Vista and 7.3 Display driver atikmdag stopped responding error

Discussion in 'AMD Graphics Cards' started by the-bullet, Mar 29, 2007.

  1. axmt

    axmt New Member

    Joined:
    Mar 22, 2008
    Messages:
    37
    Likes Received:
    0
    Trophy Points:
    0
    I find a couple of comments rather interesting as of late they are
    And the other(s) are the ones that state Running FAH have all but eliminated the Errors for folks using it ( to the right Percentage )

    The First as I had Stated many posts back had benn explained to me (by a Microsoft Tech ) as having direct links to how VIDEO hardware was handled by VISTA,, The Second Keeps the VIDEO HARDWARE --RUNNING...

    I had Hypothosed that the Problem May ACTUALLY been due to an Unforseeable and Unknown HARDWARE FLAW with The VIDEO CARDS..

    Now Disabling the Two things in Vista that the Guy I quoted first , according to what MICROSOFT told me Prevent Windows from SHUTTING down the GPU to save power,,,
    THIS IS INTERESTING..

    The SECOND Running FAH at FULL KEEPS THE GPU ENGAGED consistantly,,,
    ALSO VERY INTERESTING..

    The SAME 2 things seem to be working for NVIDIA users that are afflicted with the NVK error ...

    It seems to suggest that there is a COMMON point between ATI and NVIDIA that RELATES to the VIDEO HARDWARE.. I am not ASSUMING anything here,, but consider this,, the most basic engineering principals equate the same. There are BASE calculations and Requirements that BOTH manufacturers have to consider and at that level the TECHNOLOGY is not much different..

    BTW-- I recieved my RMA HD 3870 DIAMOND Card about a week ago and there was a out front Difference,,, the card was A DIFFERENT REVISION ( NEWER ) than ANY OF the others I Purchased. So far I have not seen a repeat of the Errors I had been seeing Consistantly ( Knock On Wood )...

    Catalystcatcher, I am of the thought that this is APPEARING to be more of a harware issue than a Software one. I think that this is a worthy thought to explore.
    How would I explain the OLDER ATI cars that are seeing the Issue ?? My explanation would be that they are possibly nearing EOSL ( end of sevacible Life ) or that they just can not handle the new power requirements efficently enough or can not endure the Stress that VISTA Puts them though thudly Damaging the card..
    As for Newer Cards ?? A Most likely SMALL unforseen complication in the Engineering ( pre- manufacture ) that translated into hardware based software glitches..

    Simalar things have happened in the past, and as in the past ther are folks who NEVER were affected.

    Just a thought...
     
  2. E-D@N

    E-D@N New Member

    Joined:
    May 2, 2008
    Messages:
    72
    Likes Received:
    0
    Trophy Points:
    0
    First post :)

    hay guys - i've reported my condition at the TDR error comm.

    just to let you know - i've found the solution for my Sapphire HD 2600 PRO AGP 512MB.

    i also had all these problems of yours - BSOD's, gray scrren and hard freeze, causing me to do hard resets many times...

    me and my fellows at Sapphire forums tried many stuff to make it work... right now i'm having a good time with the card - using the latest 8.4 ATI (Hotfix for AGP) drivers..

    and using RivaTuner - what i did is just "Underclocking" the card...

    first i've removed all my previous drivers using Drivercleaner...
    then installed the drivers + CCC...
    then installed RivaTuner...
    In RivaTuner - went to low level system settings... to the overclocking tab.
    there - enabled overclocking - it autodetected my card default clock... and gave my a safty range painted in blue... then i lowerd the Core clock to 540MHZ (For my HD 2600) and i left the memory at 500MHZ (or 495MHZ).
    then went to Fan tab at RivaTuner... enabled it also - and checked the Automatic feature...
    (saved all the RivaTuner settings and autoloading them windows starts... or restarts)

    that's it - since i've done that... not a single crush, almost finished Assassin's Creed, and finished Orochi Warriors...
    (working 24/7 for a week and 2 days)

    Good luck guys - hope it'll help someone :)

    BTW - here is a bit of my rig's spec:

    HP XW4100 Workstation... Intel Chipset 875P
    Sapphire HD 2600 PRO AGP 512MB
    PSU 280W Hippro
    1.5 GB RAM
    Pentium IV 3.2 GHZ HT CPU
    Vista 32bit Ultimate SP1.
     
    Last edited: May 2, 2008
  3. Rayline TWB

    Rayline TWB New Member

    Joined:
    Apr 15, 2008
    Messages:
    51
    Likes Received:
    0
    Trophy Points:
    0
    Hmmmmm... this may indeed be correct for some cases, but not in my case since both Adaptive Display and PCIe Link State Power Management has been off (disabled) since day one. This obviously because I sat up a high performance power plan in Vista when I configured it to run for the very first times.
     
  4. SilverDev

    SilverDev New Member

    Joined:
    Mar 25, 2008
    Messages:
    10
    Likes Received:
    0
    Trophy Points:
    0
    Still no crashes

    I saw axmt's post and I can see what you're saying about a potential underlying engineering problem. While that could be the problem, I still think the driver should be able to deal with it.

    Vista might be passing bad data, but even if it's a Vista problem, then it's up to ATI to test any input parameters for boundary conditions and validity.

    It's also up to their driver code to maintain state information related to the hardware. Correctly coded, it shouldn't be possible for Vista to cause the problem.

    At least CatalystCatcher mentioned they're working on getting some real error information to report out.

    Anyway, I'm running FAH at 50%. This was a purely random choice, that seems to be working for me. I haven't tested at points either above or below that percentage, and since its working and not causing performance problems, I probably won't bother to test other percentages.

    Glad to hear that more people are finding ways to work around the problem. Disabling Aero, by the way, has been a solution that's been around for a little while.

    It's, again, about stabilizing the load on the card, in my opinion. Either constantly high or constantly low. Disabling Aero, you're erring towards the constantly low, but then when you go into a game, the card kicks up for gaming and stays up while you're gaming, then settles back when you stop. Not that I would recommend this, but alt-tabbing in and out of your game to the Aero-disabled Vista interface would probably cause the error by telling the card to go to low-state, then back up to high-state and back and forth. Whether that's related to the driver making choices on fan speed or something else is really beside the point for those of us trying to live with the problem.

    Just 2 more cents for the pile. Yee-hah.

    Again, kudos to mrps2man for the database!

    Peace!

    SilverDev
    -=-
     
  5. CATALYSTCATCHER

    CATALYSTCATCHER ATI Guru

    Joined:
    Jul 30, 2003
    Messages:
    0
    Likes Received:
    3
    Trophy Points:
    0
    I'll add 2 cents. :)

    I would like to point out that if we did bounds checking on every bit of data being passed then we would not have much of an accelerator for graphics. Nor would any driver that spent most of the time validating data. The way things generally go for performance is to expect valid data and go with it until you know which routine is getting bad data. The fix to that kind of very predictable problem, (usually app/sw bug) is to do bounds checking to trap the bad data and hence avoid the inevitable crash.

    However, when a hardware issue is creating the bad data in a very random manner you never know which data to check. So a lot of bouds checking would need to happen in the driver and it would be really slow. The amount of asic command data not to mention all the 3D stuff being passed is quite staggering when you stop to think about it.

    Anyway, that is why hardware generated crashes are random and hard to unravel but SW bugs are usually easy to replicate and therefore a lot easier to fix.

    Also consider that hardware will usually have some mechanism to ensure data is not corrupt. Think parity or ECC memory. But usually only a server platform will use ECC memory and most user PCs use the less robust parity scheme. And we are talking system memory here. This is why AGP can be so awkward there is not even a parity checking scheme on the AGP bus. So things get problematic when the AGP data is corrupt and it is addressing areas of system memory. But these schemes work along side the data and do not slow things down.

    BTW: My TDR experience is that I get an infinite loop of TDRs when I first log into a system that has simply been left on overnight. (Almost 9000 event logged TDRs last weekend. )
    Originally the screen saver was running, so I shut it off last week. NO CHANGE!

    FAH GPU was running, so I shut it off, NO CHANGE!

    This morning with task manager running I noticed that DWM was taking 40-50% of the CPU and 70-80 MEG of memory. I don't know why, but that is way more than during normal operation with a dozen windows open.

    So , yeah I am turning of AERO for a few days to effectively disable the DWM and hopefully avoid the TDRs at login.

    If it does I will likely use the TDR reg keys to lengthen the TDR time out in order to determine if the TDR is happening because something in the system is bottle necking and the drisplay driver simply ends up waiting too long...TDR.

    I should also note that I do occasionaly see a single TDR throughout the day but I have no idea what triggers them but they also seem to happen when IE is tied up with a slow web page.
     
    Last edited: May 2, 2008
  6. CBRacerX

    CBRacerX New Member

    Joined:
    Jan 23, 2008
    Messages:
    28
    Likes Received:
    0
    Trophy Points:
    0
    Well, I had the full on crash version of TDR (Seems that I am alway getting that now) this am. I checked the FAH CPU Useage Percent and it is set at 100%.

    I have had Aero off for the past month.

    <sigh>
     
    Last edited: May 2, 2008
  7. Rayline TWB

    Rayline TWB New Member

    Joined:
    Apr 15, 2008
    Messages:
    51
    Likes Received:
    0
    Trophy Points:
    0
    Well guys - it is sad that I have to say this, but after 2 days of gaming I must admit that the thing that looked so promising - turning Aero off and folding with FAH GPU - has turned out to be a short era of happiness.

    It looked so good during the first day, but now everything is as bad as it was earlier. TDRs in abundance......

    So much for hope and faith....:-(

    But I am certain that it must have something to do with the cards behaviour, because the TDRs always happen exactly simultanously with the card gearing down and up again. But I don't have a clue to what makes the card spin down and up.....beats me.
     
  8. Jamezila

    Jamezila New Member

    Joined:
    May 4, 2008
    Messages:
    1
    Likes Received:
    0
    Trophy Points:
    0
    Hi

    I just got an ASUS G2K with Vista 32 bit, 2 GB of RAM, a Radeon HD 2600 and mod tooled Catalyst 8.4, and I get TDRs every few hours or so; pretty much once a day they'll end in a BSOD. I have an external display hooked up too.

    Anyway, I'm thinking about putting XP on it. It seems like everyone's problem occurred when they put Vista on their machine, but there was one guy who put XP back on and the problem remained. Has anyone had success with putting XP on?

    Thanks!
     
  9. reggie14

    reggie14 New Member

    Joined:
    Feb 20, 2008
    Messages:
    23
    Likes Received:
    0
    Trophy Points:
    0
    The TDR database that's being compiled has a remarkable number of dual monitor setups. Over half of the reports (which, admittedly there aren't very many yet) are using multiple monitors. That's far more than you'd ordinarily expect. I still think that's related to one of the causes of these TDRs.


    Jamezila,

    The TDR messages are unique to Vista. You simply won't get them in XP. If the video card/driver crashes in XP, the computer crashes. The TDR messages in Vista indicate the driver crashed but was restarted. So, to some extent, its nice that the drivers recover a bit more gracefully under Vista, but it stinks that crashes seem to happen a lot more.

    Out of curiosity though, you say once a day "they'll end in a BSOD". Do you just mean they won't recover, or are you actually shown a BSOD? It wouldn't make sense for it to show you a BSOD, since that would imply the graphics driver is working.
     
  10. daflory

    daflory New Member

    Joined:
    Mar 29, 2008
    Messages:
    2
    Likes Received:
    0
    Trophy Points:
    0
    I will sometimes get a TDR, which will actually recover and let me continue working. Every few days on average, however, I will get them in rapid succession, and then a BSOD.

    Interestingly, I only started getting the BSOD after I enabled the watchdog timer (WDT) in the bios. Before that I would get total video corruption and have to do a hard reset. Maybe the WDT lets the OS take control back from the unresponsive video driver.
     
  11. SilverDev

    SilverDev New Member

    Joined:
    Mar 25, 2008
    Messages:
    10
    Likes Received:
    0
    Trophy Points:
    0
    It seemed so promising

    Oh well, FAH seemed so promising.

    I got a crash this morning and had gotten a few TDRs yesterday, though they had not led to crashes.

    Thanks for that detailed reply Catalyst Catcher. You do make sense, but there have to be some boundary checks in there, granted it can't be on every call...

    Anyway, I'm pretty sure that the additional factor for me is heat. I've already got my machine sitting on one of those extra fan systems, but my office is warm. They've had trouble keeping the ambient temperature regulated. Plus, being a laptop, it runs extra-hot anyway. My two cores run around 60C.

    *sigh*

    My screen saver is just a blank screen and it crashed when going to the screen saver. (7 rapid TDRs, followed by BSOD) Again, a situation where the load on the card shifted, for what its worth. I do have dual monitors and Aero is running. If I start getting common errors again, it'll be time to try disabling Aero and dropping FAH.

    Argh. It's not like its important. Just my livelihood. :rolleyes:

    Peace!

    SilverDev
    -=-
     
  12. Biffes

    Biffes New Member

    Joined:
    Mar 31, 2008
    Messages:
    30
    Likes Received:
    0
    Trophy Points:
    0
    just a small update:

    Fah is running for quite some time now. No crashes or TDRs for me so far...

    @SilverDev are you sure about ->7<- TDRs in a row? my "normal" hard crash is 5 tdrs in a row + BSOD...
     
  13. CATALYSTCATCHER

    CATALYSTCATCHER ATI Guru

    Joined:
    Jul 30, 2003
    Messages:
    0
    Likes Received:
    3
    Trophy Points:
    0
    No doubt there are boundry checks but only at major points as determined by the developers based on SDK suggestions and risks. Also based on experience with certain applications and the risk to the overall integrity of operation. As well as when a bug investigation reveals that an app is passing bad data. This is the software side of bad data, but we can't forget that hardware also has the ability to create bad data. The best example is failing memory component.

    Heat, is certainly a major concern as it contirbutes to the overall flakeyness of the hardware. Higher clocks (OCing) generate more heat and usually require higher voltages which also generates more heat.

    That's why we always want to know if problems continue at normal setting.

    I agree that my TDRs tend to occur at some kind of transition in load. But in my experience it is returning from the screen saver or logging in after a long period of system inactivity.

    In my experience DWM seems to jump to 50% or more of CPU time and then I go into a continuous TDR loop. (Eventlog showing 8000 or more consecutive TDRs. I am glad it was the weekend and I was not here :)

    On the downside I suspect the load transition TDRs are caused by something different than the spontaneous single events I am also seeing. But maybe not as DWM does get very busy for some reason. In any case disabling AERO by simply setting 16 bit color essentially takes DWM out of the picture and seems to stop the TDRs. At least for me.

    So disabling AERO should provide temporary relief for some.
     
  14. daflory

    daflory New Member

    Joined:
    Mar 29, 2008
    Messages:
    2
    Likes Received:
    0
    Trophy Points:
    0
    Thank you for your updates Catalystcatcher.

    What I have found most interesting about my TDR's is that they do not appear to have any relation to heavy use of the computer.

    I can run 3Dmark, and make the video card fan wail like a banshee--with no video corruption and no errors. No problem with games. I can also do very heavy multitasking, with tons of windows open on both monitors, maxed out cores, virtual machines using up most of my memory, and have no problems at all.

    I most commonly get TDR's when I am simply working on a Word document, or browsing the web.

    For me, the errors seem associated with working on things that put very low power demands on the system.

    Is it possible that the power-saving modes of the video card are failing to supply enough voltage in "low power mode"?
     
  15. CATALYSTCATCHER

    CATALYSTCATCHER ATI Guru

    Joined:
    Jul 30, 2003
    Messages:
    0
    Likes Received:
    3
    Trophy Points:
    0
    That is also what I see on my system. Makes me think that perhaps many of the TDRs related to heavy duty use may well be triggered by heat or some other hardware issue. While the low usage events, are something else. I am currently waiting for a special display driver that is intended to capture more detail when a TDR occurs. At this point for all TDR event log entries my GPU is in the idle state and there aren't any other clues.

    CC
     
  16. Rayline TWB

    Rayline TWB New Member

    Joined:
    Apr 15, 2008
    Messages:
    51
    Likes Received:
    0
    Trophy Points:
    0
    Just my 2 cents about heavy load and/or triggering by heat:
    As I have stated in an earlier post, I did not have these TDRs using WXP SP2. Same card, same use, same applications, same games, same heat readings. But with Vista - TDRs in abundance.
     
  17. CATALYSTCATCHER

    CATALYSTCATCHER ATI Guru

    Joined:
    Jul 30, 2003
    Messages:
    0
    Likes Received:
    3
    Trophy Points:
    0
    :) I hear ya. I did say "may well be triggered" which is my way of acknowledging that not all of the heavy load events are due to heat. I just want viewers to understand that TDRs can, sometimes, in some instances, possibly be generated by heat.

    In my case, I also transitioned from XP to Vista on the same platform over a year ago. Actually the TDRs only started recenty and I was only ever able to create one in the past while OCing the GPU using .

    So I am still listening and gathering clues. Problem is there are no really good clues and the TDR message is of no help what so ever. Not to mention a pain in the a$$ :)
     
  18. xengren

    xengren New Member

    Joined:
    Mar 3, 2008
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    0
    Yes, XP solved it for me.

     
  19. Rayline TWB

    Rayline TWB New Member

    Joined:
    Apr 15, 2008
    Messages:
    51
    Likes Received:
    0
    Trophy Points:
    0
    Yeah I pretty much know that converting to XP again would make my errors go away, but I have upgraded my system and left XP for Vista, so I am not going to go back to that. That would render my upgrade somewhat useless.
    I have faith that some guru at ATI/Mircrosoft will eventually find a fix for this, so I keep my hopes high.

    @CatalystCatcher: Yeah I know what you mean, I just wanted to make sure that anybody could see that heatproblems might not be the answer.....in many cases ...:)

    I am 100% certain that this relates to the interaction between the drivers and Vista. I have looked at so many different forums over our little blue orb, and it looks like the common things are:

    1. It affects ATI card owners, nVidia card owners, S3 card owners and so on.
    2. The problems started almost at the same time - when more and more people migrated to Vista.
    3. The problems are pretty much similar - some problem with the correspondent "display driver (.sys) stopped responding, but succesfully recovered".

    Based on this together with all the reports of countermeasures tried, I find it sincerely hard to think that this has something to do with how people configure hardware/software - and all these problems cannot be related to common heat issues and so on. Of course in certain cases this may be true indeed, but I have never seen such a uniform complaint about the same thing in my whole computer life. It must be several thousand postings about this, and I think that all these posters cannot be noobs that do everything wrong with their rigs......:)
    I bet my Grandmothers undies on Vista/driver interaction problems....:)
     
    Last edited: May 8, 2008
  20. CATALYSTCATCHER

    CATALYSTCATCHER ATI Guru

    Joined:
    Jul 30, 2003
    Messages:
    0
    Likes Received:
    3
    Trophy Points:
    0
    :DI agree there is something not quite right here. Yes there is a common point we simply have not found it yet. Part of the problem is the TDR message. In XP you may see game crashes, spontaneous reboots or system hangs yet in Vista they all manifest as a TDR. This has a tendancy to collect a lot of people complaining about this error. (Evidence this thread and a couple of threads at NV) But it seems that this is something above and beyond that because generally speaking returning to XP cures most peoples problem. (Not everyone but certainly most)

    I have noticed this week that my TDRs are occuring mid afternoon about 3:00 so I am looking at scheduled tasks and network stuff as possible triggers. And lets not forget virus scanners and IT installed spyware which does many things on a regular basis.:D

    But so far only disabling AERO seems to cure my TDRS and I know that isn't the case for everyone.
     

Share This Page

visited