CPU usage way too high

Discussion in 'Windows Virtual Machine' started by bnguyen, Sep 4, 2012.

  1. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Try this sequence:

    - reboot host with NVRAM reset (alt-cmd-P-R)
    - start VM
    - check CPU - you should have CPU ok (after VM housekeeping)
    - shut off host (power off) the host
    - start host
    - start VM
    - check CPU - now you should have again the high CPU (quite immediately)

    I did this more than once and it happens the same every time
     
  2. DanielN26

    DanielN26 Bit poster

    Messages:
    7

    I have not tried it yet, but I did an SMC reset this morning before I booted my host and so far everything is good.
    Next time it occurs I will try the PRAM reset that you mentioned as well.
    But still I guess it is not ok that we have to reset SMC and/or PRAM each time we boot our host system. Parallels should fix that on their end.


    When I enter "sudo nvram -p" in the terminal what do I have to look out for or what do you see when the high CPU usage occurs?
     
  3. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Sorry DanielN26, I didn't explain well the situation.

    Obvously we can't boot our machine every time and resetting NVRAM, but if we find that the problem is common, this could help parallels support to solve the issue.

    Now if you see the output of nvram -p command, you see that after a NVRAM reset there is no "boot-gamma" string, but after shutting off the machine (I mean power off) this string comes up again.

    This means, I think, that we have a problem with the monitor driver of parallels and the OSX kernel.

    So if more than one user tries to check these steps and find that with NVRAM reset it works, otherwise the CPU goes up even if VM is doing nothing, this could be something to explain to parallels support and they have to investigate within this line
     
  4. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Hi Daniel,

    I found this and I agree with you that we can't reset NVRAM every time we start the machine, but if more than one customer find that this is the cause, this could help parallels Support.

    By the way if you che the "nvram -p" command you will find that after a NVRAM reset there is no more the "boot-gamma" string. If you boot again (from power off) you will find it again.

    So if there is no "boot-gamma" the VM works correctly, otherwise (in normal conditions) we have the huge CPU level.

    It's something about monitor resolution, that means tha we have a problem between Parallels monitor driver and OSX kernel
     
  5. DanielN26

    DanielN26 Bit poster

    Messages:
    7
    Sounds good, but I am not sure if that is really the problem. Hope that it is but I have entered the sude nvram -p command and get something like this:

    gpu-policy %01
    fmm-computer-name Daniel%e2%80%99s MacBook Pro
    boot-gamma %10%06%00%00%0e%a0%00%00%00%00%00%00n%00%00%00%00%00%00%00%07%00%04%04%db%05%0b%0b%b2%0e%1e%1e%95$66m>WWO`ww%09%80%a1%a1%87%a8%07%00%04%04%db%05%0b%0b%b2%0e%1e%1e%95$66m>WWO`ww%09%80%a1%a1%87%a8%07%00%04%04%db%05%0b%0b%b2%0e%1e%1e%95$66m>WWO`ww%09%80%a1%a1%87%a8
    SystemAudioVolume [
    backlight-level %ff%02


    According to your idea I should have huge spikes now, correct? But I don't. Since I booted my machine this morning with a SMC Reset before I booted everything is ok - no huge spikes or anything.
     
  6. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    I'm sorry. I did also a SMC reset but I have ever the same results:

    - doing NVRAM reset : ok
    - normal boot : high CPU (no spike, ever at 100%)
     
  7. gradkal

    gradkal Bit poster

    Messages:
    7
    I´m in the same boat here. Having always at least 15-20 % cpu-usage of Parallels 8 if Windows 7 (32-bit) is idle. With Parallels 7 it was at 5 %. What a pitty!

    Sended a problem report (#21623028) as the others above. Hope, that helps to find the bug. I´m with Parallels since Version 3, but I remember this was always a problem they did not fix from the start. Parallels, when do you learn out of your errors in the past?
     
  8. totvos

    totvos Member

    Messages:
    34
    I "ditto'd" this in another thread, but since there are several high CPU threads, I'll post here too. My MBP is continually running at >50% CPU for Parallels 8, while Windows 7 is using 0-1% according to Task Manager.

    Parallels Support, this is a REALLY BIG PROBLEM.

    Problem ID: 22530885
     
  9. Molecool

    Molecool Bit poster

    Messages:
    6
    Very same issue

    I've got a brand spanking new MBP 2.7GHz running Lion 10.8.2 equipped with 16GB of RAM. My temperature without Parallels and a ton of apps open is around 70 Celsius. As soon as I start Parallels 8 (without any Windows apps running) it climbs up into the 80s and 90s. Running any type of app pushes it over 100 Celsius in no time. This is particularly evident after suspending my VM and waking it up again. Heat issues right from the get-go.

    Parallels 8 has an issue that they are not addressing. I wonder if it may take a class action lawsuits by everyone who's hardware has been damaged to fix this once and for all, despite the fact that this problem has been known for a long time now.
     
  10. panzer07

    panzer07 Bit poster

    Messages:
    2
    Was this problem ever resolved? My system CPUs start heating up just starting parallels. In the VM I took and down to 1 CPU and turned off 3d altogether and temps are slowly dropping. Fan still going full blast. This is very annoying.

    Cheers,
     
  11. Molecool

    Molecool Bit poster

    Messages:
    6
    Resolved? I didn't even get a reply here. Next time I'll go with XEN.
     
  12. panzer07

    panzer07 Bit poster

    Messages:
    2
    Yeah, I'm getting that feeling. I may revert to VMware. Thanks
     
  13. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Unbelievable

    I agree with you!!! Last "resolution" from Parallels engineering is quite fun and borders on the ridiculous. Even if I show them that there is a problem somewhere in their code, probably around monitor drivers, they still ask to uninstall, reboot, reinstall and so on.

    And probably Release 8 has the same bug; if they don't find where is they can't fix it!!!!!

    I continue to use my work-around, the only one that permit me to work with VM. I'm testing VMware in the mean time.

    regards
     
  14. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Hi all,

    VMware Fusion 5 tested - same issue. This means that the problem is inside MAC.

    The problem resides in display management; resetting NVRAM there is no "boot-gamma" parameter.

    I'll proceed it this way, checking how to avoid this parameter to be setter
     
  15. noddy347

    noddy347 Bit poster

    Messages:
    1
    "If" the issue is with the MAC the responsibility still lies on Parallels shoulders in my opinion. They are selling a product which they have developed to work specifically with the MAC. I have purchased their product after reading their advertisements and this software is frying my new 3K Mac book. Basically if you are doing anything more intensive than opening email in windows the cpu goes into meltdown. This needs to be fixed and the consumers shouldn't be used as R&D guinea pigs.
     
  16. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Other tests

    Sorry Noddy347, but if I found the same problem with VMware, this should be something around Mac, very hard to find. I have a friend with same Mac (MB Pro 2010), same OSX, same config, but it works smootly. And boot-gamma parameter has nothing to do with it (when rebooting this parameter appears but Parallels works)

    I found another thing: if you force a NVRAM reset (cmd-alt-PR) everthing works; if you reboot the system (without shutting it off) nothing changes. I have again the problem if I shut off and start again from power-off. In some forums I've seen that someone complains about bad drivers that are loaded only when powering on.

    Pls search also for this kind of problem, may be we can find the solution! I'm quite sure that is something within drivers or OSX start-up procedures.

    The complain I can move to Parallels Engineering is that they didn't find why it happens, because doing a deep troubleshooting they should see where their software uses so much CPU doing nothing (a core dump of parallels app should be enough to find it)

    regards
     
    Last edited: Jul 3, 2013
  17. Niceguy

    Niceguy Bit poster

    Messages:
    7
    Does this CPU usage problem also exist with the newest MacBook Pro Retina, or only to other series?
     
  18. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Hi all,

    finally using the Parallels utility "trace_int" I found that I have a problem with the second HDD (mounted instead of DVD) that is using port SATA@B. This situation creates a interrupt storm with "com.apple.driver.AppleUSBOHCI" as follows:

    VECTOR CPU#0 CPU#1 TOTAL DEVNAME DRVNAME
    52 563886 0 563886 OHC2@6 com.apple.driver.AppleUSBOHCI
    SATA@B com.apple.driver.AppleAHCIPort

    I found more than one guy with same situation and same problem. This interrupt storm affect VM machine performances (Parallels, VMWare, OVM) in the same way.

    Rebooting the system with ctrl-CMD-P-R (reset NVRAM) the storm disappear. It seems there is no solution with this issue.
    Another way to do that is:

    - delete boot-gamma from NVRAM (nvram -d boot-gamma)
    - reboot (from console in command mode)

    Anyone has info about this issue?
     
  19. pborzenkov

    pborzenkov Junior Member

    Messages:
    19
    I assume you have MacBookPro 5,[123]. If this is true, it's most probably a bug in Apple drivers.
    We have seen this problem many times, but, unfortunately, failed to reproduce it locally.

    But it looks like a "race" between AHCI (SATA) and OHCI (USB) drivers, which share the same interrupt vector 52.
    Depending on which one loads first, you either will have interrupt storm, or will not.
     
  20. Peterwithes

    Peterwithes Junior Member

    Messages:
    14
    Thanks pborzenkov, I have a MBPro 5,1. By the way I can reproduce the issues as many time as I need and it seems not "random"; it happens only when started from shutoff.

    If it depends from module loading it should be random, but it's not so.

    I didn't find any notice about Apple drivers update.

    To reproduce it you should mount an extra HDD instead of the DVD; this should be enough.

    I hope you can help me to find a solution.

    thanks again
     

Share This Page