Vista and 7.3 Display driver atikmdag stopped responding error

Raduque

In my case, the problem is Aero. It's the only time i get TDRs. I've never gotten a TDR in a game or benchmark app.

eggmancs

The system is at stock speeds. All settings are on "auto" detect. Ive changed them to the manufactures settings were they do not match without changing the error crash. Ive updated the motherboard bios (1004 to 1304) which didnt change anything. Increased NB and memory voltages seperately and together - doesnt help.

All drivers are up-to-date including PCIE bus drivers. Tried all versions of ati drivers (8.5 to 8.9 and the new 8.10betas). Tried the drivers with and without CCC, with Rivatuner and with ATT. Tried disabling Aero, tried changing boot setting for memory and cpu with detect_HAL ticked and unticked in msconfig. Changed fan speeds, changed clock speeds on the gfx card including slight perminant o/c and u/c to take off PP profile switching. Taken off Catalyst AI and changed it to Advanced. Tried everything! Tried reinstalling vista multiple times. Vista has all available updates.

Somethings work in combination to lessen the problem and reduces crashes but they still happen frequently.

The only thing that takes away the problem is removal of one ram stick - which isnt a permanent solution.

Dont know what to do....

dawny

yups.. me too... i use x48 dfi lanparty ddr 3, ram corsair and vga card ATI HD4870 Crossfire. I try everything until i am stuck. And atikmdag stopped responding error still showing. I dont know what i must do again. This VGA card i buy just one week a goo...

goaty

Any room left in the "atikmdag error" club? I'd like to apply...

Anyway, posting the contents of my ticket sent to ATi, Gigabyte, Sapphire and a few other forums:

"My specs:

Gigabyte EP35-DS3R;
Intel Core 2 Duo E8400;
Sapphire Radeon HD 4870 512MB DDR5;
Transcend JetRam 2x2GB DDR2 800MHz;
CoolerMaster RealPower M700 52A@12v;
Vista Ultimate x64 fully updated;

I'm randomly getting "Display driver atikmdag stopped responding and has successfully recovered." when card is under stress. It always recovers, never experienced a BSOD or anything. Sometimes it takes 5 seconds, sometimes up to a half an hour.

Things I've tried:

- upping voltages in BIOS and tweaking memory timings;
- monitoring card's temps, they don't go over 60 Celsius but still getting driver recoveries;
- clean Vista install;
- updating OS/drivers;
- flashing card BIOS (fixed another issue when the system would lock up when playing older games, the TDR problem persisted);
- removing 1 stick of RAM;
- downclocking RAM;
- testing other driver versions (all versions from 8.6 - 8.9, including 8.10 RC2) with driver sweeps inbetween every new driver installation;
- installing only display driver (no CCC);
- attaching an additional fan blowing above the card/SB heatsink, fearing that it might be my SB overheating;
- getting a replacement card (experienced the same with my first card, same thing after replacement);

Additional notes:
- memtest/Orthos show no errors after hours of testing, so I rule out mobo/RAM compatibility as a reason;
- using my old card 6600GT doesn't produce such problems, so I rule out PCI-E slot being faulty;
- after disabling TDR in Vista through registry, I experience system (screen) lockups instead of a TDR message, but, a funny thing that I noticed: I tried shutting down the computer by pressing the Power button trying to avoid manual reboot and suddenly my cursor appeared on screen and I could move it around (the frozen screen remained in backgrounds). Unfortunately, the system didn't respond to anything after so I had to manually reboot;

Things I'm suspecting:
- possible mobo/card compatibility;
- a driver problem;
- Vista problem (haven't tried XP still, but, according to some people, even if such problems don't appear in XP, it doesn't mean that XP is the solution to the problem since XP as an OS is, allegedly, "more fault-tolerant" and I would be only avoiding the problem);

Any suggestions would be welcome."

eggmancs

dawny - think there is a new bios out for your motherboard which has fixed some issues for other users with 4870x2's. Give it a go.

Link

len1x

I'd like to say that I installed Xp 64 as my 2nd operating system ( before 2 days ).. and it's stable as a horse. not an error!

simland

So, I've read a bit and I figured I just toss this out there in case anyone can give me advice.

I've got a Visontek 4870, Vista 64bit, 8gbs of ram, E8400.

Everything runs fantastic, except when I play WoW. For some reason, only in WoW I get an atikmdag driver crash. Sometimes it just causes the screen to go black and "reboot". Other times it will freeze the entire system. This can happen after 10 seconds of being ingame or 2 hours. Any ideas?

mscrivo

Hey guys,

I have a different twist on this problem. I just built a HTPC system with an ATI 780G based motherboard. It has 2GB (2x1GB) sticks of Kingston DDR2 1066MHz RAM. There are no additional addon cards in the system, only an LG Blu-Ray writer and it's running on an Antec 430W power supply. I've installed Vista Ultimate 32-bit, with SageTV for TV and PowerDVD 8 for Blu-Ray playback and not too much else.

The system is completely stable except for when I run SageTV to watch TV and use the PowerDVD MPEG decoder to decode HD over the air Channels. I use the PowerDVD decoder since it makes use of the hardware acceleration built into the Radeon 3200 HD. In this mode, my system will crash with the dreaded ATIKMDAG.sys BSOD. It usually shows the pop-up box first saying that it has recovered from a video error, but then a few seconds later, the screen will go black and the BSOD appears. If i use the built in SageTV decoder which doesn't use hardware acceleration, I can watch TV for hours and hours with no problems at all. I'd use that mode and be done with it, but the CPU usage is high and the picture is not as smooth.

So in my case, it's clearly the UVD acceleration that's causing the problem. I don't play games on this system because it's too slow for that, so I have no idea what happens in games, but the above scenario is the only one that causes it on my system and it can happen anywhere between 10 mins of viewing to 2 hours of viewing TV, but it always happens without fail.

btw, the output is through HDMI and the resolution I'm using is 1920x1080.

I've also tried underclocking the video chip in the BIOS and it made no difference.

Anyone have a similar experience?

Thanks,
Michael

macbiter

my sony laptop with 3450 does this when im watching a MKV file, randomly. its really pissing me off. never again will I buy ATI laptop, first time ive bought one powered by ATI as I heard the IQ was great.

Nightmare.

Zardon

Interestingly enough ive seen this exact same problem with windows media player on vista with the MKV codec and videos (he has a 3450 as well). it seems to happen randomly, I have my friends laptop to look at, if I manage to find a solution to the BSODing ill let you know. been playing with it all day on and off and it seems to do this every so often. Ive tried using mobility modder on newer drivers and it doesnt appear to help.

temeteus82

That problem sounds like it would have something to do with the Video overlays in the VMR....

Decoman

*deleted by myself*

Freelancer

Just want to say this is not an ATI only problem. Some Nvidia users also suffer this problem (of course not with the same driver name).

My problem seems gone after I increased the DDR voltage manually (I was using auto and the reported voltage in windows was below spec). I have also upgraded to 8.10 beta driver, so far no problem.

P.S. I also upgraded my Realtek LAN and sound drivers to latest versions in case they indirectly cause problems.

maxwell005

Freelancer how do you know why to change the voltage ? you compare it to what? and how do you do that?

Something else: I got a motherboard with pci xpress x16, and my graphic card is pci xpress 2.0 designed... do you think its a compatibility problem ?

Freelancer

Well some software like CPU-Z will show you the "SPD" values of your memory - supposed to be the specified timings and voltage. While some software like Speedfan can show the current voltages including memory. I found that my current voltage was below spec, so I went into BIOS to change it.

eggmancs

PCI Express 2.0 is the version of PCIe, while the 16x is the speed at which it runs at - theres no problem there.

My problem is definitely a RAM issue. I can take out one stick of RAM and it runs perfectly fine. I tried changing the boot.ini advanced settings to manually say how much memory to boot up with but when I do this it just forgets what I say and puts 3Gb (instead of 4 - as shown in task manager and other programs). When I dont do the boot.ini thing it says 4gb and is alot more stable - but still crashes.

dawny

like what i say bro... i ve try everything. about upgrade my bios. I already upgrade it when my MB first active,. But.... nothing can stop " Display driver atimdag stopped responding"

simland

Ok, I posted a few back during what I like to call my "Ignorant" phase of this experience. I now understand that this error is essentially a catch all from the TDR and means jack diddly in identifying the root of the problem.

Let's try this again, once more and this time with feeling:

System Facts and History:
Mobo: Biostar I45
Memory: GSkill 4x2GB 800 DDR2
CPU: Intel E8400
PSU: Zalman Modular 750W
VGA: Visontek 4870 (Why the hell when I go to register my card online this isn't listed under the drop down?)
OS: Vista Ultimate 64 Bit
Soundcard: SB X-Fi


I built my system about 1 month ago. Fresh install of Vista 64bit. Updated all drivers and everything was running fine and dandy. I played Crysis, SPORE, Warcraft III and Orange Box all without issues. Merc 2 crashed to desktop a couple of times, but no ATIKMDAG error, just felt like game specific crash. Able to run everything at MAX settings aside from Crysis. I install World of Warcraft and start playing. Everything seems fine for a while. Then out of no where I get the screen going black and everything coming back with a ballon saying the ATIKMDAG recover message. Sometimes this would happen once other times it fires off 2-5 times. The worst is when the sound continues, the screen is frozen and no key inputs result in any change. CAD does not work. It could be that the driver was unable to recover and thus the screen is "fubared" minus a hard restart with the power button.

I put up with it for a while because I could go 3-4 hours at a time with nothing wrong. It was during the times where I had a series of hard reboots in a short amount of time that I decided to try and fix the problem.

I tried "disabling" powerplay by setting up a profile that forced a clock speed of 750 at all times. This did not seem to have any statistically significant effect. I should mention that with all these custom profiles I manually increased the fan speed to about 40%. My GPU temps are in the 40C-60C range. It was at this time that I ran MemTest. It ran 11 tests total and with 0 errors.


Finding no comfort in disabling power play in Cat version 9 I decided to try the beta 10s. These drivers actually started the ATIKMDAG error to start popping up on the desktop at, for the most part, random times. The driver recovered all times it crashed on the desktop. Using CCC and DD 10 did not solve the problem in WoW and I'd be willing to say increased the frequency of the crash. I decided to play Crysis and see if the v10 drivers were just bad all around. The game ran splendidly for 2 hours and I quit out due to real life engagements.

I decided to see if WoW was broken (should have done this first.) I ran WoW repair and to my surprise it fixed 3mb of data. Some of the data was in a divx*.dll and some of it was in the .exe. This got me really excited to see. I started up the game and without fail about 5 mins in I got 2 crashes. (The driver recovered from these.)

Irate I decided to wipe the v10 drivers off and roll back to 9 with a fresh driver install. I uninstalled all ATI software and drivers. Booted up in safemode and ran Driver Cleaner; deleted all record of ATI stuff that I could find. Installed v9 with the CCC. Rebooted again. Started up WoW and played for a while with no “powerplay” disabling profile enabled. The driver crashed when my character cast a series of spells. It seemed fine when I was just walking around. Unfortunately, it caused a hard crash. Figuring Powerplay was at the root I created a profile that forced 750 at all times and upped the fan speed.

The game again ran well until I cast a series of spells and the driver crashed 3-4 times in a row (recovering each time.) It’s at this time I jump into WoW and disable all graphically ‘intensive” settings.

That’s where I am now. I’ve not had enough time to see if the crashes continue. However, if they stop I will still be very upset. WoW should not be demanding on my 4870 in the least. Why is it that, until v10 drivers, WoW was the only app to cause the crash? My next fix attempt will be turning off TDR and seeing if the driver just needs vista to just leave it alone.

Looking for help, advice or a shoulder to cry on. Thus ends the Wall O’ Text and thanks if you actually read it all.

maxwell005

i think i got rid of it... i have 2 1gb ram stick that was on dual channel. I put the two of these on a single channel and that di the trick. Just a slight lost of perf but at least i can play games and not worry about my system crashing every 10 min

simland

Ok so I have a possible update and a possible solution. It needs a good 4-5 hours of testing before any conclusion can be made. If anything this post will highlight my ignorance on computer hardware.

So I tried a few things:
Downloaded GPU-Z to better observe the inner workings of my card.

A profile that has the clocks dialed in at 720 rather than 750. CRASH

I tried to change memory timings and voltages. Wow did I fail on that. Reset and promised to never speak of it again. SYSTEM CRASH

Disable Hardware Accel for Sound. CRASH

Turn all settings to low and sound off in WoW. CRASH

Scream Obscenities. NO EFFECT

Disable V-SYNC…

Wait a minute. 2 Hours without a crash.

I need to run it another 2 hours before I can be safe in saying that something has changed. What I have come up with is logic to support why this may be a fix. When I turned V-SYNC off, I noticed that my FPS skyrocket to 170+. Neato, but this has possible implications. GPU-Z reports that my card goes back and forth between 26% usage and 0% usage while V-SYNC is enabled. With an LCD monitor at 60hz that means that my card can only report 60FPS. So if my card can process frames at 170+, to me that means it has buffered 2+ seconds worth of frames. The default value for TDR in windows is 2 seconds. It could be that my card would process so many frames before they were needed that it just took a “nap.” Windows TDR might have interpreted this as a failing driver. I’m not sure what exactly is going on between the two but I feel that there might be something to this. It would help explain why the newer more graphically intensive apps aren’t crashing. The card can’t maintain the same crazy FPS or take a break. It could just be that my monitor is causing a problem as well. It certainly feels so far that V-SYNC is causing a communication or timing error. Will report more as info rolls in.

goaty

@simland, I suggest you try disabling TDR through registry and see if you'd experience a hard lock instead of a driver restart. That way you can be sure if it's the TDR system or not.

simland

Looked for TDR in registry and was unable to find it. I searched all keys, values and data. Turning TDR off and or turning on debug mode was going to be one of the things I tried, but I obviously couldn't find it.

goaty

You need to ADD the key, there's no such key by default. Search the Microsoft site for instructions, rather busy atm, I can try helping you a tad later if you don't manage to find it.