Page 3 of 4 FirstFirst 1234 LastLast
Results 21 to 30 of 33

Thread: Very slow SDD/HDD performance

  1. #21
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    Just a new note:
    Now the problem has occured again. Currently I have 4 days uptime with no suspend, but a few hibernates. Up until tonight things were working as they should, but not now after the latest resume from hibernate. So it's not only suspend-to-RAM that causes it.

    Next thing I should try is probably to disable Optimus. That could probably be causing all kinds of weirdness.

  2. #22
    Join Date
    Jan 2007
    Beans
    3,202

    Re: Very slow SDD/HDD performance

    Once in four days is painfully random -- makes it forever to troubleshoot. But yes, try it with Optimus disabled -- that can't hurt and might help. If you didn't already check, it would be worth looking at dmesg and the syslog around the timeframe that it slowed down.
    Intel Core i7-950 / Asus P6X58D-E / Nvidia GTX480 / siduction 64-bit on OCZ Revodrive SSD / KDE4.10.2/ Kubuntu 13.04

  3. #23
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    I tried disabling optimus, but then I couldn't get full resolution on my screen. I didn't want to struggle more with that, so I reenabled it (together with bumblebee and bumblebee-nvidia).

    Anyway, the symptoms just occured again, and dmesg shows nothing. (The last entry is 8000 seconds ago, when I inserted a USB keyboard).

    The only entries in syslog over the past two hours are a few cron jobs that don't look suspicious.

    There's a new kern.log started just recently (at 23:02), and kern.log.1 ends a few hours ago (at 20:49). I first noticed the symptoms about half an hour ago. I don't see anything obviously wrong in the kernel logs, but that gap is a bit suprising to me.

  4. #24
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    At the moment I'm trying to run with acpi=off as a boot option. I've been seeing several ACPI-related error messages in the logs, so I'm starting to suspect it has something to do with that.

    So far so good (but with only a couple of hours uptime). Of course, suspend to RAM no longer works, but hibernate does. If I can run this way for a few days, than I'll start to selectively disable some ACPI settings.

    One unexpected thing that also doesn't work with acpi=off is the touchpad and the mid-keyboard joystick. However, attaching a USB mouse works fine.

  5. #25
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    Here are my latest findings.

    I have tried a few ACPI-related boot options.
    • With acpi=off, I can't suspend at all, the fan is constantly storming, and I can't use the touchpad. Other than that, I didn't notice the slowness problem. Then again, since I can't suspend, it's not so strange. And the fan noise was a bit too much to bear for testing several days in a row.
    • With pci=noacpi, I can suspend but can't use the laptop keyboard; only an external USB keyboard. But after suspending, the slowness problem comes back.
    • With acpi=noirq, I also have no laptop keyboard
    • With pnpacpi=off, I can use the keyboard and touchpad, but the problem comes back after suspend
    • With noapic, it's the same; and that's what I currently run


    Something that I suspect is related to my problem is the following output from dmesg. After a fresh boot (with the noapic boot option), I have the following:
    Code:
     
    > dmesg |grep -i acpi | grep -i error
    [    0.526161]  pci0000:00: ACPI _OSC request failed (AE_ERROR), returned control mask: 0x1d
    
    
    > dmesg |grep -i acpi | grep -i warning
    [    8.815574] ACPI Warning: 0x00000460-0x0000047f SystemIO conflicts with Region \PMIO 1 (20120320/utaddress-251)
    [    8.815586] ACPI Warning: 0x00000428-0x0000042f SystemIO conflicts with Region \PMIO 1 (20120320/utaddress-251)
    [    8.815592] ACPI Warning: 0x00000500-0x0000053f SystemIO conflicts with Region \GPIO 1 (20120320/utaddress-251)
    [    8.815596] ACPI Warning: 0x00000500-0x0000053f SystemIO conflicts with Region \_SB_.PCI0.PEG0.VID_.GPSP 2 (20120320/utaddress-251)
    Not sure if that's alarming or not. However, when I do the same after a suspend/resume cycle, I get the following
    Code:
    > dmesg |grep -i acpi | grep -i error
    [    0.526161]  pci0000:00: ACPI _OSC request failed (AE_ERROR), returned control mask: 0x1d
    [  311.459051] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR1] (Node f742fd50), AE_TIME (20120320/psparse-536)
    [  311.459069] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR2] (Node f742fd68), AE_TIME (20120320/psparse-536)
    [  311.459077] ACPI Error: Method parse/execution failed [\ECRW] (Node f742fe40), AE_TIME (20120320/psparse-536)
    [  311.459086] ACPI Error: Method parse/execution failed [\ECG1] (Node f742fe70), AE_TIME (20120320/psparse-536)
    [  311.459094] ACPI Error: Method parse/execution failed [\NEVT] (Node f74300f0), AE_TIME (20120320/psparse-536)
    [  311.459102] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV._Q66] (Node f742fd38), AE_TIME (20120320/psparse-536)
    [  312.979329] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR1] (Node f742fd50), AE_TIME (20120320/psparse-536)
    [  312.979349] ACPI Error: Method parse/execution failed [\ECBT] (Node f742fdf8), AE_TIME (20120320/psparse-536)
    [  312.979369] ACPI Error: Method parse/execution failed [\ECG2] (Node f742fea0), AE_TIME (20120320/psparse-536)
    [  312.979378] ACPI Error: Method parse/execution failed [\ECG6] (Node f742ff48), AE_TIME (20120320/psparse-536)
    [  312.979439] ACPI Error: Method parse/execution failed [\_SB_.BAT0._BST] (Node f7430408), AE_TIME (20120320/psparse-536)
    [  313.486137] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR1] (Node f742fd50), AE_TIME (20120320/psparse-536)
    [  313.486155] ACPI Error: Method parse/execution failed [\ECBT] (Node f742fdf8), AE_TIME (20120320/psparse-536)
    [  313.486166] ACPI Error: Method parse/execution failed [\ECG2] (Node f742fea0), AE_TIME (20120320/psparse-536)
    [  313.486175] ACPI Error: Method parse/execution failed [\ECG6] (Node f742ff48), AE_TIME (20120320/psparse-536)
    [  313.486185] ACPI Error: Method parse/execution failed [\_SB_.BAT0._BST] (Node f7430408), AE_TIME (20120320/psparse-536)
    So a bunch of new ACPI errors.

    Googling hasn't given me much input on these. Maybe my best bet is to wait for the next kernel update in Kubuntu 13.04, and hope that any funky ACPI features of my Dell BIOS has been incorporated in the kernel by then?

  6. #26
    Join Date
    Jan 2007
    Beans
    3,202

    Re: Very slow SDD/HDD performance

    The error reported by dmesg is not a good thing, and may in fact be relevant to the problematic behavior. AFAIK, the warnings are not significant -- I see the same things on my desktop, which happens to be an Asus P6X58D-e board:

    Code:
    don@imerabox:~$ dmesg | grep -i acpi | grep -i error
    don@imerabox:~$ dmesg |grep -i acpi | grep -i warning
    [   11.447063] ACPI Warning: 0x0000000000001000-0x000000000000101f SystemIO conflicts with Region \SMRG 1 (20121018/utaddress-251)
    [   11.448388] ACPI Warning: 0x0000000000000530-0x000000000000053f SystemIO conflicts with Region \GPS0 1 (20121018/utaddress-251)
    [   11.448393] ACPI Warning: 0x0000000000000500-0x000000000000052f SystemIO conflicts with Region \GPS0 1 (20121018/utaddress-251)
    Here is a relevant post from a couple years ago: http://ubuntuforums.org/showthread.php?t=1882050&page=2
    Intel Core i7-950 / Asus P6X58D-E / Nvidia GTX480 / siduction 64-bit on OCZ Revodrive SSD / KDE4.10.2/ Kubuntu 13.04

  7. #27
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    Thanks for the info and the link, dibl!

    I'm currently running with the "nolapic" boot option, and this setting seems to be good. I have done three suspends and have a total uptime of 10:58, and things are still working as they should.

    The ACPI error messages in dmesg are similar to what I had when running with "noapic", but there are a some differences.

    Code:
    > dmesg |grep -i acpi | grep -i error
    [    0.073397]  pci0000:00: ACPI _OSC request failed (AE_ERROR), returned control mask: 0x1d
    [   11.325322] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR1] (Node f742fd50), AE_TIME (20120320/psparse-536)
    [   11.325328] ACPI Error: Method parse/execution failed [\ECRB] (Node f742fe28), AE_TIME (20120320/psparse-536)
    [   11.325331] ACPI Error: Method parse/execution failed [\ECG5] (Node f742ff00), AE_TIME (20120320/psparse-536)
    [   11.325334] ACPI Error: Method parse/execution failed [\_SB_.AC__._PSR] (Node f7430330), AE_TIME (20120320/psparse-536)
    [   11.325340] ACPI Exception: AE_TIME, Error reading AC Adapter state (20120320/ac-118)
    [ 1067.618169] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR1] (Node f742fd50), AE_TIME (20120320/psparse-536)
    [ 1067.618176] ACPI Error: Method parse/execution failed [\ECBT] (Node f742fdf8), AE_TIME (20120320/psparse-536)
    [ 1067.618179] ACPI Error: Method parse/execution failed [\ECG2] (Node f742fea0), AE_TIME (20120320/psparse-536)
    [ 1067.618182] ACPI Error: Method parse/execution failed [\ECG6] (Node f742ff48), AE_TIME (20120320/psparse-536)
    [ 1067.618185] ACPI Error: Method parse/execution failed [\_SB_.BAT0._BST] (Node f7430408), AE_TIME (20120320/psparse-536)
    [ 1068.452021] ACPI Error: Method parse/execution failed [\_SB_.PCI0.LPCB.ECDV.ECR1] (Node f742fd50), AE_TIME (20120320/psparse-536)
    [ 1068.452028] ACPI Error: Method parse/execution failed [\ECBT] (Node f742fdf8), AE_TIME (20120320/psparse-536)
    [ 1068.452031] ACPI Error: Method parse/execution failed [\ECG2] (Node f742fea0), AE_TIME (20120320/psparse-536)
    [ 1068.452046] ACPI Error: Method parse/execution failed [\ECG6] (Node f742ff48), AE_TIME (20120320/psparse-536)
    [ 1068.452049] ACPI Error: Method parse/execution failed [\_SB_.BAT0._BST] (Node f7430408), AE_TIME (20120320/psparse-536)
    So the first one is still there (ACPI _OSC request failed), but the system has been running fine with that error message before too.

    A few new ones have appeared, at 11.3 seconds.

    At suspend/resume there are now five ACPI errors (Method parse/execution failed), whereas with "noapic" there were ten. The ones that do not appear with "nolapic" are [\_SB_.PCI0.LPCB.ECDV.ECR2], [\ECRW], [\ECG1], and [\NEVT].

  8. #28
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    I found this thread, which describes a problem with the very same error messages that appeared with "noapic" but not with "nolapic". In that case, it had to do with an AC adapter, and apparently the problem was fixed in kernel 2.6.37.

    I tried unplugging and replugging my AC adapter, and there are no new errors in dmesg. However, the system is a bit slower now than before. Kwin is generally responsive, but apt-get update takes around two minutes. That's not too bad, and I can certainly live with it; but it is definitely slower than the 10-20 seconds it took before removing the AC adapter, and still very much faster than the 30 minutes it has taken before (and then kwin was lagging also).

    So I should be monitoring what happens with and without the AC adapter, and not just what happens at suspend and resume.

  9. #29
    Join Date
    Jan 2006
    Location
    Örebro, Sweden
    Beans
    491

    Re: Very slow SDD/HDD performance

    Now I booted with nolapic, and the AC adapter unplugged. Things going OK. Then I plugged in the AC, and now "apt-get update" is slow again. Go figure.

    There are no "parse/execution failed" ACPI error messages (they would only occur when I suspend). I don't see anything obviously wrong in the logs, but I'll paste tail of kern.log here anyway.

    Code:
    Mar 21 11:50:17 anis kernel: [  513.852178] bbswitch: enabling discrete graphics
    Mar 21 11:50:18 anis kernel: [  514.566259] pci 0000:01:00.0: power state changed by ACPI to D0
    Mar 21 11:50:18 anis kernel: [  514.566273] pci 0000:01:00.0: power state changed by ACPI to D0
    Mar 21 11:50:18 anis kernel: [  514.566341] pci 0000:01:00.0: power state changed by ACPI to D0
    Mar 21 11:50:18 anis kernel: [  514.566347] pci 0000:01:00.0: power state changed by ACPI to D0
    Mar 21 11:50:18 anis kernel: [  514.566354] pci 0000:01:00.0: enabling device (0004 -> 0007)
    Mar 21 11:50:18 anis kernel: [  514.620093] nvidia: module license 'NVIDIA' taints kernel.
    Mar 21 11:50:18 anis kernel: [  514.620095] Disabling lock debugging due to kernel taint
    Mar 21 11:50:18 anis kernel: [  514.627352] nvidia 0000:01:00.0: power state changed by ACPI to D0
    Mar 21 11:50:18 anis kernel: [  514.627356] nvidia 0000:01:00.0: power state changed by ACPI to D0
    Mar 21 11:50:18 anis kernel: [  514.627359] nvidia 0000:01:00.0: enabling device (0004 -> 0007)
    Mar 21 11:50:18 anis kernel: [  514.627370] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=none
    Mar 21 11:50:18 anis kernel: [  514.627714] NVRM: loading NVIDIA UNIX x86 Kernel Module  304.51  Tue Sep 18 17:36:24 PDT 2012
    Mar 21 11:50:19 anis kernel: [  515.815378] NVRM: GPU at 0000:01:00: GPU-21148771-abf9-5d8e-f9a3-6997c8be31c3
    Mar 21 11:50:23 anis kernel: [  519.781681] bbswitch: disabling discrete graphics
    Mar 21 11:50:23 anis kernel: [  519.797232] pci_raw_set_power_state: 46 callbacks suppressed
    Mar 21 11:50:23 anis kernel: [  519.797235] pci 0000:01:00.0: Refused to change power state, currently in D0
    Mar 21 11:50:23 anis kernel: [  520.097342] pci 0000:01:00.0: power state changed by ACPI to D3
    Mar 21 12:02:54 anis kernel: [ 1268.613609] cfg80211: Found new beacon on frequency: 5220 MHz (Ch 44) on phy0
    Mar 21 12:04:53 anis kernel: [ 1387.926244] cfg80211: Found new beacon on frequency: 5180 MHz (Ch 36) on phy0
    The pci_raw_set_power_state message seems to lead in the direction of bumblebee. So maybe my next step would be to try to disable bumblebee/Optimus again.

  10. #30
    Join Date
    Jan 2007
    Beans
    3,202

    Re: Very slow SDD/HDD performance

    I can't see how using the AC adapter would be relevant to a software operation like apt-get, unless somehow the wifi chip's performance is being compromised. Doesn't make much sense to me. Might be a design bug in the hardware.
    Intel Core i7-950 / Asus P6X58D-E / Nvidia GTX480 / siduction 64-bit on OCZ Revodrive SSD / KDE4.10.2/ Kubuntu 13.04

Page 3 of 4 FirstFirst 1234 LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •