*HP HDX 9000 DRAGON Owners Lounge, Part 2*

Discussion in 'HP' started by 2.0, Sep 3, 2010.

  1. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    :D Since you're right. My mistake, I try to explain it exactly. I wish that it can understand any step by step.

    Better would be the short version:

    1. Install new drivers (uninstall and delete the old first)
    2. "NVPMM" to "fixed performance level" for AC and Battery. (Never use performance level "MinPerf / MaxPowerSave", otherwise there is "HDX" crash)
    3. For fine adjustment use "Nvidia Inspector".

    ready

    There was always user to use the claimed new drivers without problems. I can not agree. All my tests, on all 6 HDX with different OS, confirm the instability. Also for 301.42.

    Have just tested again for 301.42. With automatic "GPU frequency" I can "HDX" bring within 1 minute crash. (my test method)

    Did you determine, do not you use "NVPMM".
    "Powermizer" even after your installation method would be more active.
    This also means, "301.42" will work for you with automatic GPU frequency adaptation.
    That I can not believe. If this were so, then we would have never had a problem with new drivers as 257.21.

    A driver extraction is not necessary. When you first install the new drivers must also "Nvidia GeForce 8800M GTS" be removed from the Device Manager.
    (in addition to the driver uninstall, and delete)

    Also, it makes sense to not be online. (Only with automatic Windows Hardware update)

    Still, I'm testing now your installation method. Exactly, even with automatic detection Nvidia driver.

    Result follows ....
     
  2. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    result:

    "HDX" crash

    This is not surprising, because your method will result in a normal installation. It happened otherwise.

    My new test method for driver stability is very simple.

    After installing and restarting I open GPU-Z for frequency control.
    Then I open an Explorer window (eg Computer drives Overview)

    I wait until the GPU works at 200MHz and then I resize and scale of this Explorer window.
    (At the upper edge hold and then pull up and down.)

    Here, the GPU frequency to 500MHz jumps. At that moment, "HDX" crash.

    If it does not crash immediately, then you have to repeat this procedure.
    Prior to the Explorer window size changes, the GPU must always operate at 200 MHz.

    DigitalWindow please report when your HDX crash. Even if you are not dealing with the "GPU Drivers" look together.

    Perhaps you have on your computer a program that always keeps the GPU at a higher level of performance.
    If this is not the case, then your GPU would work with 200MHz idle.
    And then there must also come to the crash.

    Please controlling this with GPU-Z!
     
  3. hovercraftdriver

    hovercraftdriver Notebook Deity

    Reputations:
    361
    Messages:
    785
    Likes Received:
    2
    Trophy Points:
    31
    That was a good write up Rikki...my question is this: Is there a valid reason that nVidia wouldn't have corrected this behavior? It obviously didn't occur on earlier drivers, i.e. 257.21 and earlier. Have you contacted nVidia with your findings(each driver version usually has a corresponding forum post on nVidia support site to flesh out bugs and problems)? Shouldn't it be possible for them to alter a setting in their driver packages to make said driver perform as before?

    I suggest contacting nVidia support as above and include this test method AND your "settings" findings/fix. Or with your permission, I will use your words and contact them myself. Probably won't get anywhere as our system is older and this issue doesn't affect all nVidia mobile systems, just a select few...but any chance is worth it. I have many times contacted them with my issue, but never with an identified problem. Maybe this would help change things...or maybe they are aware and simply choose not to address the issue because so few on the grand scale experience the problem.

    The reason I suggest this, is that despite your strong recommendation to not use new drivers, I and others have noticed significant performance increases of varying kinds, whether it be temp drops, FPS increase, or whatever. I can only assume you suggest not to use the newer drivers because it is difficult or simply a pain in the butt for the average user to set up profiles as you have outlined.

    All the more reason to keep after nVidia to implement a correct fix.

    I folowed your procedure(and others) to no avail...can't get a complete, successful install using the nVidia installer. I did take it apart and cleaned fans. Now sounds like a hang-glider as opposed to an F-18 :D I suspected that, but was curious how the newer games performed on someone's HDX besides mine.



    I still get the same old crashes after a successful back door install...until I apply the NVPMM fix. Drivers do perform very well for me...significant increase in FPS and noticeably smoother playing Diablo 3.
     
  4. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    Thank you :)

    I know of no reason.

    No, and I have no motivation. This is what I had done it in a letter in 2008. (4 full pages long) time it was about "Powermizer" and its peculiar switching behavior. In addition to the high latencies when jumping from 200MHz to 500MHz.
    I never got an answer. Even a phone call with a Nvidia manager and his promise to answer were unsuccessful. :mad:

    Of course. It went so well the other way around. We not only know the true cause. For every problem there is always a solution.

    You can do this gladly. Your English is certainly better. ;)
    For me personally, this is not important. I can use any driver without problems. I turn "Powermizer" always "off", even if it is a driver would not be necessary. (257.21 or 301.42 "PM" will always be switched off)

    The reason: See my test method! This type of GPU frequency increase is not reasonable. From 200MHz to 500MHz and step down in a circle I feel as feebleminded since 2008.

    This would, despite Nvidia's new driver fix, not an option for me. This would result in the best case with automatic clocking no crash.

    Exactly, this is the only reason. I highly recommend the new driver at all times. But it is important to the functioning of his "HDX" and the "GPU" is to be understood.
    Otherwise there will be problems.

    Someone who does not understand this has the following problems:
    1) Without the use of "NVPMM" with attitude "MedPerf / MedPowersave" or "MaxPerf / MinPowersave" there is some point to the "HDX crash".

    2) Who "NVPMM" only with the setting MedPerf / MedPowersave uses, which lacks the "GPU Power" to play. Frequency is only 275MHz yes.
    Ranches for performance must be "NVPMM" temporarily set to "MaxPerf / MinPowersave".

    3) Who "NVPMM" only with the setting MaxPerf / MinPowersave uses, which has a very high temperature, GPU iddle and loud fans. Frequency is always at 500MHz.

    4) The manual control with "Nvidia Inspector" (option is) aggravating circumstance.

    All these points do not allow a general recommendation for drivers over 257.21.

    Anyone who understands these issues, however, can use the new drivers with no problems. I'm using the latest drivers at any time.

    Another benefit as described above and explained to me since 2008:
    No interference from latency spikes. (neither audio nor video)

    These problems are now, by the arbitrary performance requirements of the programs, largely resolved.
    "Flash browser requests" from 720p = 383MHz. Similarly, the WMP.
    This time was different. It was then when playing the meaningless cycle frequency.
    Is still "Powermizer" always better off. The GPU then jumps to never directly 500MHz, but directly to, the particular program, the required value. (mostly on 383MHz)
    The Adobe Flash Player (browser add-on), however, has problems when closing the browser window. In this case, the GPU is usually at 383MHz (PerfLevel2) and will not reset. WMP makes it better. This resets always the frequency.

    For inexperienced users, in any case. :D


    Did you follow my advice sometimes. --> GPU delete from device manager. (in addition to the driver uninstall and deletion)

    I had this problem before. After deleting the GPU installation worked without problems then. Only the first time. After that, change drivers, as always.


    This is every "HDX" exactly as I have described. Digital Window is also no exception. If his iddle is at 200MHz in performance mode 0, then its "HDX" crash to 100%. (sooner or later)

    If the GPU iddle but held in a different performance level (through whatever), then explained it to his observation.
    This may not be a recommendation for this thread.

    Just as then "KindHeart" with its information, with F.41, all new drivers are working fine. All this is nonsense.

    However, I am curious what in "Digital Window" is different. Perhaps we know it soon.


    ---------------------------------------------

    rather hovercraftdriver, thank you for your contribution!

    Surely it makes virtually no one fun to read my letter maze.
    The more I am glad your dealing with my contributions.

    This goes for everyone else.

    @Digitalwindow

    I hope the translation makes no attack my words.
    That would not be in my mind. We only want to clarify all issues relating to the "HDX". And this is exactly as possible.

    @ all

    Please write if your driver is newer than "257.21", without the use of "NVPMM" and used without problems!

    Or write at all times here! Anyone who reads this and has never been written?
    Sign up and write! Now!

    EDIT:

    By the way ..... This contribution has not gone past me. So once again my link.
    "2.0" seems indeed to be absent.
    Here's some more come to light as He would have liked?
    Whether that seem to point to his whereabouts? :D
     
  5. hovercraftdriver

    hovercraftdriver Notebook Deity

    Reputations:
    361
    Messages:
    785
    Likes Received:
    2
    Trophy Points:
    31
    My responses in blue ;)
     
  6. Digitalwindow

    Digitalwindow Notebook Evangelist

    Reputations:
    139
    Messages:
    377
    Likes Received:
    1
    Trophy Points:
    31
    I don't know guys, still no problems.......except when trying to run 3DMark06. I get momentary blackscreen and auto-reboot about half way through the bench. It's possible I just don't use it enough to experience all the crashes I'm supposed to.

    I did over a solid hour with Sniper Elite V2, and let an HD video run for over two hours while I was away. No crashes. :confused:

    I may not have several pages of technical to convince of 'rep', but these are honest reports. I have mobo 5.92, 8800m 1.36 (I think :eek: ).

    :D :p

    Which reminds me that my gpu fan only makes an 'audio' appearance when running the heavy gpu apps, but RickiBerlin's explanation would support that a crash would happen no matter the temp activity.
     
  7. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    This is exactly confirm my prediction.

    The crash may never take place under load. It is only with the jump from "Performance Level 0" to a higher "performance level".

    You could run for 1 year a HD video can without crashing. (Note as long videos are very rare) :D
    But when the movie finishes and the GPU in "Performance Level 0" is returned, then the start of a new video cause the "HDX" crash.
    (need not, but may / likely about 1 to 5)

    This example in figures:

    HDX started. GPU is at 200MHz. You start the 1 year video. GPU frequency jumps with the help of "Powermizer" automatically at Performance Level 3 to 500MHz. (this time no crash)
    Now (15 seconds later), the frequency is governed by the "player program" required performance level. (for example, performance level 2 - 383MHz)

    The video is now 1 year with this performance level. The "HDX" will not crash.
    But alas, the video is over. Then, the frequency drops in performance level 0. The next "Jump Up" can now be the crash.



    3DMark06 has a number of these jumps. Tailored to each individual test. (I think there were 4)
    If you're lucky, then it goes through. It can, however, even after the first test, just the beginning of the next crash.

    Now to your game. In a game the GPU is also under constant load* . The only risk is the start of the game. If the successful launch, then you can play as long as you want.

    An interval of play (real time off without GPU load) is again a risk.


    * Load is always higher than performance level 0
    Of course there are fluctuations. This move, however, always remaining between the performance levels 1 to 3.
    Always between 275 MHz and 500 MHz. Never the way down to 200MHz.

    ---------------------------

    Your review has shown me yours "HDX" is like any other.

    And if you would use GPU-Z, then you would understand my explanation. Crash only by Performance Level 0 (with you 200MHz) to a higher level)

    My recommendation: Use "NVPMM" or a driver to 257.21.
     
  8. JimWid

    JimWid Newbie

    Reputations:
    0
    Messages:
    7
    Likes Received:
    0
    Trophy Points:
    5
    Bricked Dragon

    Can anyone provide any advice?
     
  9. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    @JimWid

    This is very unusual. Once the motherboard is receiving power and the power button is actuated, the fans go into operation.

    I have just been thinking about it. Unfortunately I have no idea. Sorry!

    EDIT:
    The main set, I had forgotten. Even if you had forgotten all cables (except fans). Have you removed the CPU?

    EDIT 2:

    To further explain:

    When the CPU is not in contact, then do not run the fan. (as I recall)
    I might check this morning for you.
    Of course, the "HDX" do not work then. Only the Quick Launch bar lights.
     
  10. JimWid

    JimWid Newbie

    Reputations:
    0
    Messages:
    7
    Likes Received:
    0
    Trophy Points:
    5
    RickiBerlin Thanks for your response.

    No I did not remove the processor and have now checked all connections several times.

    I have also looked the whole thing over for damage (the motherboard was pretty difficult to remove) and everything looks OK.

    It is interesting that all of the blue lights come on except for the LED on the power button itself.

    I did wear an antistatic bracelet but wonder if I could have zapped something by accident.

    Thanks

    EDIT - perhaps I will try to take the CPU out and replace to see if this helps. Shame the HDX is so difficult to pull apart!
     
Loading...

Share This Page