Intel turbo boost is on but not reporting turbo speeds [SOLVED]

haigioli · 2025-05-24 13:40:19

LTS still has this issue.

It's like an update I did some time last week switched a hidden persistent bit on the BIOS, which I can't switch back on the BIOS because interface is childish.

To solve the problem, I have to make changes on what the CPU capabilities are before GRUB loads. That's how screwed up this problem is. This is NOT a solution. It's not even a patch. It's a pathetic hack.

@AidanDaOne, if your BIOS is also an overly simplistic one, I think we need a third person to confirm the same issue with mighty laptop i9's with a similar pathetic BIOS and we have confirmation on the combination of things that's causing this.

I've reached a point where I have to try and write my own BIOS using Coreboot. I don't know if I want to do that. You'd imagine somebody somewhere still believes in service, nowadays, and would actually help with this problem. I guess when AI arrived on the scene, people checked out completely, except for people in communities, such as this.

Anyway, I firmly believe that this is an issue with the BIOS that was somehow affected by an update, or just enough time with a buggy BIOS that just made some switches on its own behind the scenes. But, when the kernel reports that the CPU has no turbo because the BIOS turned it off, or the CPU doesn't support it, and you can't access pstate switches in sysfs no matter what, then there's an issue with the way your system is talking to your OS.

haigioli · 2025-06-02 21:27:54

After weeks of trying all manner of things, my solution to the problem was to disable intel_pstate altogether. It turns out that on this line of CPUs, bu my estimation, it's geared far more towards Windows gaming, than serious workstation demands from a driver point of view. The problem is with the interaction between the BIOS, which is locked, and the intel_pstate driver in its current versions in kernels 6.14.6 onward (6.14.9, as of writing this post). I dissected the driver in the kernel, trying to find ways to patch it. The patch that worked was tantamount to disabling the pstate driver entirely.

Putting the driver into passive mode in kernel, as I used to, is now - for this CPU - completely useless. Passive mode doesn't not turn off HWP, as it once used to. And in the confusion, turbo is turned off by default. Those were on the good days. On the bad ones since this started, turbo isn't reported as being available at all.

Originally, the solution was to writing a DSDT that would inject itself after BIOS and before GRUB through UEFI. Which, in essence, makes the rEFInd bootloader more suited as the system bootloader and I'm not cool with that for my setup and needs. Besides, it's just waaaaay to much hacking before getting off the ground, for my taste.

So, the most practical things to do was to disable intel_pstate with the kernel parameter:

intel_pstate=disable

And then write a bunch of new profiles for tuned, which is what I use for CPU frequency scaling management, but cpupower will do just as well, depending on your needs.

I've found that with realtime profiles, I can get each core going at max, which for the high-power cores is a whopping 5.6GHz, and the efficiency cores are 4.1GHz. With a modified throughput-performance profile, I can get very nice speed scaling on the mid 3.6GHz side, jumping to 5.6GHz range on a moment's notice. You can turn boost on, but turbo is useless, as it's a pstate driver exclusive switch. The other cool thing about this solution is, with the acpi_cpufreq driver that the CPU defaults to with pstate off, you have access to the schedutil governor. On profiles that you create in order to save power on low-demand operations, the latency between frequency jumps is much quicker then with the ondemand governor. Very handy for more workstation-oriented computing.

I have an open bug report at Intel on this, as well as a thread on one of their support forums, but I'm guessing that this will be a fix that will come way later than I'm willing to wait.

I hope this helps anyone else who is suffering from this very weird problem from a driver that works near-flawlessly on other machines I'm aware of. I'm marking this as SOLVED.

AidanDa0ne · 2025-06-03 08:15:18

Hi, I think I had tried that already and it did not solve the issue for me. Had to switch over to windows 11 for now which fixed the problem as I can barely run 3 VMs at once with an i7 ?

haigioli · 2025-08-11 11:54:51

This is definitely a BIOS bug combined with an Intel driver and intel firmware bug, because if you roll back the kernel and firmware enough, you get the right behaviour. Intel isn't being very cooperative on this and calling it an exclusive BIOS issue, and the BIOS providers are dragging their heels, leaving people like me who bought a premium system for a premium experience with a big fat lemon.

The fact that this doesn't happen on Intel's regular mainline CPUs is beyond frustrating, but the only thing one, such as I, can do is bring it to the attention of the right people and wait. Meanwhile, my solution - which is really a compromise - is all I've got until someone randomly decides to fix this.

FYI, the kernel, firmware, and ucode that seems to work with no problems is:

intel-ucode 20250211-1

linux-firmware 20250311

linux-lts 6.12.21-1

Not an ideal solution, but alas, it's documented on the intel community boards, as well as bug tracker.

Last edited by haigioli (2025-08-11 15:33:24)

seth · 2025-08-14 19:04:07

@AidanDa0ne if you still running some linux somewhere, can you please post the output of "lscpu"?

arraybolt3 · 2025-09-10 16:28:32

For what it's worth, I ran into this problem on Ubuntu 24.04 with a 6.14 kernel. My machine is very similar to OP's (Clevo with an i9-14900HX). I was able to bisect the kernel to find the commit introducing the issue, and detailed more about that here: https://bugs.launchpad.net/ubuntu/+sour … ug/2122531

I also see someone on Gentoo running into what appears to be the same thing: https://forums.gentoo.org/viewtopic-p-8 … fe8f60b0d3

The LKML thread where the patch was first posted is here: https://lore.kernel.org/lkml/2025042921 … intel.com/

kratusch · 2025-10-09 17:01:40

Iam having similar issues with turbo boost on intel_pstate(active).

$ sudo sh -c "echo 0 > /sys/devices/system/cpu/intel_pstate/no_turbo"
sh: line 1: echo: write error: Operation not permitted

$ sudo dmesg | grep -F -i intel_pstate
[    0.675728] intel_pstate: Intel P-state driver initializing
[    0.676180] intel_pstate: HWP enabled
[    5.655493] intel_pstate: Turbo disabled by BIOS or unavailable on processor

The best workaround seems using acpi_cpufreq instead of intel_pstate with kernel parameter intel_pstate=disable.

Arch Linux

#26 2025-05-24 13:40:19

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

#27 2025-06-02 21:27:54

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

#28 2025-06-03 08:15:18

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

#29 2025-08-11 11:54:51

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

#30 2025-08-14 19:04:07

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

#31 2025-09-10 16:28:32

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

#32 2025-10-09 17:01:40

Re: Intel turbo boost is on but not reporting turbo speeds [SOLVED]

Board footer