core i7 i5 (Lynnfield) turbo boost not working

The i5 support seems to be a little behind, even on the latest from Koala beta (which I went for having seen the same Phoronix article explaining that the 2.6.31 Kernel was required for compatibility).

In addition to the Turbo Boosting not working, I have also found performance to be on par or below that of a Core2Duo chip of similar Ghz, lm_sensors unable to detect/monitor the chip, and issues with the CPU scaling defaulting to a maximum (https://bugs.launchpad.net/ubuntu/+bug/436356 and https://bugs.launchpad.net/ubuntu/+bug/359559).

Given the popularity of these processors, I'd hoped that these things would be ironed out for the Koala release.

If I can supply any further information, please let me know.

MG

I have an ASUS G51J laptop with a Core i7 720QM processor. I myself found that the Turbo Boost is definitely not working.
I'm running Karmic Koala with the 2.6.31-15 Kernel.

From my experience, I found that the ACPI support might be broken. When I COMPLETELY disable ACPI support through the kernel boot options, I see a major performance boost when benchmarking using single threaded applications, compared to the same tests when using ACPI support. I tried other ACPI settings, but non of them seemed to have positive effects, except for disabling it entirely. Of course when I disable ACPI support, I loose all power management together with cpufreq support, plus I'm only able to access one of my 4 cores.

I've read that the recent kernels support this architecture including the Turbo Boost feature. Apparently some desktop users have posted to get their system fully working. This are the only informative sites I found, specificly about Turbo Boost on Linux:
http://kbase.redhat.com/faq/docs/DOC-17124
http://kolbusa.livejournal.com/71066.html

I've had issues with both the HP dv6t and Alienware m15x.

Here are some logs with TurboBoost enabled and disabled.
http://www-personal.umich.edu/~mjbommar/m15x/

I've done some performance testing here:
http://global.phoronix-test-suite.com/index.php?k=profile&u=mjbommar-10157-25021-1778

You can see that Ubuntu virtualized under VirtualBox on Windows 7 outperforms Ubuntu 9.1 running natively by quite a bit. This was the case with the dv6t, though there were also thermal issues with that machine (buggy BIOS).

I also compiled 2.6.32-rc8 without ACPI, SMP, or SMT. The scimark2 scores were around 530 again, indicating that there was no performance gain even when the OS was only accessing one core.

I'd say this is a pretty big issue. Under native Ubuntu, an M4400 with DDR2 and an C2D X9100 currently beats the m15x with DDR3 and an i7 820.

Michael,

I ran your phoronix test on my Core 2 Quad Q9550: http://global.phoronix-test-suite.com/index.php?k=profile&u=nikolay-16115-4413-8672

I just ordered my core i7 upgrade - i7 920 2.66GHz. I will post results with it as soon as I get the parts and get my box upgraded.

The Core 2 Quad beats your core i7 results, but since your results are with Core i7 820 @ 1.73GHz and the Q9550 runs at 2.83GHz I am hopeful that I will see a performance benefit from my upgrade even if Ubuntu still has problems with Turbo Boost.

Cheers,
Nikolay

... and here's the results from my ThinkPad T61 with Core(TM)2 Duo CPU T7800 @ 2.60GHz, just for the sake of comparison and because PTS makes it so easy to do this (kudos to Michael Larabel for giving us such a great tool!):

http://global.phoronix-test-suite.com/index.php?k=profile&u=nikolay-14005-26954-1202

Well, note that the Q9550 is not as fast as Ubuntu 9.1 under VirtualBox on Windows 7 (the fifth column).

I have benchmarked a wide variety of calculations that I run on Ubuntu/VirtualBox and the architecture is certainly faster for most tasks even under virtualization - the issue is just that we could unlock even more performance if this were not running virtualized.

I discovered this interesting article about Turbo Boost and operating system support.

http://computing-intensive.blogspot.com/2009/09/how-to-make-turbo-boost-work-under.html

It mainly says that, for an OS to support Turbo Boost:

   1. The said OS must support ACPI
   2. The BIOS in use must support ACPI
   3. EIST and Turbo Boost BIOS option must be turned on
   4. CState should be set
   5. The said OS must turn on its power management features, for both Pstates and Cstates
   6. The Pstate entries, P1 should be corresponding to the chip default frequency

Someone should check for these features in the linux kernel (mainly 4-6)

Windows 7 supports it, why not Ubuntu.

Just ran the test on core i7 920. It appears that Ubuntu runs fine under the 920 Bloomfield series processor: bare-metal Ubuntu is slightly faster than Ubuntu under VMware or VirtualBox on Windows 7:

VMware:
http://global.phoronix-test-suite.com/index.php?k=profile&u=nikolay-6078-12767-16686

VirtualBox:
http://global.phoronix-test-suite.com/index.php?k=profile&u=nikolay-31301-25415-1722

Cheers,
Nikolay

Hello nikolay,
  Right, I have three i7 920 desktops on the X58 that all run fine with TurboBoost.

  It's my mobile i7's on the PM55 (720,820,920) that have given me problems.

my mobile i7 720 only show frequencies up to 1,6Mhz. I've tested it with karmic and jaunty. when i try to switch from gnome to console or reboot the machine, i get a message "overheat of processer"

the notebook is not realy hot, but the overheating message comes on every shutdown or reboot.

http://code.google.com/p/i7z/

On my m15x (laptop) with an i7 it says Turbo boost enabled, but when I try to max out one cpu and look at it the multiplier does not go to full speed, so I'm definitely not getting 100% out of it.

http://code.google.com/p/i7z/issues/detail?id=1

Some discussion there trying to figure out, seems I dont enter C6 at all.

I've also seen this mentioned a lot of places:
"Turbo is enabled if cpufreq selects the (max+1) Mhz frequency"

Checking scaling_available_frequencies I see this:
1597000 1596000 1463000 1330000 1197000 1064000 931000

Checking scaling_cur_freq under load I see:
1597000

But running a task * 4 is exactly as fast as running it * 1, which leads me to believe turbo boost is not working properly.

same problem with a dell 1557 (i7 q720). frequencies go up only to 1.6ghz. windows 7 show correctly frequencies up to 2.6ghz.

sorry, I forgot to say that I have also tested my processor with lucid alpha 2 and I observed the same problem.

Now posted here http://bugzilla.kernel.org/show_bug.cgi?id=15064

There is a lot of confusion about turbo boost. /proc/cpuinfo will not report correct freq (only reports non turbo frequencies)
Up to recently I thought my turbo boost was not working on my i7 laptop as cpufreq reported max 1.6ghz. With the following util I found out that my frequencies where up to 2.8ghz No cpufrequtils needed, it all done on the hardware...

http://code.google.com/p/i7z/

http://code.google.com/p/i7z/
gave me some hight frequencies too (over the standard 2.8 GHz of my i7 860), but i`m not quiet positive about its validity, since from time to time it gave me frequencies like 7 GHz.

http://bugzilla.kernel.org/show_bug.cgi?id=15064

Resolved upstream.

Worth mentioning that this is in 2.6.33, which wont be in karmic or lucid, and it's a very simple fix (wrong assumption in the acpi code). It should be atleast in lucid's kernel.

[WORK AROUND], [SOLVED], [NOT SOLVED]. ( I have inserted these flags due to the frustration of users requesting "Turbo Boost" and the advantages of a 1.2 GHz increase in CPU speed.)
Work around # 1: (Masters) You can recompile your current kernel and include this module here http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/utils/. This has been patched into 2.6.32 and 2.6.33 (Confirmed by me for 32 and 64 bit).
Work around # 2: (Seasoned) Download KernelCheck, autocompile the 2.6.32 kernel and your modules, (no need for the 2.6.33 rc5 or 6, the 2.6.32 mods work fine), and enjoy. (Note, you will have to reconfigure or search for the fixed DEB packaged that allows KernelCheck to download from the correct source).
Work around # 3: (Familiar) (What I did after playing with KernelCheck), Goto http://kernel.ubuntu.com/~kernel-ppa/mainline/, download 2.6.33 RC5 or later, install the kernel and sally forth.
NOTE*** After installing the 2.6.33 RC5 kernel and modules, Virtualbox requires that you D/L a beta version due to the changed in the kernel and if you have NVIDIA drivers, you must apply a patch in order for the binaries to install on your system. If you need some help with the after-effects, I will start a new thread. I am tryinf to keep this related only to the lack of turbo-boost.

If anyone would like to create a more step-by-step process, feel free. I challenge all of you to create these processes so that the average MS to Ubuntu/Linux cross-over user an tweak their systems.

Have fun!

Stupid keyboard...

@Bacchus: Thanks for your posting on the workarounds. What patches for NVIDIA do you recommend be applied?

I recommend going here. http://www.nvnews.net/vbulletin/showthread.php?p=2159641 This page will guide you through the process. I don't want to get too technical and go off topic. This page is pretty step-by-step and the patch files are posted for download. I will start another thread on applying patched for NVIDIA drivers if people request additional help. Also, I had some quirks with the 195 beta driver on 2.6.33 RC5, I recommend installing the stable 190.53 with the patches attached. x-86 and x-86_64 have been tested on my end; ALL JOY.

Also worth mentioning, ext4 performance has been suffering due to some kernel regressions. (Especially in 2.6.32 and .33) Since the turbo boost issues are predominant on mobile platforms and mobile platforms have a battery, I thought it to be appropriate to include this link which will give your ext4 a "kick in the pants". http://blog.loxal.net/2009/04/tuning-ext4-for-performance-with.html I know this link was made for solid state discs (SSD's), but they work extremely well for conventional HDD's as well. Instead of a commit time of 100, I personally use 30. Default value is 5. I also applied some other tweaks in my /etc/fstab... (included).

UUID=bb5155eb-2b46-4c70-870a-367ebeecc3c5 / ext4 barrier=0,noatime,nouser_xattr,nobh,commit=30 0 1
# swap was on /dev/sda6 during installation
UUID=b786b071-fbee-4b2c-a5a8-b838a2489316 none swap sw 0 0
/dev/scd0 /media/cdrom0 udf,iso9660 user,noauto,exec,utf8 0 0
/dev/sda1 /media/sda1 ext4 users,user,noatime,nodiratime,barrier=0,nouser_xattr,nobh,commit=30,data=writeback 0 0

Bacchus wrote on 2010-02-02: #20
"...This has been patched into 2.6.32 and 2.6.33 (Confirmed by me for 32 and 64 bit)...."

When you say this has been patched into 2.6.32, are you referring to the mainline kernel (now 2.6.32-9), the Ubuntu kernel in Lucid (now 2.6.32-15)?

If I am running the distribution kernel in Lucid, is the turbo mode for i5/i7 mobile processors currently recognized and functional?

The latest ubuntu lucid kernels work.

I'm about due for a rebuild of Ubuntu 9.10 on my HP DV-3085DX (i720). I will re-assess the easiest method to get Turbo Boost up and running with Karmic and post my results. (I only rebuild due to my frequent experimentation with everything...) I have confirmed that Lucid 10.04 (32 and 64 bit) work like a champ with Turbo Boost. Unfortunately 10.04 is still a little too buggy for my to rid myself of 9.10.
I know the 2.6.32 generic mainline has been fixed. I don't know if any of the 2.6.32-xx for download at http://kernel.ubuntu.com/~kernel-ppa/mainline/ have been updated. I will go through the build logs and look for the patch. When I find it I will post my testing results with any quirks (if any) that I come across.
-Stand by and happy patching!

I have a notebook with i7-720qm and I use Ubuntu 9.04 (which is still supported and updated): is this bug patched also there? Kernel version is 2.6.28 line.

But if I use a bugged linux kernel version in VirtualBox as a guest machine with hardware virtualization (using VT-x, nested paging or EPT, VPIDs or Virtual Processor Identifiers) in a Windows 7 host, is this bug still valid? Or everything is managed by Windows 7 host?

Thanks!

The bug is patched in 10.04 but not Karmic/Jaunty, so you'll need to make
the jump or compile a new kernel yourself if you want to stay on
Karmic/Jaunty.

You should get full speeds (less virt overhead) on the Windows 7 host, as
Win7 will have already set the machine in a state allows TB.

Thanks,
-Michael J. Bommarito II
University of Michigan
http://www-personal.umich.edu/~mjbommar/

On Wed, Mar 10, 2010 at 10:52, Teox <email address hidden> wrote:

> I have a notebook with i7-720qm and I use Ubuntu 9.04 (which is still
> supported and updated): is this bug patched also there? Kernel version
> is 2.6.28 line.
>
> But if I use a bugged linux kernel version in VirtualBox as a guest
> machine with hardware virtualization (using VT-x, nested paging or EPT,
> VPIDs or Virtual Processor Identifiers) in a Windows 7 host, is this bug
> still valid? Or everything is managed by Windows 7 host?
>
> Thanks!
>
> --
> core i7 i5 (Lynnfield) turbo boost not working
> https://bugs.launchpad.net/bugs/429036
> You received this bug notification because you are a direct subscriber
> of the bug.
>
> Status in The Linux Kernel: Fix Released
> Status in “linux” package in Ubuntu: Confirmed
>
> Bug description:
> According to
> http://www.phoronix.com/scan.php?page=article&item=intel_p55&num=4,
> Intel core i7 and i5 (Lynnfield) processors do not work well with ubuntu
> (both Ubuntu 9.04 and Ubuntu 9.10 Alpha 5).
> In detail, turbo boost does not work and occasionally creates performance
> problems.
>
> To unsubscribe from this bug, go to:
> https://bugs.launchpad.net/linux/+bug/429036/+subscribe
>

Thank you for answering.

But if turbo boost is managed by Win7 host, I will use turbo boost also in Ubuntu 9.04 guest machine running on Win7 host, even if 9.04 is still bugged, right? And is this true also if I use hardware virtualization (VT-x, nested paging or EPT, VPIDs or Virtual Processor Identifiers)?

Another question: if I install Ubuntu 9.04 directly (not as virtual machine), could I harm my hardware, since it isn't patched? Or just turbo boost doesn't work?

Thank you again.

Dear all,

how can i test if the turbo is working? I run the latest lucid kernel in karmic and i do not see deep c-states in "sudo powertop -d".

Are there any other tests?

Cheers!

http://code.google.com/p/i7z/ should show it.

-If you are using a windows 7 host, Ubuntu won't care about TB. Windows 7 will manage everything just fine. In fact, you could run good 'ol MS-DOS 5.0 in VBox and turbo boost will work on in your VM.

-Here is an easy way to find out if turbo-boost is actually working (in a real world scenario)...

sudo apt-get install hardinfo && hardinfo

Run CPU N-Queens. The result for my turbo-boosted (properly working 2.6.32 kernel) is about 0.88 sec.
When I run CPU N-Queens on the "stock" 2.6.31 Kernel the result is over 2 seconds.
(N-Queens is a single threaded benchmark).

(A new menu entry should show up under "System Tools" called "System Profiler and Benchmark". (Just in-case you would like to re-run the program in the future.))

Just as a reference, I have an I720 1.6 GHz; 2.8 w/ Turbo.

Teox wrote"
Another question: if I install Ubuntu 9.04 directly (not as virtual machine), could I harm my hardware, since it isn't patched? Or just turbo boost doesn't work?

Thank you again."...

Lack of Turbo Boost on a native (bare-metal) ubuntu 9.04 or any other OS will NOT harm you computer in any way. However, since the CPU can not enter deeper sleep states, batter performance will suffer. (If you are running a desktop, higher electric bill.) I was about to get an extra 45 min out of a charge when I installed the 2.6.32 kernel. Is there a reason why you are running Ubuntu 9.04?

Edit.... batter = battery

Extension from comment #27.... I'm about due for a re-build... Ubuntu 9.10...."
Here is a pretty step by step method.... Confirmed for Ubuntu 9.10 32 and x86_64 versions.

***Warning, you will have broken modules after this upgrade, i.e. Virtual Box, Nvidia drivers, BFCutter, etc...
    Your modules will have to be re-configured / re-installed. This is why I highly recommend step #1. (Of course the is no different than a normal kernel upgrade when Canonical pushes a new kernel update.)

1) I recommend (but not required) to start with a clean install.

2) Perform all of the distro's updates.

sudo apt-get update && sudo apt-get upgrade
***Do NOT install any proprietary drivers, do NOT install Virtualbox, etc... If you can help it, don't worry about your wifi drivers (if needed) until after the kerenel upgrade.

3) Lets update the kernel the easy cheesy (but works very well) way....

For 32 Bit, open a terminal

### This downloads the Ubuntu PPA 2.6.32 32-Bit Kernels to your desktop.

cd Desktop && wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32.9/linux-headers-2.6.32-02063209-generic_2.6.32-02063209_i386.deb && wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32.9/linux-headers-2.6.32-02063209_2.6.32-02063209_all.deb && wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32.9/linux-image-2.6.32-02063209-generic_2.6.32-02063209_i386.deb

For 64 Bit, open a terminal

cd Desktop && wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32.9/linux-headers-2.6.32-02063209-generic_2.6.32-02063209_amd64.deb && wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32.9/linux-headers-2.6.32-02063209_2.6.32-02063209_all.deb && wget http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.32.9/linux-image-2.6.32-02063209-generic_2.6.32-02063209_amd64.deb

4) Install the new kernel.
### This will install the new kernel.
sudo dpkg -i linux-*.deb

5) Restart.

Turbo-boost should be working like a champ!

Bacchus wrote:
"... Is there a reason why you are running Ubuntu 9.04?"

I'm running 9.04 also because I have a big software compiled in it and I have no time to re-install it. And I don't want to risk that after an upgrade it doesn't work.
Ubuntu 9.04 support will end on April, then I will install Ubuntu 10.04.

Teox wrote:
"...I'm running 9.04 also because I have a big software compiled in it and I have no time to re-install it. And I don't want to risk that after an upgrade it doesn't work."

I concur with your decision not to experiment with your kernel. However, I would recommend a program called image for Linux http://www.terabyteunlimited.com/image-for-linux.htm . There is a 30 day free trial. If you like it, pay the $29 bucks or so for a license. The only reason I recommend Image for Linux (IFL) is because it is extremely simple to use, backs up all of your master boot records, and restoring is a snap. It would allow you to make a snapshot, play around with your system, and if you break it, you can be up and running in less than 10 min. Frequently, I make a restore DVD for anyone who wants to break form Windows but would like an easy recovery system.

Bacchus: Even if you are not affiliated with TeraByte, I object to your advertising products here. This is a bug report for a specific kernel issue, and not a forum for advertising. If you felt that this product was applicable to Teox's comment, you could have sent your message directly to him/her without broadcasting an advertisement to all subscribers of this issue.

Sorry about that... won't happen again... and I'm not affiliated with any companies.

CPU: core i5 750

MB: intel DP55WB
BIOS: WBIBX10J.86A.0251.2010.0302.2003 (03/02/2010)

OS: ubuntu 9.10 (karmic)
Kernel: 2.6.31-20-generic

gnome cpu scaling frequency applet reports 1.20GHz to 2.67GHz

i7z reports multiplier 9x to 23.96x

can i assume turbo boost is working for even a standard karmic system on 2.6.31?

Bacchus said hardinfo CPU N-Queens is single threaded and his result is 0.88s, but what i get is all 4 cores max out 100% @ 21x with a result of 8.65s! any idea what's up?

Elsie wrote "Bacchus said hardinfo CPU N-Queens is single threaded and his result is 0.88s, but what i get is all 4 cores max out 100% @ 21x with a result of 8.65s! any idea what's up?"

Try this,
1) Shutdown compy, wait, turn-on computer.
2) Open hardinfo
3) Generate report
4) Post your results.

2.6.32 Kernel
Blowfish 3.778
Cryptohash 309.672
Fibonacci 2.593 *
CPU N-Queens 0.887 *
FPU FFT 1.516 *
FPU Raytrace 5.811

2.6.31 Kernel (9.10 Default)
Blowfish 4.466
Cryptohash 252.071
Fibonacci 4.115
CPU N-Queens 1.540
FPU FFT 2.538
FPU Raytrace 21.473

As you can see the 2.6.32 has some MAJOR performance advantages when using a CPU with Turbo Boost. Especially the benchmarks denoted with a *. A side effect is that SMP programs do see a slight boost as well. I assume that as the program initially runs, the first thread will throttle to max boost,. The second thread that starts brings the 2 working cores to approx 80% max TB, and when all of my 8 cores kick in I'm running 57% 1600/2800 of Turbo.

Elsie, I recommend (if not on a production machine) following my steps in post #36. Skip step #1. When you are installing the new kernel, make sure (when prompted), you install the maintainers version so grub will update with both kernels. (You current kernel and the new one you are installing will be available for selection at boot.) If everything works well then you can re-install all your modules / headers, etc.
Hope this helps!

I failed to mention that everything should continue to work fine on your old kernel even after installing the 2.6.32 until you rebuild your mods into the new kernel. Take note of the hardinfo report on both of your kernels.

I had another thought.... You could also D/L Ubuntu 10.04, burn to USB, enable third party repo's, install hardinfo, benchmark and see what results you get that way. When you create your "live" space for storage, increase the default 128 MB to 512 or so. Otherwise the you won't have enough space in your persistent file to d/l the app cache and hardinfo. This method will allow you to temporarily see if the 2.6.32 kernel will activate your Turbo Boost without modifying your system at all.

Sorry for my late response - i made a fresh install of 10.04.

Here are my benchmark results (i7-Q820):
Blowfish 3.69
Cryptohash 299.91
Fibonacci 2.46
CPU N-Queens 0.84
FPU FFT 1.48
FPU Raytrace 5.61

Lucid rocks !

dfriel, good to know that Lucid works for you but Lucid is still very broken. If you need full functionality AND the performance benefits of 10.04, I would recommend installing 9.10 then following the steps on #36. I have had several headaches with 10.04.
Good Luck!

Hi Bacchus,
i have lucid running on my netbook since alpha2 without major problems, so i decided to install it on my main notebook. I have had no isses so far - i'll keep my fingers crossed ;-)

Thanks for your support here!

Cheers, dfriel

Hmm interesting.

On my i7-860 that tool is showing I'm getting turbo boost - but even for something that only uses one cpu that
tool is showing all of the cores running at 2.933GHz rather than the normal 2.80GHz; I thought turboboost was
only supposed to raise one cores frequency? i7z-0.21.4 on 2.6.32-19-generic #28-Ubuntu SMP Thu Lucid.

i7-860 on an Asrock P55M Pro

Dave

On my Dell Studio 1557 turbo boost now works as expected. Also idle temperature is 5-10 °C lower than with Karmic.

Updated to linux 2.6.33 but how do I check if I have turbo enabled? I can't make i7z:

i7z.c:21:21: error: ncurses.h: No such file or directory
i7z.c: In function ‘main’:
i7z.c:233: warning: implicit declaration of function ‘initscr’
i7z.c:234: warning: implicit declaration of function ‘start_color’
i7z.c:235: warning: implicit declaration of function ‘getmaxyx’
i7z.c:235: error: ‘stdscr’ undeclared (first use in this function)
i7z.c:235: error: (Each undeclared identifier is reported only once
i7z.c:235: error: for each function it appears in.)
i7z.c:236: warning: implicit declaration of function ‘refresh’
i7z.c:240: warning: implicit declaration of function ‘mvprintw’

I should point out that my i7 is a Clarksdale, not a Lynnfield.

I just purchased a hp dv6-2190us and loaded the latest ubuntu today (Lucid). I also dist-upgraded to make sure i was running the latest kernel. My laptop contains an i7 720qm, and turbo boost does not work at all. I don't see how this bug is fixed at least for my laptop.

removed deprecated team assignment.

~JFo

@Dan Weber
This bug was solved with a patch that fixed a broken acpi implementation in the kernel. There can be many reasons why it might not work on your specific hardware, but the original title of this bug report is far too broad.

Also please make sure your checking the right way, this will not show up correctly in a lot of graphical tools. Attaching turbostat (should be in the newest sysutils if I remember correctly), you can use that to check.

Hi there Anders,

So I'm sure I'm checking the right way, but I'm not sure how to
alleviate the situation. I have an i7 720QM in my laptop. I run in
powertop it says I'm in turbo mode. I also check i7z and turbo stat.
When I'm running the latter two applications it never exceeds 1.6ghz
under full load. Turbo boost on my laptop should be up to 2.8ghz. I
have yet to see this on Linux. On windows it properly gets to that
number while benchmarking. I'm not sure what it is. Let me know if you
have any ideas.

Thanks,
Dan

I have a Dell Latitude E6410 with i7 processor (4x Intel(R) Core(TM) i7 CPU M 620 ). I am running Ubuntu 10.04.1 LTS, after reading above (and also https://bugzilla.kernel.org/show_bug.cgi?id=15064 ) I updated the kernel to "Linux 2.6.35-020635rc1-generic (x86_64)", I have a rather strange behaviour: if I keep my AC power cord pluged in all cores stay fixed at 1.2Ghz (minimum speed), If I remove the power cord they "sometimes" (meaning that the behaviour does not seem "deterministic") they jump to higher speed values (but never to the maximum possible speed).

this is what I see using Turbostat with power cord IN (while running make on a large program)

 cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
            11.54 1.20 2.66 43.67 9.83 34.97 6.01 11.02
   0 0 5.83 1.21 2.66 56.73 5.15 32.30 6.01 11.02
   0 2 30.01 1.20 2.66 32.54 5.15 32.30 6.01 11.02
   2 1 5.75 1.21 2.66 42.11 14.51 37.63 6.01 11.02
   2 3 4.56 1.22 2.66 43.30 14.51 37.63 6.01 11.02
 cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
            24.05 1.20 2.66 48.51 8.22 19.22 0.00 0.00
   0 0 6.66 1.20 2.66 70.43 5.25 17.66 0.00 0.00
   0 2 46.43 1.20 2.66 30.66 5.25 17.66 0.00 0.00
   2 1 21.50 1.20 2.66 46.53 11.20 20.78 0.00 0.00
   2 3 21.62 1.20 2.66 46.40 11.20 20.78 0.00 0.00
 cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
            23.71 1.20 2.66 46.22 8.76 21.31 0.00 0.00
   0 0 7.08 1.20 2.66 50.27 12.23 30.42 0.00 0.00
   0 2 26.57 1.20 2.66 30.78 12.23 30.42 0.00 0.00
   2 1 35.55 1.20 2.66 46.96 5.29 12.20 0.00 0.00
   2 3 25.63 1.20 2.66 56.88 5.29 12.20 0.00 0.00

and this is what I get with power cord out
 cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
            18.82 2.10 2.66 47.81 7.21 26.16 4.90 6.22
   0 0 26.06 2.10 2.66 43.97 8.18 21.79 4.90 6.22
   0 2 28.85 2.10 2.66 41.18 8.18 21.79 4.90 6.22
   2 1 8.88 2.11 2.66 54.34 6.24 30.53 4.90 6.22
   2 3 11.50 2.12 2.66 51.72 6.24 30.53 4.90 6.22
 cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
            27.29 2.13 2.66 50.30 3.60 18.81 0.00 0.00
   0 0 16.80 2.13 2.66 83.20 0.00 0.00 0.00 0.00
   0 2 86.23 2.13 2.66 13.77 0.00 0.00 0.00 0.00
   2 1 3.43 2.13 2.66 51.74 7.20 37.62 0.00 0.00
   2 3 2.70 2.13 2.66 52.48 7.20 37.62 0.00 0.00
 cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
            19.92 2.12 2.66 49.84 4.25 25.98 0.39 5.11
   0 0 68.50 2.13 2.66 20.93 0.96 9.61 0.39 5.11
   0 2 4.75 2.09 2.66 84.68 0.96 9.61 0.39 5.11
   2 1 3.35 2.10 2.66 46.74 7.54 42.36 0.39 5.11
   2 3 3.10 2.10 2.66 47.00 7.54 42.36 0.39 5.11

It seems to me that there are still some issues with I7 mobile processors (btw, I also have a NON mobile i7 920 at home that does not seem to give me any problem with standard Ubuntu 9.10) ....

The bug subject for this is too generic, the issue was fixed in most cases with the acpi patch.

You should open a seperate bug for your problem.

thegreasi, thank you for reporting this and helping make Ubuntu better. Karmic reached EOL on April 30, 2011.
Please see this document for currently supported Ubuntu releases:
https://wiki.ubuntu.com/Releases

We were wondering if this is still an issue in a supported release? If so, could you please test for this with the latest development release of Ubuntu? ISO CD images are available from http://cdimage.ubuntu.com/releases/ .

If it remains an issue, could you please run the following command in the development release from a Terminal (Applications->Accessories->Terminal), as it will automatically gather and attach updated debug information to this report:

apport-collect -p linux <replace-with-bug-number>

Also, could you please test the latest upstream kernel available following https://wiki.ubuntu.com/KernelMainlineBuilds ? It will allow additional upstream developers to examine the issue. Please do not test the kernel in the daily folder, but the one all the way at the bottom. Once you've tested the upstream kernel, please comment on which kernel version specifically you tested and remove the tag:
needs-upstream-testing

This can be done by clicking on the yellow pencil icon next to the tag located at the bottom of the bug description and deleting the text:
needs-upstream-testing

If this bug is fixed in the mainline kernel, please add the following tags:
kernel-fixed-upstream
kernel-fixed-upstream-VERSION-NUMBER

where VERSION-NUMBER is the version number of the kernel you tested.

If the mainline kernel does not fix this bug, please add the following tags:
kernel-bug-exists-upstream
kernel-bug-exists-upstream-VERSION-NUMBER

where VERSION-NUMBER is the version number of the kernel you tested.

If you are unable to test the mainline kernel, please comment as to why specifically you were unable to test it and add the following tags:
kernel-unable-to-test-upstream
kernel-unable-to-test-upstream-VERSION-NUMBER

where VERSION-NUMBER is the version number of the kernel you tested.

Please let us know your results. Thank you for your understanding.

Helpful Bug Reporting Links:
https://help.ubuntu.com/community/ReportingBugs#Bug_Reporting_Etiquette
https://help.ubuntu.com/community/ReportingBugs#A3._Make_sure_the_bug_hasn.27t_already_been_reported
https://help.ubuntu.com/community/ReportingBugs#Adding_Apport_Debug_Information_to_an_Existing_Launchpad_Bug
https://help.ubuntu.com/community/ReportingBugs#Adding_Additional_Attachments_to_an_Existing_Launchpad_Bug

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

I think that, according to phoronix, this is fixed even before 12.04 (which i currently use.)
If anyone can also confirm that, please change to fixed, as i`m not quite sure.

As noted earlier the original bug was confirmed fixed in 2.6.33 upstream. And the only activity here seems to be in kernel 2.6, so I doubt anyone is still running into this issue.

thgreasi, glad to hear it is fixed for you in Precise. Did you need a backport to a prior release, or may we close this as Status Invalid?

Actually I got then next version of ubuntu right 1 month after I submitted this bug. I don't know if a backport of the patch ever got released. Just close this bug with any type of status you think is appropriate (I can't decide), since as noted in the comments, the version of ubuntu this bug was reported for is no longer even supported.

thgreasi, well the Status it will be toggled to is dependent on your needs. If you need a backport to a supported release prior to Precise (ex. Oneiric, Natty, Lucid, and Hardy Server), could you please explicitly mention this? Otherwise, this report may be closed as Status Invalid.