Contribute
Register

Z590 w/ 11th Gen Intel Rocket Lake CPU

Status
Not open for further replies.
I've revisited overclocking config of new 11900K. Letting Asus AI find the OC resulted in higher temps than 10900 (as many elsewhere have reported). My rev-limited is set 98C (Tjunction is 100C for these parts) and the 11900K was bumping into this limited at 5.1 GHz all-core.

Ffmpeg h264 transcode is my all-core workload.

I've been able to reduce average temps by about (8C) 3C by using the Asus Extreme Tweaker VF Point Offset to pull down Vcore by up to 0.1V at top points. This brings average full load temps to around 90C, which is where the 10900K has been running very reliably for the last year. Now I never bump into the rev limiter. I could maybe bring temps down a bit more as I've had no stability glitches with the 0.1V reduction. Update, my full-load measurements were upset by incomparable workloads or ambient. I don't know what's going on but it's not as good as I thought.

Re performance: Full load ffmpeg performance on 8-core 11900 at 5.1 GHz all-core is about 6% lower than 10-core 10900 all-core at 5.0 GHz. GB5 multicore of 11900 is almost exactly the same as 10900K. 11900K single-core performance is 25% higher than 10900K.

For those who judge by such things, the 11900K idle temp is 36C.

I use a BeQuiet 4 air-cooler.
 
Last edited:
As mentioned in previous posts, one of my impressions about new 10th generation i9 kit has been a feeling of surprise that in spite of big benchmark gains, my new kit feels to me less fast than I would expect from the numbers as compared to my 2008 Mac Pro. My opinions are seat of the pants.

I've since replaced i9-10900K with i9-11900K, which measures a solid 25% single-core benchmark bump. This got me wondering more about perception of performance vs quantitative performance.

Here is an actual seat-of-the-pants performance comparison a simple audio import+export using Audacity.

I did screen-grabs of the import+export progress and overlaid them in iMovie. Here's what the upgrade looks like (this capture is crude as I have no skillz with this kind of presentation):

i9-11900K vs 2008 Mac Pro Audacity.gif


As I was comparing the run-times in iMovie, I decided to graph observed performance against GB5 scores.

Here's a chart which extrapolates effective GB5 scores from observed run-times, including a conjecture about 10900K performance based on local GB5 measurements and measured audacity 11900K performance.

i9-11900K-vs-2008-Mac-Pro-Audacity.png


Conclusion: The substantial performance gains are there.

But whether you have to wait a little or a lot, once you have to wait for a job to get done, the perception is that of having to wait :)

Regards,
 
Here are some numbers for my 10th to 11th z590 upgrade

To recap, i9-11900K at 5.1 GHz multicore

NOTES

I've been mystified by a long-term problem on my build that kernel_task will eat from 35–75% of a core.

A 35% wasted CPU had reduced my 10900 single core benchmarks by 12% and multi-core proportionally, about 1.5%. On this 11900, kernel_task was constantly burning 75% of a core! I felt I had to get to the bottom of this before benchmark results would have real meaning.




Tl;dr background:

On my 10900 build, I discovered by accident that the system would properly idle if OpenCore is loaded from from a thumb drive instead of the OS NVMe. As I recalled copyng the thumb drive EFI from the same source as the NVMe, I didn't exhaustively search for differences. After installing the 11900, kernel_task is using even more CPU.

A huge aside — kernel_task is an odd program, because it's a kernel process that schedules a ton of what's called "bottom-half" driver code in old Unix (e.g., BSD / Sun Solaris parlance). Top-half kernel code gets called from user-space when a program wants a device to do something for it, while bottom-half kernel code gets called (pushed) by devices (interrupts) as those devices get work done. MacOS is based on CMU's Mach microkernel which has a much richer kernel structure for device schedluling and inter-processor communication than typical Unix derivatives. The Mach microkernel is a gangly collection of SW threads which implement messaging — and therefore interlocks — that let a lot of device activity run concurrently, which is good for system performance and stability. (I don't actually know how Linux has evolved in this regard but MacOS is mosdef not Linux kernel, nor is it BSD, in spite of strong associations with the latter.)

So one of the many, many device-level chores that kernel_task performs is the efficient wasting of time when a mac mobile CPU starts to overheat or a it looks like it's overheating because temperature sensor is broke, which led o forums full of Macbook users complaining that kernel_task is hogging their CPU and slowing down their Mac, which it in fact does. But it also does a million other device-related chores, and any sort of device pathology might lead to kernel_task wasting time.




A SOLUTION APPEARS

By searching for info on Mac SW performance profiling I learned about spindump in Activity Monitor. It collects code usage statistics on the running system.

I ran spindump on my otherwise idle system, then searched the output for threads using a lot of CPU time. What stood out was a digital-audio related threads with lots of activity

On a pure hunch, I thought "AppleALC — I've never understood my own audio layout choice for boards Realtek ALC4080." It's not even mentioned in the AppleALC notes, but some reading told me that its an upgraded ALC1220, which has about 20 layout ID possibilities. I just guessed at one when originally creating this build and it worked.

As I mentioned in previous msg, trying to solve a GPU blank-screen bugaboo I found by that WhateverGreen actually isn't essential for my build. Actually, WEG figures out most configs by itself these days, so DeviceProperties for graphics is often no more than a hint to WEG that it should do a default something, which OpenCore users mistake for having made an intelligent config choice.

I have just gone through a process of elimination of various OpenCore elements to learn more about what my system really needs versus pure boilerplate.

So, once I discovered that kernel_task was busy wasting time futzing with audio, I thought maybe I should just remove my layout choice and let AppleALC decide what's best. Voila! System idle CPU is now truly idle.

OK, SO HERE ARE MY RESULTS

Asus Hero XIII z590
i9-11900K, 5.1 GHz, DDR4-3600
Samsung 980 Pro in PCIe 4.0 slot:


View attachment 548899
View attachment 548900
View attachment 548901View attachment 548902View attachment 548903View attachment 548904View attachment 548905View attachment 548906

Just for reference...
Western Digital SN750 (PCIe 3)

View attachment 548907

LAST THOUGHTS
  • Best 10900K GB5 single core was 1460.
    The 11900K provides a solid 25% single core bump.
  • 11900K needs 5.1 GHz to multicore at same score as i9-10900K. This translates into 4C higher full-load temp of 96-98C in 78F room. The system is stable overnight though.
  • PCIe 3.0 vs 4.0 makes no difference on GB5 Metal score.
  • After I sorted out the CPU waste, my W5700 GPU metal score went down 10% !! Lord giveth and taketh.
  • PCIe 4.0 NVMe shows a strong 75% throughput gain. 980 Pro is regarded as top-performers in IOPs.
This system makes no sense. It costs a fortune, is hot as hell, and just holds its own with a Studio M1 Max.

But it's been fun and challenging learning experience.

Thank you community!
Any chance you want to share your final EFI, as I've been struggling with some of the exact same stuff with idle and 11900k weird stuff on full load? Much appreciated!
 
Any chance you want to share your final EFI, as I've been struggling with some of the exact same stuff with idle and 11900k weird stuff on full load? Much appreciated!
Here's my EFI.

In the end, the CPU waste was caused by trying to use AppleALC with USB audio. By trying to run AppleALC with a layout ID, this caused kernel_task to waste tons of cycles.

Of all the zillions of posts you can find about audio layouts, the only thing you'll never see mentioned is that USB audio doesn't need AppleALC! I was confused about this forever. USB audio is its own animal.

In the attached EFI, AppleALC is enabled but is not given a layout-id, so it's dormant. You could disable/remove it.

This EFI also includes but doesn't enable NVMefix, because its not needed for the SK Hynix Platinum P41 I use.

Unrelated to EFI, I ran into thermal limits of 11900K and air-cooling which limit my build to about 4.9 GHz sustained. I use Asus Allcore Remove All Liimits and AI training in BIOS and set a temperature ceiling of 96C. Tjunction is 100C for this chip, so its well inside the green zone. The board throttles to keep temps at this level as ambient changes. It runs all-day all-night at full bore with no errors.

Geekbench5 under MacOS at 1840 / 11500, which are solid numbers for this CPU.
Same score with Windows. Under Ubuntu I see 1920 / 12500. This is as good as it gets with air cooling.

SK-Hynix benches at full gen4 speeds, no Trim issues.

I am using MSI Mech RX6600 XT for GPU, which is a good value at 98,000 Metal for $329.

Happily my build is a daily driver which runs completely reliably.

Only gaps are I skip Intel Wifi/BT in favor of M.2 Broadcom for full MacOS compat, and I've never tried to make Thunderbolt work (Maple Ridge), although I believe it's possible.
 

Attachments

  • _EFI i9-11900K 0.8.0 12.5.1.zip
    10.7 MB · Views: 134
Very cool and thanks for sharing that! I'm still tweaking the last few things, but this definitely helps on the USB audio, I thought I was going crazy.
 
@c-o-pr @OOBIDUB,

Regarding the USB Audio, you could try using the AppleALCU.kext driver instead. That was what I used in my Gigabyte Z590I Vision D build to run the Realtek ALC4080 > https://www.tonymacx86.com/threads/...igabyte-vision-d-i9-11900k-amd-rx6600.317472/

As for the Thunderbolt, it should work if it is properly defined and enabled in BIOS. You will have to add SSDT-DTPG.aml as well as SSDT-TB3-HP.aml. As for TB firmware rom flashing, I don't know of any modded Maple Ridge roms available yet. But there is a TB firmware repo here > https://github.com/utopia-team/Thunderbolt
 
As for the Thunderbolt, it should work if it is properly defined and enabled in BIOS. You will have to add SSDT-DTPG.aml as well as SSDT-TB3-HP.aml. As for TB firmware rom flashing, I don't know of any modded Maple Ridge roms available yet. But there is a TB firmware repo here > https://github.com/utopia-team/Thunderbolt
Thanks for hint. Now that you mention it... Maybe this board is a Ridge before Maple? I don't recall.

Back at Big Sur the TB HW would appear in System Report but it disappeared after an update.

A $20 NVMe enclosure becomes $100 when it's TB. I wonder why an interface chip costs as much as whole SSD?? The cheapest TB device I've found is an Apple TB3 to TB2 adapter for $50. A TB4 hub costs 1/4 of a Mac Mini! It's a goofy situation, like industry doesn't care if it gets adopted, or is a funky profit center, idk.

I have no good reason to acquire TB devices and no way to investigate it's moot for me.
 
BTW, board USB audio works natively, through the 3.5mm line in/out, but I can't figure out a way to name the audio ports. They appear as generic.

Also don't see ports for TOSlink (SP/DIF) and surround etc, but I've never tried connecting these so maybe they manifest on the fly?

It seems there's still a audio port layout config needed, but it's not handled by any Acidanthera kext?

Idk how USB audio is supposed to work out?

I found an old Apple sdk that lets USB audio device makers tailor port config details to their product but it wasn't obvious how it works, and it's a legacy tool.

Again I have no pressing need, but it would be great to know more about how this works.
 
BTW, board USB audio works natively, through the 3.5mm line in/out, but I can't figure out a way to name the audio ports. They appear as generic.

Also don't see ports for TOSlink (SP/DIF) and surround etc, but I've never tried connecting these so maybe they manifest on the fly?

It seems there's still a audio port layout config needed, but it's not handled by any Acidanthera kext?

Idk how USB audio is supposed to work out?

I found an old Apple sdk that lets USB audio device makers tailor port config details to their product but it wasn't obvious how it works, and it's a legacy tool.

Again I have no pressing need, but it would be great to know more about how this works.
Regarding the TOSlink ports they should be embedded within the 3.5mm ports iirc so you need a 3.5mm adapter for the TOSlink cable.

As for the USB audio it works the same, you need to swap the AppleALC out with AppleALCU.kext in your config.plist and enable alcid=1. If 1 doesn’t work check the AppleALC Realtek table.
 
AppleALCU.kext
Huzzah! I never noticed this version of AppleALC... This is the secret I've been looking for. Thank you!

I think months ago you even posted a reply to my question about layouts with something to the effect of "el duderino, doesn't your board have USB audio?" to which I thought Yeah... But now in context reads helpfully as "don' forget to use the U version of AppleALC" which unfortunately I've never noticed.

I went back the Acidanthera documentation and realize I've read right past it multiple times. But it never mentions USB, nor ever discusses USB in the layout docs. There's just a throwaway sentence in the ReadMe overview:

AppleALC An open source kernel extension enabling native macOS HD audio for not officially supported codecs without any filesystem modifications. AppleALCU can be used for systems with digital-only audio.
Doh! I casually overlooked the "U" of the second sentence and read it as a supportive clause for the first; as in 'AppleALC works for digital-only-audio' and read right past it thinking '
Yah sure these systems use digital audio sheesh ok great whatever digital-only means aren't all these systems digital only? Omg my config.plist circus has high-wire and elephant acts starting in rings 2 & 3!

It makes me wonder what other treats are lying around in the thousands of files that make up my OC trove. Someday I'm gunna have to read the entire "Configuration.pdf", which I think everyone can agree is the most appropriately named document of all time!

I will gripe that tech writesups are full-full-full of smeagol-my-precious little contextualities that only make sense when read at backwards from point of view of already knowing how things work and just breadcrumbs for the initiate.

I see it in my own writing. I composed a huge how-to for USBmap but doubt that I will ever post it because the more complete and detailed I tried to make it, the more I realized how many aspects I don't know and can't figure out.

2+2=5 for large values of 2.

There's a strand of McLuhanesque thought that explanations are generally not helpful. Show & tell works better. Show the student working parts and relationships and they'll put it together. Explanations become three-ring circus of undefined terms.

Esp with USB!

Note to self: lone "U" means USB.
 
Status
Not open for further replies.
Back
Top