Intel Core i5 12600K / Core i9 12900K "Alder Lake" Linux Performance

[sophisticles]

At the risk of asking a dumb question, isn't this a single-thread performance comparison ? How is the number of cores/threads relevant ?

You probably know this better than I, but there is no such thing as a single-threaded application, including benchmarks.

There are always background tasks running and all GUI apps are multi-threaded to a certain extent. at a bare minimum there is at least one thread for I/O, one thread for the GUI and so on. If there isn't then the GUI will lock up and I/O operations will stall.

Now granted the actual impact of having more cores is minimal in so-called single-threaded benchmarks, but I would think in the case of Alder Lake, if one could lock the I/O threads and the GUI drawing threads to the E-cores, that would give an advantage to the P-cores compared to Intel and AMD processors of equal "performance" cores that lack E-cores.

[Laughing1]

Okay, what about Clear Linux distro? https://clearlinux.org/ It is optimized for Intel after all.

[Laughing1]

Alder Lake: Fedora 35 and the 12900k

[sophisticles]

Okay, what about Clear Linux distro? https://clearlinux.org/ It is optimized for Intel after all.

No it's not, the only "optimization" is that it used the performance governor by default and the compile target is Skylake AVX-512.

To run Clear on Alder Lake Michael will have to disable the E-cores, thereby enabling AVX-512. Don't know if his motherboard allows for that.

[yump]

yump,

Thank you! I have been torturing myself with the choice between the hotter/louder eleventh generation CPUs with the better integrated GPU vs. the quieter, less capable tenth generation CPU/GPU integrated solution. Quiet is primo for me - I'll probably opt for quiet over performance.

Then again . . . YouTube videos are of interest . . .

Next year for an appropriate 12th generation / Ubuntu solution is farther out than I had hoped. I think will pull the trigger sooner.

Abysmally slow? Abysmally slow? Quiet - it knows what you are saying.

This computer is not slow, as such. I just think of it as being very - thoughtful.

I just looked at the link that you were kind enough to post - it's making sense. Any idea of the noise level on an eleventh generation vs. a tenth? I just have my reading and conjecture - no direct experience.

For the sorts of tasks you describe using your computer for (web browsing, email, printing), CPU generation will make very little difference to noise, because those are very low power workloads. The only exception is internet video, which can require a moderate amount of power to decode, if the CPU has to do it. For that reason, I would expect the 11th generation to be somewhat quieter, because it has hardware acceleration for AV1 video, which is used on YouTube and will be more widely adopted in the future.

Noise in your sort of application depends almost entirely on the design of the cooling system (how big the CPU heatsink is, how the fan control is set up, how well ventilated the case is, etc.). If you were building it yourself, you could have a computer that would handle those tasks without any moving parts, by choosing an over-sized power supply with low-load-fan-stop, and similarly configuring the fan control to stop the fans below some moderate temperature. At that point, the only noise would be "coil whine", which is high-pitched squealing sounds produced by inductors and capacitors on the circuit boards. Unfortunately, coil whine is pretty much never mentioned in spec sheets, and if anyone is trying to minimize it, they aren't saying so publicly. How whiny any particular board is also has pretty big manufacturing variation, so you can't rely on other people's reports for the same motherboard.

[yump]

You probably know this better than I, but there is no such thing as a single-threaded application, including benchmarks.

For one thing, I have a single-threaded laptop on the shelf behind me.

For another thing, those threads you mentioned are almost completely idle while the benchmark is running, and there's no reason for the kernel to schedule them on the same core as the benchmark worker thread.

[Archestratus]

Yump wrote:

"For the sorts of tasks you describe using your computer for (web browsing, email, printing), CPU generation will make very little difference to noise, because those are very low power workloads. The only exception is internet video, which can require a moderate amount of power to decode, if the CPU has to do it. For that reason, I would expect the 11th generation to be somewhat quieter, because it has hardware acceleration for AV1 video, which is used on YouTube and will be more widely adopted in the future."


Yump.

I have not yet figured out how to run quoted material through the board with the shaded blue background - so quoting is a bit makeshift for now.

On to content -

Thank you! I did not know. I was just going by TDP and reputation of 11th generation CPUs running particularly hot. I have never actually been near one.

So - TDP is just when things are revved up? Not hotter idling and running non-demanding tasks?

In that case - I might as well request the 11th generation CPU - especially considering what you are writing of "hardware acceleration for AV1 video" in regards to YouTube. All news to me. I probably do access YouTube daily or so.

You wrote further:

"Noise in your sort of application depends almost entirely on the design of the cooling system (how big the CPU heatsink is, how the fan control is set up, how well ventilated the case is, etc.). If you were building it yourself, you could have a computer that would handle those tasks without any moving parts, by choosing an over-sized power supply with low-load-fan-stop, and similarly configuring the fan control to stop the fans below some moderate temperature. At that point, the only noise would be "coil whine", which is high-pitched squealing sounds produced by inductors and capacitors on the circuit boards. Unfortunately, coil whine is pretty much never mentioned in spec sheets, and if anyone is trying to minimize it, they aren't saying so publicly. How whiny any particular board is also has pretty big manufacturing variation, so you can't rely on other people's reports for the same motherboard."

My preference would be to upgrade an old rig - with oversized heatsink and fan running low and slow - I prefer a big downblowing fan - cool the whole board - CPU possibly running undervolted. I have a real nice Microtel Walmart Lindows beige box - the one with the zoomin' 800MHz Via C3 processor - looking for a little bit of an upgrade. But my kind benefactor prefers to go through Dell - we all have our reasons for what we do - so a home build is just not going to happen.

I used to follow SilentPCReview - Mike Chen's wonderful web reviews and community - and I did put what I learned from that site to good use. But Mike moved on. The site lost it's soul. Mike, we miss you, wherever you have gone.

Ave atque vale.

In the past, as per SilentPCReview, Dells have been notorious for being completely resistant to aftermarket SilentPC modification - everything fixed and proprietary, both with nonstandard physical construction and nonstandard wiring - so I want to try to get this as right as I can out of the gate without having to try to modify a Dell. The integrated CPU/GPU is a win with the absence of a separate GPU card and likely audible GPU cooling fan. The SSD should get rid of some spinning hard drive noise. That leaves only the the CPU HSF, the power supply fan and any case fans. I have never experienced a bothersome coil whine situation - wish me luck. If I do have to quiet it down, on a Dell the motherboard BIOS will likely be locked down to prevent undervolting. Heat-sink fan modification may be impossible, as well, due to proprietary wiring and Motherboard construction - I dunno. So the only tactic I would have left to quiet the thing if needed - besides an ungainly sound reduction box - might end up being to put the CPU, RAM, SSD and floppy drive into another box and abandon the rest of the Proprietary Dell. All I would need would be a ~$100 motherboard (guessing) - and I may have to upgrade a ~2008 (300-350 Watt?) Corsair Power Supply, I do not know.

It would be best, instead, if I just obtain a relatively quiet box to start with and not have to go down the aftermarket quieting path at all.

Eleventh generation i5 is beginning to look like my sweet spot.

Any idea if current Dells will take aftermarket HSFs? I can't seem to find anything with a web search - so it seems unlikely.

Thanks, again -

Archestratus

[birdie]

deepl.com is the new cool kid on the block regarding translations - give it a try if you have a few minutes; for me it works better in most cases.

I perfectly know about deepl, been using it for a few years but they don't have a browser add-on to translate web-pages on the fly which makes reading articles with tables and pictures quite a tedious task.

And yes deepl is heads and shoulders above any AI translator I've ever dealt with.

[name99]

The M1 Pro and Max do have Turbo for the big cores. I do not say they are not impressive, far from it, but they have a node advantage over AMD and Intel, because of the second generation 5nm TSMC process (vs. 7nm TSMC for AMD and Intel 7, formerly 10ESF).
As for x86 days being over, a recent interview on AnandTech makes some great points about the relevancy of ISA wars to core design. Basically Zen was designed in tandem with an ARM core and the ISA has far lesser impact on microarchitectual design than people give it credit for.

This ("turbo") is somewhat unclear.
In the most constrained environments (definitely phone, possibly iPad) Apple very mildly reduce their peak frequencies, so for A14 in iPhone when only 1 P core is active, it can run at 2.98 GHz, when all 2+4 cores are active the P-cores run at a max of 2.89GHz.
Other Apple SoCs are similar, though the frequency deltas have become flatter over the years (ie less and less of this "turbo'ing" -- with A15 the single-core peak is 3.24, the all-core peak is 3.18) .

So there is an extremely mild drop off in the phone environment.
It's unclear that even this trivial drop-off translates to the mac environment, ie it's possible that those cores are just all allowed to run at peak frequency if they want, with only thermals determining frequency reduction (ie if the fans aren't doing their job -- and maybe this thermal overload can simply never happen unless the fans are somehow blocked?).
I don't think we've ever seen numbers for how the OS either configures the max core frequencies in AS macs, or how they look on real code.

[smitty3268]

It's unclear that even this trivial drop-off translates to the mac environment, ie it's possible that those cores are just all allowed to run at peak frequency if they want, with only thermals determining frequency reduction (ie if the fans aren't doing their job -- and maybe this thermal overload can simply never happen unless the fans are somehow blocked?).
I don't think we've ever seen numbers for how the OS either configures the max core frequencies in AS macs, or how they look on real code.

The M1 can only run a single p-core at 3.2GHz if all three other p-cores are in deep sleep mode.

The moment you spin up a 2nd p-core they both drop to 3.1GHz.

3 or 4 p-cores active throttles all of them down to 3.0GHz.