not have the smaller Navi chip ready by then.
Alexvrb - Friday, August 23, 2019 - link
Put down the crack pipe, old man. *Existing* APUs are 5th gen, Polaris is 4th gen. So at worst they would CONTINUE to use 5th gen "Vega" graphics, albeit with a new process and likely faster memory.wow&wow - Thursday, August 22, 2019 - link
"In other words if you need both the best 6-core 15W CPU with the best iGPU you need to wait a few months for the upcoming Zen 2 based APUs by AMD, which will also have faster Navi iGPUs."Well said, 7nm Zen2 APU likely in Q1 2020 (demo at CES 2020), a simple common sense from what happened in CES 2019.
peevee - Thursday, August 29, 2019 - link
"(and *much* weaker if we include the 6-core part in the comparison)"If by "6-core part" you mean i7-10710U, it is a complete sham. It is only 1.1GHz when all 6 are operating.
qlockandaipeesee - Monday, September 2, 2019 - link
It's 3.9 All Core Turbo, that's the frequency you will most of the time, and is in fact quite impressive. Most OEMs will have decent cooling and it can stay at those frequencies for long time.The base frequency is what heavy LINPACK will run at if you run it for hours and restrict power consumption at no more than 15W, but non-LINPACK will run at higher freqs even for long term. So, no sham.
danjw - Thursday, August 22, 2019 - link
That would take them moving to a whole new architecture. "Lake" indicated these are all based on the Skylake architecture. Which is getting old.wut - Friday, August 23, 2019 - link
Ice Lake is using new Sunny Cove microarchitectureartifex - Monday, August 26, 2019 - link
He buys a Whiskey Lake, he buys a Vodka LakeHe buys a Lager Lake, he buys a Cider Lake
cpugod - Tuesday, August 27, 2019 - link
I get clocked down, but I get up again...Intel never gonna keep me down
cpugod - Tuesday, August 27, 2019 - link
Intel can go Jump-in-a Lake... they're obfuscating to make up for their F-ups and to ensure no one has any idea what processor is what and just think they're getting the latest and greatestyankeeDDL - Wednesday, August 21, 2019 - link
1.1GHz base clock ... The 10nm process seems a bit weak at this stage.yankeeDDL - Wednesday, August 21, 2019 - link
Forget that: it's still 14nm. Can it get any more confusing?The Hardcard - Wednesday, August 21, 2019 - link
How can you forget it, though. For a lot of people who need a laptop this year, Comet will be the best option.If you can wait, hopefully AMD made some big bets about their next APUs. Currently, Ryzen Mobile is more competitive, but it not the aggression that they have on desktop and in server.
just4U - Wednesday, August 21, 2019 - link
I've been looking at some of the laptops being offered with AMD processors.. It's nice to see Asus bringing out some interesting models with 120HZ panels and all the little goodies one would expect in a higher performing laptop. I've been thinking of buying one since the prices are pretty good overall.yankeeDDL - Thursday, August 22, 2019 - link
I wholeheartedly disagree. Being still on 14nm it means that they are only anecdotally better that the previous generation (at least in terms of performance per watt). The Ryzen 3*** mobile are one generation behind the desktop, still on the 12nm (which I think we can safely compare to Intel's 14nm+++).The complexity of Intel's offer is mind boggling, and completely useless.
I can get a 4C/8T Ryzen 3700U (15W), with Vega 10 (https://www.anandtech.com/show/14784/asus-launches... The CPU should be on-par, or close to Intel's top of the line (except for the 6C/12T, which is interesting) and the GPU will run circles around the 620. The Ryzen 4*** mobile should be based on Zen2 and 7nm, and it should trash any 14nm offering from Intel in terms of price/performance.
0ldman79 - Thursday, August 22, 2019 - link
Theoretically, yes, but so far Intel has been ahead of the game in the mobile platform.They're getting excellent speeds and power usage out of 14nm and on that front I'm not sure the 12nm process can compete. On desktops, absolutely. Super low power usage, not sure.
There is a glaring lack of HQ processors from Intel and the highest I've seen from AMD was 35W.
I've still got my 45W Skylake i5, which to be fair almost never goes above 30W, so it could just be they dialed back the "standard" or whatever you want to call it.
yankeeDDL - Friday, August 23, 2019 - link
I agree with you: Intel's offering today is better than AMD's on 12nm, for sure.That said, I have a Core i7 5500U (2C/4T) and I cannot wait to move to - at least - a 4C/8T. I am holding off to see the Ryzen 4***, because the feeling is that it will wipe out anything 14nm.
peevee - Thursday, August 29, 2019 - link
"(except for the 6C/12T, which is interesting)"It is not. Pay attention to the speed the 6cores are operating on. 1.1GHz.
qlockandaipeesee - Monday, September 2, 2019 - link
pay attention to 3.9 All Core Turbo - which is in fact what you get most of the timeSantoval - Wednesday, August 21, 2019 - link
Intel will keep on outputting 14nm based parts for the entire 2020, at least, and they'll probably keep on doing it for the first half of 2021 as well. Ice Lake and Tiger Lake are both expected to be dual release nodes (10nm + 14nm). After Tiger Lake (in 2H2021) Intel *might* have fixed their mess. Or they might have not and 10nm and its variants will be dual release nodes until their end of life.qlockandaipeesee - Monday, September 2, 2019 - link
forget 1.1 ... man, Intel's marketing is asleep! ... people are looking at "1.1" and think it's relevant ... it's 3.9 GHz All Core Turbo - "Turbo" can run for minutes on end on a laptop OEM did not screw up with ... and Intel makes sure, for the most part, they don't ...Vitor - Wednesday, August 21, 2019 - link
Quite incredible how Intel lost so much of its lust. And that imagem of the "hand covered in space Internet" looks like a tarantula.evilspoons - Wednesday, August 21, 2019 - link
Wouldn't change the processors (or the confusing way everything is laid out in the product lineups...) even if the picture was an *actual* tarantula.JJ Wu - Wednesday, August 21, 2019 - link
Comet Lake series processor does not support LPDDR4x. it should be LPDDR3.Refer to https://ark.intel.com/content/www/tw/zh/ark/produc...
Ryan Smith - Wednesday, August 21, 2019 - link
Ark is wrong in this case (it happens). LPDDR4X is one of the marquee features of Comet Lake-U.neuen - Wednesday, August 21, 2019 - link
As per ark, neither i7-10710U (6-core) nor i7-10510Y support Thermal Velocity Boost while i7-10510U (4-core) does. Could you comment whether ark is wrong on that one? This is likely some more hundreds MHz of difference.Ryan Smith - Wednesday, August 21, 2019 - link
I don't have any detailed information on TVB. Sorry.close - Wednesday, August 21, 2019 - link
Ryan, based on your experience (and any other AT editor that might care to weigh in) is there any number claimed on those slides that users should take with a grain of salt, or require some clarification? The question was triggered by the +41% in Office 365 scenarios which lacks any kind of disclaimer (from Intel and AT) and suggests it will be generally the case for *any* O365 workloads.Ryan Smith - Wednesday, August 21, 2019 - link
All of the Intel performance numbers should be taken with a disclaimer - or at the very least, you'll want to read the footnotes. Intel is doing 25W vs. 15W here, so it's not an apples-for-apples comparison in terms of power limits.As for the O365 scenario in particular, it's measuring "the time it takes Microsoft Office to perform 3 tasks in a multitasking scenario (export PowerPoint file to video, export word doc to pdf and excel calculation).
fasterquieter - Wednesday, August 21, 2019 - link
Shame Apple killed that teeny-tiny MacBook. Looks like it could finally have been fast enough.dudedud - Wednesday, August 21, 2019 - link
The air also uses Y-series but with a fan, so at least those are gonna have a huge performance uplift.JJ Wu - Wednesday, August 21, 2019 - link
Does intel make mistel on the document they released? On their product web site, Comet Lake-U does not support LPDDR4x.Anandtech said "i7-10510Y is Comet Lake-Y". intel product web site said it is "Amber Lake-Y"
JJ Wu - Wednesday, August 21, 2019 - link
Anandtech said "i7-10510Y is Comet Lake-Y". intel product web site said it is "Amber Lake-Y"https://ark.intel.com/content/www/tw/zh/ark/produc...
YB1064 - Wednesday, August 21, 2019 - link
Please don't include worthless marketing slides. This article reads like a giant Intel ad.Ryan Smith - Wednesday, August 21, 2019 - link
It's quite literally the only Comet Lake material we have right now. So it's that or a wall of text.YB1064 - Wednesday, August 21, 2019 - link
Come on man, you have done far better than this in the past. Instead of using marketing slides, say what you have to say efficiently. This particular Intel announcement does not merit a wall of text. I'm not giving you hard time to roll. (I genuinely enjoy your u-arch deep dive articles.)YB1064 - Wednesday, August 21, 2019 - link
*trollduploxxx - Wednesday, August 21, 2019 - link
Have to agree with Ryan, there is nothing wrong with these marketing slides and some add info.Actually i use it a lot to gather data and information in prof life. Actual testing will be available afterwards.
coder543 - Wednesday, August 21, 2019 - link
Ryan is correct. By publishing those slides, AnandTech is providing all currently available information. Doing anything else would be to provide less information, and most people come to AnandTech to get as much information as possible.Orange_Swan - Wednesday, August 21, 2019 - link
exactly, if I want a quick and dirty overview, I'll use Hexus or a similar site, if I want ALL the information available I turn to AnandTechshabby - Wednesday, August 21, 2019 - link
How else will you know it's 1-2% faster?MonkeyPaw - Wednesday, August 21, 2019 - link
They should publish those, if only for posterity and history. The article doesn’t read like an ad. It takes what Intel claims and responds in kind. Until they have a sample to review, it’s all that can be done. These soft launches tell us all we need to know!plopke - Wednesday, August 21, 2019 - link
This is like so confusing , regular people like me just goign to see GHz / cores , so they would see comet lake-u as the better option but then they see Ice Lake with the what looks on paper , waaaaaaay better GPU and memory speed.So if you buy a laptop without any dedicated graphics , you would never go for Comet Lake?
I serisouly would not know what to buy.
nevcairiel - Wednesday, August 21, 2019 - link
If all you do is light office work, browse the web and watch YouTube, the GPU really doesn't matter, either.Everyone that buys a laptop should evaluate the whole system performance (through reviews of individual laptops), and not get hung up on individual components like you would on a desktop, because more often then not the entire package has a far bigger influence, be it thermal limitations, or the overall system configuration/tuning done by the manufacturer.
RSAUser - Wednesday, August 21, 2019 - link
GPU should matter, e.g. FF now uses it to accelerate web rendering.If both offer the same performance, then you would want the smaller node as less power draw/longer battery life, but manufacturer should usually make a claim to those ends.
eastcoast_pete - Wednesday, August 21, 2019 - link
I agree. Also, one aspect I have been disappointed by with Ice Lake on Intel's side is that the significant compute power of the more capable iGPU is not put to use for GPU-assisted computing. Ok, it's not like having two Titans in a desktop workstation, but couldn't those 64 EU be put to use for occasional number crunching?isthisavailable - Thursday, August 22, 2019 - link
This is so false that "you don't need good good for regular web browsing". Having half decent integrated GFX has a huge impact on overall usability. Especially on higher res screens. Intel's current integrated graphics can't even draw windows 10 UI at a consistent 60fpsisthisavailable - Thursday, August 22, 2019 - link
*good GPUrrinker - Monday, August 26, 2019 - link
Say what? My work laptop is an I7-6600U, that's a 520 iGPU. here is no discrete GPU in this machine. It runs dual monitor Windows 10 just fine. It runs my EDA software just fine, both schematic drawing and PCB layout and rendering. It is more than perfectly usable. My newest machine at home has an I3-6100, stepping me up to 530 graphics - now we're cooking. Runs Windows 10, same EDA, and other stuff in a perfectly usable fashion at 2560x1440. No, you do NOT need more just to get acceptable usability, especially for an everyday use machine where the hardest thing it will ever do is play Candy Crush on Facebook.John Q Public who is not a gamer, and just emails the grandkids and looks at their Facebook walls does not need some ultra performacne computer. COmputers have been fast enough for that for probably 10 years now (my main desktop at home - that's got an E3-1230 V2, it's 7 years old, though it DOES have a GTX970 from 4 years ago - but it plays everything I actually play, which isn't much, WoT and Kerbal mostly - and it certainly handles all other tasks with ease. Granted, dual monitor 1080, but at on the same size screen I can't even see higher resolutions without using scaling, so I see no point. The I3 machine has a bigger display so the 2560 is fine.
ANd video playback on all 3 machines is perfectly fine - pretty much every TV show or movie I watch is watched on my computer, and I watch plenty of YouTube content at max resolution - even picking 2160p on a 1080 display improves the image quality.
sorten - Wednesday, August 21, 2019 - link
Anyone who doesn't need to dig into the details of processor specs would be best served by whatever is least expensive at Best Buy or Amazon. Either one of these will offer blazing fast web browsing and movie watching experiences.regsEx - Wednesday, August 21, 2019 - link
Any news on CML-S? I wanted to upgrade to 3700X, but motherboards are way too expensive. And not a single midrange X470 with BIOS Flashback in retail.Ryan Smith - Wednesday, August 21, 2019 - link
"Any news on CML-S?"Nope. Sorry. Today's launch is solely about low-power mobile processors.
yeeeeman - Thursday, August 22, 2019 - link
For what do you need a x70 class motherboard?shtldr - Thursday, August 22, 2019 - link
3700X runs perfectly fine on e.g. MSI B450M Mortar.ballsystemlord - Thursday, August 22, 2019 - link
What is CML-S? Online searches turn up nothing of use.katsetus - Friday, August 23, 2019 - link
Desktop Comet Lake. You find nothing, because it doesn't exist.qlockandaipeesee - Monday, September 2, 2019 - link
of course it does ... google 10-core Comet Lake ...Targon - Friday, August 23, 2019 - link
Asus ROG Crosshair VII Hero for $250, but that may be too expensive for you. It's worth the money, but it isn't a cheap motherboard by any means.jcc5169 - Wednesday, August 21, 2019 - link
Intel appears to be giving up on servers and workstations for mobile devices.PeachNCream - Wednesday, August 21, 2019 - link
You're watching the weather (a single product announcement) and assuming it's an indicator of climate (all of Intel's product development).dudedud - Wednesday, August 21, 2019 - link
I dont get it, the 9th gen is better than 10th? Seriously?dullard - Wednesday, August 21, 2019 - link
Which 9th generation chip and which 10th generation chip are you discussing here?shabby - Wednesday, August 21, 2019 - link
Loleddman - Wednesday, August 21, 2019 - link
Is segmentation really the reason for the lack of 6 core ICL parts? Considering the 10nm+ node is supposed to be more power efficient, I was expecting to see a higher core count, despite the better GPU. Is intel still struggling with yield and/or power consumption issues even with the reworked 10nm+?Also, has intel canceled ICL-H and S parts? I haven't seen any mention of them, or perhaps I haven't looked hard enough?
shabby - Wednesday, August 21, 2019 - link
Segmentation? Lol no, 10nm just isn't upto snuff so 2/4 core is all we get.qlockandaipeesee - Monday, September 2, 2019 - link
lol, funny that I end up "defending" Intel ... but should not we all give it a break, it made one mistake with the Si process in 40 years (and then only because they over pushed it, went for >>2X area scaling), they will fix it, their 10nm+ is already better by all characteristics than TSMC 7nm ... just a hiccup, Intel is a GOAT of Si design ... few games does not make a damn difference ...NICOXIS - Wednesday, August 21, 2019 - link
If AMD releases a 7nm EUV Zen 2 based APU by next year, Intel is in real trouble.eastcoast_pete - Wednesday, August 21, 2019 - link
Wish they would. Right now, that's a big "If". Has AMD announced anything at all along those lines?Rudde - Thursday, August 22, 2019 - link
"Renoir (Zen2) APUs will likely launch in the first few months of 2020, following the regular cadence the company has stuck to since 2017."Targon - Friday, August 23, 2019 - link
For laptops, I am expecting 7nm APUs in November-February(November would get AMD more competitive in laptops sooner, rather than waiting for a CES release event). It doesn't even need to be EUV, just the current 7nm node.Trikkiedikkie - Wednesday, August 21, 2019 - link
MArketing has had a lot of weight in this blah blah. A comparison between 2 processors but with different settings is BS.Intel is getting very sad. Avoid
qlockandaipeesee - Monday, September 2, 2019 - link
lol, right, go get AMD laptop ... try it ... seriously :)eastcoast_pete - Wednesday, August 21, 2019 - link
Thanks Ryan! Two thoughts:1. I have the strong suspicion that the key reason for keeping Comet Lake-U's memory speed limited to 2933 Mhz on LPDDR4 is to avoid the otherwise likely situation of the cheaper, older node CPU kicking Ice Lake's butt. Ice Lake-U's 4C/8T vs. the cheaper Comet Lake-U's 6C/12T might otherwise just make the old 14 nm+++(+++++) Comet Lake come out on top. So, better slow it down by limiting memory speed.
2. The Comet Lake-Y situation shows what happens when one company has an entire market (low and very low power CPUs) for itself - incremental progress at best, all at high prices. With AMD essentially out of this market for the foreseeable future, I find myself rooting for Qualcomm's 8cx chip, if only to light a fire under Intel's rear end.
HStewart - Wednesday, August 21, 2019 - link
You ever thought that Skylake's memory controller can not handle 2933 and only IceLake's can handle it U.Also I never really care for the Windows for Qualcomm computer with dog slow emulation. Also it looks like from Ice Lake Y specs the only difference is lower power and frequence to keep the power down. Cores and architexture look the same.
eastcoast_pete - Wednesday, August 21, 2019 - link
Comet Lake is an updated design, not just a renamed SkyLake. Thus, it was Intel's decision what memory controller to implement. From their perspective, it makes sense to not give the cheaper chips the same fast memory as Ice Lake. Just sucks for those of us who get stuck with the lower-priced Comet Lake machines. I agree that emulated Windows on Qualcomm sucks/sucked. However, the 8cx notebooks will run a native version of Windows courtesy of MS, as well as several productivity programs also adapted to run native on 8cx and likely other A76 SoCs, so that should be interesting.shompa - Sunday, August 25, 2019 - link
Emulators don't need to be slow. Remember when VirtualPC7 with G5 was the fastest windows computer. MSFT solved it fast by buying VirtuaPC and killing G5 support for a couple of years. Also, remember that Universal apps for Windows are natively compiled for ARM. And here is the real reason why non X86 processors are slower today in windows: Intel has a habit of adding more and more instructions like AVX256/AVX512. Killing non X86 on Windows is one reason for it. If you compare real CPU vs CPU strength RISC have always been faster than X86. Adding more instructions is a way for Intel to try to keep its monopoly including stopping AMD. Remember that a 10 core Intel with AVX512 is faster in media tasks than AMDs 16 core with 128x2 AVX. Intel could off course compete with AMD if they wanted. Taking of the iGPU that takes 50-80% of the die and add more cores. Removing AVX512 that is at least 5% of die area. Add eDRAM cache that would boost performance 10-20%. Intel don't care. Its more fun selling Xeons for 10K/each. Laptops is over 50% of PC market, and all high end is Intel. PCs over 1K dollars is 90% intel (because 90% of computers sold over 1K is Apple). Just a reality check. The best stuff never wins. X86 will never be real 64bit, something that RISC was in 1990. Same with Windows. They could make a working OS using a Unix/Linux core, but they need to ship crap for people to upgrade. The famous case of Novel 3.12. It worked = they did never upgrade = death of a company.qlockandaipeesee - Monday, September 2, 2019 - link
innovating is hard, you can blame Intel for many things ... but it is hard to blame them for furthering the tech ... and I hope you are not referring to ARM as "RISC" :) ... they are the king of CISC, have you seen Neon? ... and then it all does not matter as all are "RISC" inside ... New instructions do help performance/efficiency (yes AMD would rather you not use them as they optimize of legacy, non-AVX, non-FMA is where they kill Intel now), ARM has been changing the whole ISA (current one is ARM v8 mind you) and is adding new instructions too it still, they learned the lessonHStewart - Wednesday, August 21, 2019 - link
What would be interesting is compare ai7-10510U (Comet lake) vs i7-1065G7 (Ice Lake)
Both are quad core different core architexture.
poohbear - Thursday, August 22, 2019 - link
Yes this is what we're all wanting to know. I'm assuming since they're both 10th gen, they'd be performing on par. The only reason they're doing Comet Lake is because they simply can't produce enough 10nm CPUs to meet demand.qlockandaipeesee - Monday, September 2, 2019 - link
and the frequency - these 10nm+ Ice Lakes still have a problem going much above >4GHz - Intel's so much refined their 14nm++++++++ it's amazing, it beats TSMC's 7nm on frequency (5GHz 9900K, and overclocks like mad, 3x00 Ryzen is stuck at ~4.5-ish and not a half a bin to overclock above the label) ... my bet is that next 14++++++++ desktop, Rocket Lake, is using Sunny Cove, but then may Intel fixes should-have-been-called-7nm++ by then ... they got hurt bad ... there will be a come back from Intel, it will be a treat to see ...sdnic - Wednesday, August 21, 2019 - link
Which socket or motherboard connection does CometLake-Y utilize?Is it the BGA layout the same as KabyLake-Y which would allow vendors to re-use existing board designs with existing BGA or is this a new BGA layout? Did Intel provide any information on backwards compatibility between designs?
In other words, perhaps an older board design could be upgraded to newer CometLake-Y if BGA pinout was very, very close.
Ryan Smith - Wednesday, August 21, 2019 - link
Assuming Ark is correct here (and that's a big if at the moment), it looks like CML-Y is using a new layout. ABL-Y used FCBGA1515, which is a 20mm X 16.5mm package. However CML-Y, while not listing a specific BGA interface name, is listed as being a 26.5mm x 18.5mm package; significantly larger than the old one.I have to assume that, owing to the larger CPU die, Intel needed to move to a larger package (they were already really packing it in there with ABL-Y). Coincidentally, this is the same package dimensions for Ice Lake-Y.
https://images.anandtech.com/doci/14664/ICLU-wm.jp...
chucky2 - Wednesday, August 21, 2019 - link
Would be nice to know when AV-1 codec is going to be supported.Oxford Guy - Thursday, August 22, 2019 - link
So, are all of the security risks mitigated or are there some that don't appear in your chart but which are already known?Oxford Guy - Thursday, August 22, 2019 - link
It would also be better to have a generational comparison chart, to show what has changed (along with any additional security vulnerabilities that have yet to be fixed — without resorting to big performance hits via software/BIOS).ksec - Thursday, August 22, 2019 - link
The article could be summarised as :Comet Lake, A 6 Core 14nm+++ Skylake CPU with a few hardware security fix and LPDDR4X Support plus Icelake's PCH.
Haawser - Thursday, August 22, 2019 - link
Think they should stop calling it '4.9 GHz Turbo' and pick a more accurate description, like 'Micro Burst', because that's surely what it will be.Rudde - Thursday, August 22, 2019 - link
Not if OEMs put the power limit at 45W.ET - Thursday, August 22, 2019 - link
How can both Ice Lake and Comet Lake be "10th gen"? They're quite different. That's going to be quite confusing for buying getting "10th gen" laptops.shabby - Thursday, August 22, 2019 - link
Intel is betting on confusing people it seems.LuckyKnight - Thursday, August 22, 2019 - link
Are we likely to see any new NUCs? I was hoping for 10nm, without seperate GPU before upgrading.zodiacfml - Thursday, August 22, 2019 - link
Depressing. Sounds like AMD is going to own mobile too, like how they did in the desktopshtldr - Thursday, August 22, 2019 - link
Depressing only if you are an Intel fanboy.poohbear - Thursday, August 22, 2019 - link
This would be a whole lot less confusing if Intel just said: "We can't produce enough 10nm Ice Lake CPUs for everyone, so Comet Lake (14nm+++) is our way of mitigating that." Simple as that.They both perform roughly the same hence why they're both 10th generation. The average consumer won't even care, techies are the only ones confused.
viksh - Thursday, August 22, 2019 - link
<a herf = '' https://www.vktechnical.com/2019/01/qualcomm-snapd... > FIRST QUALCOMM SNAPDRAGON 855 PROCESSOR MOBILE</a>you are doing well
ballsystemlord - Thursday, August 22, 2019 - link
One technical error:"Even then, it's the 4 core i7-10510U that gets the highest turbo clock 4.9GHz for a single core, which is incredibly high for a 15W part and only 100MHz behind what Intel's top desktop parts turbo to."
The table has that processor at 4.5GHz.
Machinus - Friday, August 23, 2019 - link
There are three plus signs in a row now? It looks like a joke.qlockandaipeesee - Monday, September 2, 2019 - link
yep, I dunno, should have called it 13.9nm instead ...Samus - Friday, August 23, 2019 - link
Impressive for 12MB cache and 15w, but I'm willing to bet now once the reviews are in, the 4C i7 will actually be faster in most mobile tasks (productivity) due to its much higher clocks for base, turbo and all-core turbo.That said, I think the Y series is the real star of the show here. At 7W, the i7-10510Y could be a monster at those clock speeds.
qlockandaipeesee - Monday, September 2, 2019 - link
I like 4C too, damn AMD with their cores war pushing Intel in a wrong direction and using Cinebench-marketing (which for CPUs testing is like testing Buicks in Formula 1 run - it's not for real life) - I want 1.5X bigger core, and same 4-cores, but please boost my ST by 25%+ ... people lost their mind ... who again runs ray tracing on 15W laptops??Pessimism - Friday, August 23, 2019 - link
Is anyone else sick of these soldered, low power, low performance, heavily throttled parts? I miss fully socketed, user upgradeable 45w CPUs like the i7-3740QM. We have the 8750H which now surpasses it in performance but is soldered and frequently subject to subpart cooling which cripples its performance.qlockandaipeesee - Monday, September 2, 2019 - link
well, I'm with you ... or rather would be if I did not like my X1 ... light and thin is cool (if only proverbially speaking :) )Zingam - Saturday, August 24, 2019 - link
In 2027 will release the brand new SkyLake+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ on a most advanced 14++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ process!Suck it, AMD! You dead! AMD is for peasants!
qlockandaipeesee - Monday, September 2, 2019 - link
14nm to infinity and beyond :)Timur Born - Sunday, August 25, 2019 - link
The tables list "1C Turbo", but at least on Windows you never get true 1 core processing, so 2C frequency is more practically relevant.qlockandaipeesee - Monday, September 2, 2019 - link
not true ... to clear suspicion run Cinebench 1T - you will see the highest turbo clocks ...FunnyWifiNames - Sunday, August 25, 2019 - link
Just like the article mentioned, it will have a weaker CPU performance watt for watt (and *much* weaker if we include the 6-core part in the comparison) but stronger iGPU performance. Ice Lake also has a better video encoding engine, faster and more efficient wireless (which you actually require multiple devices to take advantage of, i.e. have a WLAN with 3+ routers all supporting Wi-Fi 6) and better AI inference performance.Maxiking - Monday, August 26, 2019 - link
AMD FRAUD CONFIRMED BOYS.https://www.reddit.com/r/pcgaming/comments/cusn2t/...
poohbear - Tuesday, August 27, 2019 - link
Comet Lake is just a rebadged Whiskey Lake with slightly faster clock speeds. They made the highest model a 6 core part, but that's it. THERE IS NOTHING NEW HERE. There are already 2 sites calling Intel out on this, why isn't Anandtech? When Nvidia and AMD rebadge their GPUs, we call them out. Why not Intel? They're just not innovating anymore, that's fine but now they're trying to deceive consumers by saying last year's Whiskey Lake CPU is suddenly 10th Gen because of a slight speed bump. It's unethical!Gastec - Sunday, September 1, 2019 - link
And what do you want, to quarrel online for the sake of quarreling? Fuggedaboutit! JUST BUY IT! IT JUST WORKS!peevee - Thursday, August 29, 2019 - link
6 cores at 1.1GHz vs 4 cores at 1.8 GHz. 6.6 vs 7.2. What a fake "improvement".qlockandaipeesee - Monday, September 2, 2019 - link
forget the "base" freqs - Intel CPUs almost never runs at those (unless you run matrix muls for hours) - it's 3.9 All Core Turbo, which is what you will actually see most of the time, and it's quite impressiveGastec - Sunday, September 1, 2019 - link
Just smack me over the back of my head with the big-lettered, colored slideshow.Oliver44 - Tuesday, September 10, 2019 - link
To avail subsidy under Pradhan Mantri Awas Yojana CLSS scheme, the eligibility criteria are defined for Lower Income Group/Economically Weaker Section ...https://pradhanmantriyojana.net
Korguz - Wednesday, September 18, 2019 - link
and yet another random site, filled with trojans, spyware and malware ???nikolayo - Tuesday, September 10, 2019 - link
I am tired of all the Intel "lakes" and I do not keep track of them. It appears that Intel has finally managed to squeeze somewhat decent yields out if its 10nm process at least for the U-series and this is good news. The frequencies of these U series chips appear to be already comparable to the 14nm U series. So given the higher IPC these should hopefully be faster too.