Intel Launches 11th Gen Core H-Series "Tiger Lake H"

[Damian_Ivanov]

I love how this turns out. I just bought a new configuration with a
Ryzen 5600x & 32 GB Ram and WD SN850 SSD.
Fedora boots in ~6-7sec (cold boot) I have some services turned off.
(A Fedora 34 Silveblue VM installs in 3-4 minutes)

I would not recommend an Intel processor and never did, my last one was FX8320.
My tablet/laptop hybrid Acer SA5-??? with Intel i7 was already in repair once because the battery/processor
combo which is solded under display could have (?) blewn off like a Samsung.
The display was severely swollen "off" and detached from the hull.

After repair as well I saw in dmesg
some warnings regarding the temparature, but what can I do?
It's too late now cause I use the tablet only occassionally now (after repair) and with Windows on it....
(also it's second camera is not working on Linux), neither are parts of the touchscreen of the surface
because of some windows file with closed intel firmware blob (AFAIK).

A thing I would really appreciate is like laptops certified with Intel logo must support Linux out of the box.


Certainly in the future Intel could provide competitive processors but it is a long way.
I would certainly consider buying an Intel GPU (I already have NVidia and AMD card),
I split them for virtualization. Also both brands for wayland compositor testing.

And this is even better since this will bring a bit energy into the
closed source nvidia and not closed source amd market.

A huge win!

[smitty3268]

20% lead over ZEN3 in gaming is a bit suspicious. Something is going on there, power-related most likely. Either way, looks like a decent CPUs. They should have released them in the desktop too instead that Rocket Lake garbage.

Intel's marketing "benchmarks" are a total joke, and you can't take any of those kind of performance numbers at face value. We'll see how these parts really perform once they get released and 3rd parties start testing them.

That said, it's great that Intel finally got these out and they should at least provide good competition for AMD.

[birdie]

The GPU's power draw does not matter at all when benchmarking the CPU! You always want to remove a GPU bottleneck when showcasing the CPU performance in games because you want to see how the CPU effects the performance. This is benchmarking 101. Intel was only making a statement about maximum possible performance, not battery life. It is possible, for example, that we could find these processors with 45W dGPUs and 12+ hour non-gaming battery life. Also, would it be valid to make a general claim that Rocket Lake is better than Zen 3 because you compare the performance of a system with a 5800U to a 11900K when both systems have a 2000W GPU? The logic of your argument says it is because the GPU power dwarfs everything else.

You are probably right that cooling probably does not matter, but we would need to examine the model used to know for certain. Apple has famously not not turned on their active cooling on Intel based laptops until the chips were already thermally throttling and many other OEMs are just really cheap.

That said, the CPU will operate differently depending on the TDP it is set to. For example, a CPU could have an effective 3.6 GHz all core turbo when set to a 45W TDP and a 4.2 GHz all core turbo when set to 65W due to changes in the turbo algorithm from the TDP alone (and this has nothing to do with the operating temperature). Last I checked, on Intel systems the programed TDP is one of the key factors that determines the the tau value on laptop SKUs (desktop SKUs typically have weird motherboard overrides in place). Intel could have fairly easily acquired the AMD chips from a distributor and used an open air bench platform for both systems with the CPU cTDPs set to the same value - they have done this before.

We are discussing gaming benchmarks. This article and press release are about gaming performance. The fook you're trying to drag CPU benchmarks into it? The fook you're trying to talk about a mere 20W difference for systems which consume over 150W under load? I'm done with you. Can you please stop changing the topic FFS?

[t.s.]

I don't care about ICL and TGL power consumption too much but what I care about is bloody DPTF which breaks things left and right, including for Windows users. There are literally thousands of threads on the internet from people complaining that due to DPTF their performance is subpar and DPTF is not even available for Linux in any shape or form. Why has power managerment become a proprietary technology requiring support from an OS is beyond me. Why Intel, why? Not to mention that Intel made it impossible to configure low-level power settings for TGL CPUs under Windows. Under Linux it's even worse as Linux developers disabled access to MSR in the name of "security" and "stability". And TGL results under Linux in the absence of DPTF are abysmal: 40% slower in MT mode than under Windows. And if you start blabbering about AMD Ryzen mobile CPUs - they have pretty much the same issue under Linux as under Windows they perform a lot better, sometimes substantially better.

Now I'm awaiting Volta with a new portion of insults against me.

What a weird Intel shill I am, publicly disparaging the company and its practices.

When your CPU can burn 110W power, you absolutely doesn't want people (whom according them, is amateur/ novice) other than a few selected by you to mess with the power consumption. I think it'll remedy itself when their CPU competitive again (perf/power) against AMD.

[smitty3268]

Apparently Intel's same tests show that their current (a.k.a 2 generations old) processors are already 5% faster than Zen 3. The new ones are just ~15% faster than that.

[Guest]

"Intel's top-end Core i9 11980HK model is reported by the company to outperform AMD's Ryzen 9 5900HX by 11% to 26% in various (Windows) games."

With the IGPU? Sure, DG? No... TGL is too far behind amd in power efficiency numbers, in single core workloads, the mobile 5000 amd chps hits 4.5Ghz at a peak of 15w, while a TGL requeres 30w 4.8Ghz, and they score mostl the same. There is a huge efficiency gap between these process nodes, and amd is still on 7nm, with avaiable 6 and 5. 6nm is a drop in upgrade path.

False. TDP for both Intel and AMD is at base frequency only with all cores/threads. Anything beyond that uses more power than specified TDP.

Also, Ryzen can't actually reach that specified turbo frequency on all cores, infact it's only guaranteed to reach that turbo frequency on a single core - see https://www.anandtech.com/show/14873...ncy-metrics-/3

[Guest]

Meanwhile big OEMs are still selling gen-8 Skylake CPUs in their products because the price and performance are still compelling and "good enough" for the vast majority of users. There better be something more compelling to Tiger Lake than the benchmark scores with no realistic comparisons and over-hyped 'security' they're pushing with PR, or they're going to continue looking like a creaky dinosaur both in the PC tower and the server room.

It's a new core architecture and 10 nm, performance and power efficiency is definitely much better than the 8-series........

And duh, for normal person's use, older CPUs are more than fast enough. Why do you think newer, faster CPUs would offer anything compelling for typical desktop users, who just browse the web and watch videos?

Tigerlake does have AV1 decode though, so that's a good incentive, but typical desktop users don't know what that means, so they're not going to bother until their 1080p/4K 60 FPS AV1 videos start stuttering and they think "time to get a new computer".

Edit: Also the vast majority of users don't even know what CPU their computer is running, or how much RAM it has. All they do is use it, and say "it's slow", "it's really fast", or "works well enough for me".

I mean, there's a reason people switch from HDD laptop, to Mac with an SSD and think "Oh wow, Mac is so fast and awesome, PC sucks!" because they don't know anything technical about the computers they're using.

[Guest]

20% lead over ZEN3 in gaming is a bit suspicious. Something is going on there, power-related most likely. Either way, looks like a decent CPUs. They should have released them in the desktop too instead that Rocket Lake garbage.

Have you actually seen Zen 3 CPU power usage? I mean actual power usage. Even the mobile chips use up quite a bit of power. Zen 3 H-series has 35 W TDP which is only for base frequency all-core, actual full turbo speeds uses up way more than 35 W, and AMD CPUs can't even reach the specified turbo speeds on all cores.............

They're literally using liquid metal TIM on some of these gaming laptops, because of how hot they are......

[Guest]

I don't care about ICL and TGL power consumption too much but what I care about is bloody DPTF which breaks things left and right, including for Windows users. There are literally thousands of threads on the internet from people complaining that due to DPTF their performance is subpar and DPTF is not even available for Linux in any shape or form. Why has power managerment become a proprietary technology requiring support from an OS is beyond me. Why Intel, why? Not to mention that Intel made it impossible to configure low-level power settings for TGL CPUs under Windows. Under Linux it's even worse as Linux developers disabled access to MSR in the name of "security" and "stability". And TGL results under Linux in the absence of DPTF are abysmal: 40% slower in MT mode than under Windows. And if you start blabbering about AMD Ryzen mobile CPUs - they have pretty much the same issue under Linux as under Windows they perform a lot better, sometimes substantially better.

Now I'm awaiting Volta with a new portion of insults against me.

What a weird Intel shill I am, publicly disparaging the company and its practices.

Hm, I have seen rather bad performance on Windows. On my laptop here with an i5-6200U, Windows is kind of laggy, while Linux (GNOME on Wayland) is fast. And this is on an NVME SSD. Windows aggressively sets the performance to the lowest state, but can't actually perform well at that low state, while Linux is fine on lower frequencies.

[t.s.]

Hm, I have seen rather bad performance on Windows. On my laptop here with an i5-6200U, Windows is kind of laggy, while Linux (GNOME on Wayland) is fast. And this is on an NVME SSD. Windows aggressively sets the performance to the lowest state, but can't actually perform well at that low state, while Linux is fine on lower frequencies.

Don't forget that in windows land, the background processes are a-plenty. Antivirus that will turn on when there's something happening, optimizing windows, memory optimatization, search for updates, chrome-firefox-adobe-and the likes that all have background process, etc.. etc.. Try windows 7 without that kind of BS. It's blazing fast.