Thursday, April 2nd 2020

Intel 10th Gen Core "Comet Lake-H" Mobile Processor Lineup with Desktop-class Performance Announced: 5.3 GHz, 8 Cores, 16 Threads

Intel today announced its 10th generation Core "Comet Lake-H" mobile processor family with a promise of bringing premium desktop-level performance to conventional notebook form-factors, including some bordering the thin-and-light form-factors (under 2 cm thickness). The higher-end of the lineup is geared not just toward serious gaming and creative work, but also toward PC enthusiasts wanting to overclock the processors. For gamers, Intel is guiding its notebook partners to come up with new designs with high refresh-rate and DisplayHDR 1000 displays. The company is also providing OEMs with engineering support to better design their cooling solutions for these processors.

Compared to 9th generation "Coffee Lake-H" processors, the strategy with "Comet Lake-H" appears to be increasing clock-speeds at given price-points, along with introduction of Turbo Boost Max 3.0 technology to this segment, over from the company's HEDT processor lineup. Four out six SKUs hit the magic frequency figure of 5.00 GHz. Two of the SKUs even feature CPU overclocking support. The Core i5 family consists of 4-core/8-thread chips. The Core i7 lineup consist of two 6-core/12-thread parts, and one 8-core/16-thread part. The sole Core i9 SKU is an 8-core/16-thread, with the highest clock speeds and unlocked base-clock multiplier. The "Comet Lake-H" processors are built on 14 nm process, and have identical CPU IPC to "Skylake."
Intel 10th Gen Core-H Comet Lake Lineup Comet Lake-H
The series begins with the Core i5-10300H and i5-10400H are 4-core/8-thread parts with 8 MB shared L3 caches. The i5-10300H is clocked at 2.50 GHz nominal with 4.50 GHz boost; while the i5-10400H ticks 100 MHz faster. The Core i7 lineup consists of three SKUs, the i7-10750H and i7-10850H being 6-core/12-thread parts with 12 MB L3 caches; and the i7-10875H being 8-core/16-thread with 16 MB L3 cache. The i7-10750H is clocked at 2.60 GHz with 5.00 GHz boost; while the i7-10850H runs at 2.70 GHz with 5.10 GHz boost.

The i7-10850H is "partially unlocked," in that you can manually overclock it by up to four "bins" (base clock multiplier values, i.e., by up to 400-ish MHz). The Core i7-10875H doesn't feature an unlocked multiplier, but is clocked at 2.30 GHz, with up to 5.10 GHz boost. Leading the pack is the Core i9-10980HK, clocked at 2.40 GHz nominal, 5.30 GHz boost, and a fully unlocked multiplier. Besides support for Extreme Tuning Utility (XTU), Intel is introducing a new tool that simplifies overclocking, called Speed Optimizer.

It's interesting to note that unlike previous-gen "Coffee Lake-H" series, "Comet Lake-H" chips only support conventional DDR4 SODIMM memory, with native support for DDR4-2933. LPDDR3 support is dropped, and there's no support for LPDDR4 or LPDDR4x replacing it, either. The processors support up to 128 GB of dual-channel memory. Between "Comet Lake-H" and AMD "Renoir," there are quite a few platform tradeoffs. Renoir supports LPDDR4X up to 4266 MHz dual-channel, PCIe gen 4.0, AMD's equivalent of Smart Sound technology (ability to give voice-commands even when the notebook is standing by), and a more advanced power-management solution. Intel's platform-level advantages are solely in the area of PCIe, with 40 total gen 3.0 lanes, which should enable multiple M.2 NVMe slots, Thunderbolt 3 ports, etc.
On the platform side of things, not a lot has changed. The HM470 chipset is the core logic of choice. Combined with the 16 PCIe gen 3.0 lanes from the processor, the chipset puts out 24 gen 3.0 lanes of its own, totaling 40 on the platform. This enables notebooks to have multiple M.2 NVMe slots, more than one Thunderbolt 3 port, and other bandwidth-hungry onboard devices. Intel is standardizing 802.11ax Wi-Fi 6 (gig+) support using the company's new AX201 WLAN card. Some notebooks could even feature 2.5 GbE wired networking.

The complete slide-deck follows.
Add your own comment

44 Comments on Intel 10th Gen Core "Comet Lake-H" Mobile Processor Lineup with Desktop-class Performance Announced: 5.3 GHz, 8 Cores, 16 Threads

#1
The Quim Reaper
I look forward to seeing them hit 5.3Ghz....for about 3 seconds.
Posted on Reply
#3
truehighroller1
The Quim ReaperI look forward to seeing them hit 5.3Ghz....for about 3 seconds.
1 second one core and limited power and 115C temps with in one microsecond with 4 external 360radiators and two strong pumps.. Freaking joke seriously.
Posted on Reply
#4
ShurikN
The last time a chip was so obsolete at launch was probably bulldozer.
And it would be fine if the prices reflected that, but it's Intel so you know it's gonna be overpriced
Posted on Reply
#5
btarunr
Editor & Senior Moderator
VulkanBrosis this 7 nm or 14 nm?
14 nm, same IPC as Skylake.
Posted on Reply
#6
laszlo
The Quim ReaperI look forward to seeing them hit 5.3Ghz....for about 3 seconds.
1 ms maybe just to appear in monitoring soft...
Posted on Reply
#7
R0H1T
Reminds me of :laugh:


Posted on Reply
#8
Vayra86
The Quim ReaperI look forward to seeing them hit 5.3Ghz....for about 3 seconds.
You give them way too much credit. 3 entire seconds?!
Posted on Reply
#9
john_
The most unfortunate thing here is that, we will see more laptops with these processors than Renoir. At higher prices than the Renoir models, offering lower performance and probably battery life. And people will prefer to buy these models than those with AMD inside.
Posted on Reply
#10
R0H1T
Well there's only one way that can happen, no need to spell it out. Renoir is unmatched on most metrics, especially battery life, & shouldn't need huge "marketing" budget to gain traction in the retail space.
Posted on Reply
#11
ppn
Comet lake will boost to at least 3.5 on all core all day. so long you are not slamming prime-avx on it, and why would you, just to say look look it can survive the absolute torture test, so it is perfectly fine. 14nm is also perfectly fine with 43Mtr/mm2 is already 30% better than TSMC 14nm 30 Mtr/mm2, and TSMC 7nm is actually 50Mtr/mm2. 7nm TSMC is not even close to 7nm intel. so that is coming just a matter of time. until then refreshes and gap filler.

You can see it in the size, 2 cores 14nm Intel = 25mm2, 2 cores 7nm zen = 19mm2, really, really not that much big of a deal smaller.

Perhaps the unlocked CPU carries Z490, not H470.
Posted on Reply
#12
R0H1T
And 7nm Intel isn't even Vaporware atm let alone ready for primetime, at least 10nm with CNL was vaporware a couple of years back.
Posted on Reply
#13
amehmet24
I will buy the 100 MHz faster parts and pay whatever they ask for it !!!

Jerks ... they still have the nerve to sell parts that are 100 MHz apart from each other. What cost is involved in producing 100 MHz faster parts ? Nothing. All chips are sorted in the same testing process anyway and etched with the same number of characters anyway and packed into same packaging anyway ???

Jerks... They still don't get this "milking the customer by increasing CPU speed by 100 MHz" put them into this position. 5 years ago, they were claiming to be 18-24 months ahead of everybody in chip manufacturing process; now others , even AMD, are using 7 mm process and will move to 5 mm process soon; while they are still on 14 nm process.

They deserve what they will get.

Based on these specs, 6 and 8 core parts seem both to be produced as 8 core parts - and 6 core parts seem to be disabled cores after sorting process and core speeds are adjusted to use the unused thermal leeway . Did I say "Jerks" before ?
amehmet24I will buy the 100 MHz faster parts and pay whatever they ask for it !!!

Jerks ... they still have the nerve to sell parts that are 100 MHz apart from each other. What cost is involved in producing 100 MHz faster parts ? Nothing. All chips are sorted in the same testing process anyway and etched with the same number of characters anyway and packed into same packaging anyway ???

Jerks... They still don't get this "milking the customer by increasing CPU speed by 100 MHz" put them into this position. 5 years ago, they were claiming to be 18-24 months ahead of everybody in chip manufacturing process; now others , even AMD, are using 7 mm process and will move to 5 mm process soon; while they are still on 14 nm process.

They deserve what they will get.
Posted on Reply
#14
R0H1T
amehmet24Did I say "Jerks" before ?
Nope, must've missed it :slap:
Posted on Reply
#15
ppn
Clearly they are not milking the same custommer with 100Mhz, it is a new born fresh custommer that is just building a system. Almost nobody is replacing a perfectly good cpu just for the 100Mhz bump. Where are those preposterous claims coming from. AMd are not using 7nm. it is 7nm-ish(10-11nm), and we haven't seen 5nm yet, could be in 1 year, could just fail to launch. Given the choice between 5nm-ish, and 7nm Intel that is back on track for 2021, I prefer to wait another year, I upgrade once every decade so it would be worth it.
Posted on Reply
#16
Caring1
ppnClearly they are not milking the same custommer with 100Mhz, it is a new born fresh custommer that is just building a system. Almost nobody is replacing a perfectly good cpu just for the 100Mhz bump. Where are those preposterous claims coming from.
Circa 1985 :eek:
Posted on Reply
#17
Valantar
So IIRC TVB requires CPU temperatures to be below 65C. Anyone care to bet how long a chip like this will stay below 65C while under a heavy single-core load in a laptop? I'm guessing the answer is a very, very short time.

Also congratulations to Intel for now officially surpassing the 2x mark for boost clock vs. base clock. That is certainly impressive, but not in the way Intel wants it to be.

It really is too bad that the engineering effort to fit these into existing laptop designs will be trivial compared to designing new motherboards for Ryzen 4000-series chips, meaning we'll inevitably see far more laptops with these chips out there despite them being fundamentally inferior. Intel's massive marketing support of course doesn't help this either. But the ~2x delta in perf/W isn't going anywhere, meaning anything using these chips will be bigger, hotter, louder and no more performant (likely less) than AMD-equipped alternatives.
amehmet24Based on these specs, 6 and 8 core parts seem both to be produced as 8 core parts - and 6 core parts seem to be disabled cores after sorting process and core speeds are adjusted to use the unused thermal leeway . Did I say "Jerks" before ?
Wait, is there something wrong with binning chips to reduce waste by disabling defective parts now? Please explain. I mean, this series is obviously not a match for Renoir, but binning like that is literally what everyone does. Do you think the <8 core AMD chips have different silicon? There is literally one Renoir die that is used across the entire U, H and HS lineups, regardless of core count, CU count and so on.
Posted on Reply
#18
Vya Domus
ValantarIt really is too bad that the engineering effort to fit these into existing laptop designs will be trivial compared to designing new motherboards for Ryzen 4000-series chips, meaning we'll inevitably see far more laptops with these chips out there despite them being fundamentally inferior.
john_The most unfortunate thing here is that, we will see more laptops with these processors than Renoir. At higher prices than the Renoir models, offering lower performance and probably battery life.
I highly doubt that, even if OEMs will want the Intel platform everywhere it just wont be possible. For one thing there aren't going to be many of these produced by Intel, the clocks are just too high and binning process will be atrocious. Secondly, the cooling will have to be ramped up considerably, don't be fooled by the 45W TDP, hell AMD's 4900HS is more like 55W under load imagine how these things are going to behave.

There just aren't going to be many laptops around featuring these CPUs (except for the i5s but those wont really be that much faster anyway). It's just something meant to steal the show.
Posted on Reply
#19
Valantar
ppnComet lake will boost to at least 3.5 on all core all day. so long you are not slamming prime-avx on it, and why would you, just to say look look it can survive the absolute torture test, so it is perfectly fine.
...that depends on the cooling design of the laptop. Or are you saying that even lower-bin CML will maintain 3.5GHz boost within TDP? In that case, why have a ~30% delta between sustainable boost at TDP and base clock? That doesn't make sense. And what top-bin chips can do, or what they can do in a laptop with overbuilt thermals, really doesn't apply to the whole series.
ppn14nm is also perfectly fine with 43Mtr/mm2 is already 30% better than TSMC 14nm 30 Mtr/mm2, and TSMC 7nm is actually 50Mtr/mm2. 7nm TSMC is not even close to 7nm intel. so that is coming just a matter of time. until then refreshes and gap filler.
Transistor density is quite literally the least interesting metric when comparing how nodes perform. The relationship between density and performance is far too complex to be used as a direct comparison, so please stop. Beyond that, nobody is saying TSMC 7nm is equal to Intel 7nm. The difference is that TSMC 7nm CPUs have been on the market for quite a while now, while Intel 7nm chips are likely 2 or more years out. The refreshes and gap fillers are running into walls left and right, with thermals and power draw quickly becoming unsustainable.
ppnYou can see it in the size, 2 cores 14nm Intel = 25mm2, 2 cores 7nm zen = 19mm2, really, really not that much big of a deal smaller.
Again, how is this important? Zen2 is still smaller, even in total die size with a much more powerful iGPU, performs the same or better depending on the workload, consumes much less power. How does the die size deficiency not being that bad alleviate any of this?
ppnPerhaps the unlocked CPU carries Z490, not H470.
These are mobile chips. No desktop chipsets used there, mobile chipsets are different.
Posted on Reply
#20
Vya Domus
www.anandtech.com/show/15687/intel-details-10th-gen-comet-lake-h-for-45-w-notebooks-up-to-53-ghz

Holy shit :
The base frequency of this chip is 2.4 GHz, and it has a regular 45 W TDP (sustained power), which can be run in cTDP up mode for 65 W. Two other plus points on this chip is that it is unlocked, for when an OEM provides more thermal headroom, and it supports DDR4-2933, which is an upgrade over the previous generation. Intel's recommended PL2 (turbo power) for the Core i9 is 135 W, and Intel says the recommended 'Tau' is set to 56 seconds for the i9, and 28 seconds for all the other CPUs. OEMs don't often adhere to these values for notebooks, but they are provided as a guide. It does mean that in order to hit 5.3 GHz, the Core i9 is by default allowed to take 135 W across two cores, or 67.5 W per core. Even at 60W per core, you're looking at 50A of current per core... in a laptop.
135W in order to hit the 2 core maximum boost clocks in a laptop. What are they even doing at this point ?
Posted on Reply
#21
Valantar
Vya DomusI highly doubt that, even if OEMs will want the Intel platform everywhere it just wont be possible. For one thing there aren't going to be many of these produced by Intel, the clocks are just too high and binning process will be atrocious. Secondly, the cooling will have to be ramped up considerably, don't be fooled by the 45W TDP, hell AMD's 4900HS is more like 55W under load imagine how these things are going to behave.

There just aren't going to be many laptops around featuring these CPUs (except for the i5s but those wont really be that much faster anyway). It's just something meant to steal the show.
No need to ramp up cooling; they are 45W chips, so cooling capable of dissipating 45W will keep them at or slightly above base clocks, which technically is all that matters. No need to change that just because Intel unlocks a stupidly high boost clock that will never see use in real life. It will simply throttle (well, technically not throttle, just boost less) once it gets too hot. Which for TVB is above 65C. So, these bosts are essentially non-existent. Some OEMs will no doubt go balls-to-the-wall on cooling to be able to say they have the highest clocks, but those will be the minority. As with 10th gen U-series chips, with the same core counts actual performance differences between 8th/9th gen laptops and "new" designs with 10th gen chips will be negligible if they exist at all.

The 4900HS has a ~65W boost for a short time, 55W boost for a longer time (a few minutes from reviews I've seen of the Asus G14) and then settles at 35W at ~3.5GHz for anything beyond that. These Intel chips will do the same, only the first number will be in the ~120-150W range, the second likely ~80, and the actual clock at TDP will be ~1GHz lower than AMD's. As for binning ... meh. These boosts will be so short as to make that a non-issue. Just feed it enough voltage and nearly any 14nm+++ chip can do 5.1GHz for a few seconds. It doesn't need more than that to reach Intel's (very misleading) spec.

As for i5s, try checking availability of 9th gen i5 H-series chips vs. i7s. There are far more SKUs with the latter, as Intel generally doesn't want OEMs to sell i5s with high end GPUs in the laptop space, so they cost essentially the same (or the purchase of the i7s comes with a bunch of rebates, engineering support, marketing support, etc.). The i7s will rule the roost despite being hot, slow, and generally sub-par. And that will apply for all of the 100+ H-series laptops Intel is announcing they have coming. There's no way on earth the majority of those will be i5s, and no way on earth AMD is getting anywhere near 100 designs for Renoir no matter how superior it is. Such is the power of being a market leader for a decade with massive coffers and strong ties to all OEMs. And the end user is the one who suffers.
Posted on Reply
#22
john_
Vya DomusWhat are they even doing at this point ?
Marketing. People will buy the brand name and those frequency numbers and they will probably never realise that they payd more for something inferior and for numbers that they will never see.
Posted on Reply
#23
Valantar
john_Marketing. People will buy the brand name and those frequency numbers and they will probably never realise that they payd more for something inferior and for numbers that they will never see.
Yep. Enough marketing, with enough preexisting mindshare and strong enough ties to OEMs to make sure enough designs are out there, and you keep your market leader status for another generation. There are limits to how long this can be sustained, but for a generation or two it can tide them over. This is essentially the "too big to fail" theory of product design as their current "momentum" is what carries them through, though this didn't exactly work out too well for the banks back in 2008, but Intel might be able to launch a compelling product in the next couple of years. We'll see.
Posted on Reply
#24
Aldain
Versus a 3 years old pc

Intel is cnn of marketing.
Posted on Reply
#25
bencrutz
truehighroller11 second one core and limited power and 115C temps with in one microsecond with 4 external 360radiators and two strong pumps.. Freaking joke seriously.
and at 135W (PL2) :wtf:
ValantarSo IIRC TVB requires CPU temperatures to be below 65C. Anyone care to bet how long a chip like this will stay below 65C while under a heavy single-core load in a laptop? I'm guessing the answer is a very, very short time.
IIRC tom's hardware article mentioned below 50°C
Posted on Reply
Add your own comment
Sep 26th, 2024 22:32 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts