Intel Unveils Full Tiger Lake-H Processor Line-Up For Higher Performance Laptops (hothardware.com) 39
MojoKid writes: In January, Intel officially announced its Tiger Lake-H mobile platform, but today disclosed full details on the new, higher-end variant of Tiger Lake manufactured using 10nm SuperFIN technology, that brings with it a few significant platform enhancements beyond just its clock speed and core count boost. Intel is refreshing the lineup with higher-power and higher-performance Tiger Lake-H45 processors, with up to 8 physical cores (16 threads). In addition, the CPUs feature 20 reconfigurable PCI Express 4.0 lanes attached directly to the processor, which enable PCIe 4.0 NVMe RAID -- a first for any mobile platform. The platform features all of the latest IO and connectivity technologies, like Killer Wi-Fi 6 / 6E, Thunderbolt 4, and support for Resizable BAR. There are an array of consumer and commercial Tiger Lake-H based 11th Gen Intel Core H-series processors coming down the pipeline. The top-end consumer SKU is the Core i9-11980HK, which is an 8-core / 16-thread processor, with a base clock of 2.6GHz and maximum turbo clock of 5GHz on one or two cores. What also makes this particular processor interesting is that it is fully unlocked and overclockable via Intel's XTU utility. Intel has shipped millions of units volume to laptop OEMs already and expects to have laptops in market from all of the majors this month.
Re: Great, another 'stopgap' processor is all we n (Score:1, Funny)
Apple is sitting at the kids table. Albeit by their own choice.
Re: Great, another 'stopgap' processor is all we (Score:2)
Thing is, for the laptop market there's not much of a difference since heat output has to be limited due to form factor concerns.
Re: (Score:2)
Well, not for Apple laptops perhaps. Apple certainly have a habit of making laptops that thermally throttle. Fortunately they warn you about this with the howling shriek from the cooling system. Other manufacturers that don't mind the looks of having large vent holes on the side don't seem to have this problem.
I'm out of the loop. (Score:1)
Where are we on the whole Spectre/Meltdown, IME, multi-core performance (with the aforementioned avoided), and chip availability thing?
All I know that Intel's "10nm" is equal to TSMC's "7nm" and they're both really about four times that, so they're both fraudulent thugs already.
Re: I'm out of the loop. (Score:4, Informative)
What's the problem with 20 lanes?
That processor is for laptops. In a laptop the only things you will connect through PCI-e are going to be a graphics card and NVMe because lots of things will go through USB.
graphics card typically get 16 lanes. That leaves 4 lanes for the NVMe. PCI-e 4.0 cat get you 2GB/s per lane. So that is 8GB/s for Disk IO traffic (probably cap at 7 realisitically). That's a lot of storage speed already.
Am I missing something?
Re: (Score:2)
I don't know. Did you intend to joke by ignoring the 8 DMI 3.0 lanes [anandtech.com] going to the mobile chipset, the 24 PCIe 3.0 lanes coming off of the mobile chipset, and all of the HSIO lanes associated with the mobile chipset IO (like USB)?
NVIDIA doesn't even do SLI for anything below a 3090, and everybody's given up on it due to that fact. So what are you going to do with the extra PCIe 4 lanes in a laptop, exactly?
Re: I'm out of the loop. (Score:1)
NVIDIA doesn't even do SLI for anything below a 3090, and everybody's given up on it due to that fact.
mGPU; catch up.
Re: (Score:2)
You're hilarious! How may games support that now-4-year-old technology, exactly?
Re: I'm out of the loop. (Score:1)
that now-4-year-old technology
That's a very desperate attempt to make an entirely immature, alpha-version technology sound like it's past its prime and missed its mark.
Re: (Score:2)
If the shoe fits [nvidia.com]. You're the one claiming that mGPU should be driving a hardware requirement for more PCIe 4.0 lanes (I'm referring to all that other material that you helpfully clipped)
Meanwhile: [anandtech.com]
Re: I'm out of the loop. (Score:1)
You're the one claiming that mGPU should be driving a hardware requirement for more PCIe 4.0 lanes
The voices in your head notwithstanding, I merely corrected your claim that SLI is dead.
Re: I'm out of the loop. (Score:1)
Re: (Score:2)
It is. [hexus.net]
Please don't call mGPU SLI ,or vice versa. They're not the same thing. And now you're wrong on three counts. SLI, mGPU, and "voices in my head" that have official web pages for some reason.
Re: (Score:2)
I mean support for it is pretty solid in Vulkan and DX12.
I've been playing around with it for years. It works, and it works well. It's just a weird enough programming model that you don't see real adoption in things like games.
Mixing the work of unalike discrete GPUs together after the fact in your own code with their own pipelines and scheduling... isn't trivial.
SLI is much easier to work with.
I think SLI and mGPU are equally "dead in the water" due to one simple fact: 2 GPU
Re: (Score:2)
20 lanes of PCIe 4.0 blows any AMD offering out of the water and into orbit.
Current top end AMD mobile proc has 16 lanes of PCIe 3.0.
Re: I'm out of the loop. (Score:1)
And finally: I like my backdoored CPU wireless-networking-free and my wifi non-murderous, thank you very much.
Re: I'm out of the loop. (Score:2)
It will be funny because when they get to the angstrom feature size, because of the fraudulent size marketing, they will still be able to claim sub-atomic feature sizes. IBMâ(TM)s 2nm process claims 330 million transistors per millimeter squared. That means that means along each side there are 9,000 blocks per transistor. Each of those blocks is actually 110 nanometers in length, not 2 nanometers .. the reason being you need spacing wiring other devices etc. those distances are dictated by the current
Re:I'm out of the loop. (Score:4, Interesting)
Where are we on the whole Spectre/Meltdown
About the same as AMD.
My 9th gen i9:
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced IBRS, IBPB conditional, RSB filling
Vulnerability Srbds: Mitigation; TSX disabled
Vulnerability Tsx async abort: Mitigation; TSX disabled
My zen2 ryzen:
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Full AMD retpoline, IBPB conditional, STIBP conditional, RSB filling
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
You can obviously ignore the TSX shit, since there is no analogue to TSX on any AMD procs.
The important question (Score:5, Interesting)
Is is anywhere near close to matching M1 at the price/performance/powerdraw ratio?
Re: (Score:2)
Is is anywhere near close to matching M1 at the price/performance/powerdraw ratio?
I'm betting the M1 crushes it when both are running on battery alone. I'm also betting that the fans will be very loud next to the fan-less MacBook Air.
more ram then M1 and pci-e video + m.2 storage (Score:2)
more ram then M1 and pci-e video + m.2 storage (not apple locked in over priced storage)
Re: (Score:3)
If your average x86 chip is superscalar, then the M1 is ultra-fucking-scalar.
You can expect the big guys to (begrudgingly) accept the new status quo and become competitive within a couple generations.
On the power side... nobody is going to touch the M1 until they move to 5nm alon
Re: The important question (Score:3)
Don't forget: SSDs not soldered directly to the mainboard.
The RAM was bad enough, but that is enough to blacklist any product that does it immediately.
Re: (Score:2)
But now for the pros:
This passively cooled wonder easy keeps up with my i9-9980HK for an entire 7 minutes before it throttles- and it can do that using about 1/5th the power.
That is fucking remarkable.
If you cut through all the apple hype, this processor really does represent a paradigm shift.
So, I don't regret the money I spent. I've spent a lot more on laptops far less cool. But you're right: the SSD and the RAM would have been deal breakers for any "regular" machine I was looki
Re: (Score:2)
Rosetta is awesome, and it does a fucking fantastic job at running x86_64 macOS apps.
However, it would be super nice (for me) if they supported doing their dynamic translation inside of the HVF, so that my qemu virts could have that level of speedup when running an x86_64 image.
And on the 32-bit side, that's a pure Apple-ism, and just an assaholic business move.
They have decided they no longer support 32-bit anything, so R
Re: (Score:2)
Only Microsoft makes screens comparable to Apple's (across its entire product lineup). ...
You might get gimmicks like 120Hz and 144Hz but you usually lose one or more of brightness, contrast, viewing angle, color accuracy, improved color space,
(which doesn't mean that you can't find _one_ particular laptop screen that _for you_ is better than anything Apple offers - like 17 inch screen laptop for example).
However, limited RAM, limited SSD space, huge markup for more RAM/SSD, nothing user-upgradable are huge
Re: (Score:2)
Only Microsoft makes screens comparable to Apple's (across its entire product lineup).
What are you talking about? I literally have several 4k laptops with higher pixel density, and superior screen technologies (OLED) and color space than my M1.
You're making shit up.
Believe me, you haven't seen beautiful until you've seen 10-bit images on an OLED screen.
As for better GPUs, x86/32 bit support, not having to use MacOS - those might not matter. M1 is basically made for "thin and light" laptops not for desktop replacements, mobile workstations or gaming powerhouses.
the new M1 iMac, and M1 MacBook Pro would like to have a word with you.
Herp (Score:1, Insightful)
Herp Derp
Eeeh Emm Dee.
(Can't believe it wasn't already in the comments)
20 + DMI or just 16 + DMI for pcie lanes? (Score:2)
20 + DMI or just 16 + DMI for pcie lanes?
Re: (Score:2)
DMI is good for another 24 of PCIe 3.0.
Re: (Score:2)
It has better density than TSMC's 7nm node.
Initial beta released in Houston yesterday! (Score:3)
Deprecated after brief test.
They are so locked in (Score:2)