Display driver amdkmdap stopped responding and has successfully recovered

Jun 26, 2015
35
0
4,530
So I just finished building a new PC. I had a small problem with a loud fan and made a thread about it, but it's been solved. But now I have a new issue. My system keeps crashing and when I check the event log, I see this message:

display driver amdkmdap stopped responding and has successfully recovered

There are a lot of these. The PC will freeze, the display will go out, then it will come back, then go out, then come back. I tried the dxdiag program in Windows 8.1, it found no issues. I installed everything off of my mobo disc, and I used the AMD auto detect program and it says I have the most up to date drivers.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813157528 - this is my motherboard

http://www.newegg.com/Product/Product.aspx?Item=N82E16814202137 - this is my GPU.

Nothing in the system seems off. Everything recognizes the GPU, the fans work, nothing is overheating. The only thing in my device manager is my SM Bus Controller has an error with it. What does this do and could it be causing the problem?

I bought StarCraft 2 for the new PC and when I try to install it, the screen flashes 3 times and I get 3 error messages saying device driver AMD has stopped working but recovered. It's when I am doing general things on the PC like using the web browser and it crashes that it really messes the PC up.

Idk if I should send the card back or try Windows 7.
 
I would suggest doing a Safe Boot (hold down F8 while booting), select Safe Boot with networking. Once there, uninstall all AMD video card related drivers and Cataclysm. When done, download the free copy of CCleaner and run it. It will clean up various parts of your computer, including anything that might have been left over in the registry when uninstalling the AMD drivers. When CCleaner is done, Restart the system going back into Safe Mode just like the first time.

This time, download the latest available video drivers from AMD, and install them. Same with Cataclysm. When done, reboot and go into Windows normally. Hopefully, things will go a little smoother for you this time.
 
Jun 26, 2015
35
0
4,530
This unfortunately did not work. I went into safe mode, I uninstalled all things AMD and my driver. I downloaded and ran CCleaner, I dow loaded the new Catalyst which installed the same driver I already had.

I began the SCII installation and it actually ran for about 10 minutes, but the machine just flashed the screen a couple times and crashed. And even just now as I try to go to the event viewer it is slowly flashing from desktop to blank screen and will probably require a restart.

So I have restarted and checked event log. Lots of display driver amdkmdap stopped and recovered warnings, also some errors about Service KMSELDI terminating unexpectedly. It seems each batch of driver display warnings is preceded by the KMSELDI service shutting down.

Idk what to do. I might give Windows 7 a try and if it doesn't work, send the card back.
 
Jun 26, 2015
35
0
4,530
I thought I had, but maybe not. I will try when I get home. I have also been reading it could be a factory OC GPU. The Newegg description does say factory clock 950Mhz or something, and 1000Mhz boost clock, maybe I should try lowering it to 950?

I also read it could be bad memory. I will run memtest86+ when I get home. Lastly, this is the PSU I bought for the new build. 700W should be enough, yes?

http://m.newegg.com/Product?itemNumber=N82E16817153167
 
Right now, I am more inclined to suspect the software than the hardware. Because the video driver is failing, and that its normally a software problem, but not always.

99.9% of the time, factory overclocks are fine.

We keep a list of power supplies here on Toms. The Thermaltake TR2 has not yet been rated. That normally means the company has not bothered to send units out for reviews, which I tend to take as a sign they do not want reviews on them. So while I cannot say that it is a bad or good unit, I would be suspicious of it right now. Other power supplies from Thermaltake range from Tier 2 to Tier 4 (good to junk), so I cannot base it on that.

If the power supply is a quality power supply, then that would be enough capacity. If on the other hand, its a junk unit, it might not be. But I do not suspect the power supply right now.

Get the Intel drivers installed and see if that makes a difference. And yes, testing the memory cannot hurt.
 
Jun 26, 2015
35
0
4,530
I installed the latest Intel drivers but it made no difference. It has crashed 3 times in the last 20 minutes since installing them. I do see artifacts on the screen both when Windows loads and on the Windows 8 start screen. Idk if this means the card might be bad or if faulty software issues can make that happen.

I downloaded memtest86+ but have no idea how to use it :\ I'm also at a loss on how to underclock my GPU. I'll look into it.
 
On memtest, there are 2 sites... the one I use is http://www.memtest.org/

Before you run memtest in the next step, I would like you to shutdown the system, and remove your video card. That will eliminate it from being a source of any errors during the memtest run. You can plug your monitor into the back of the motherboard and use the IGPU in the CPU. Memtest doesn't do graphics, so it won't matter.

Download the ISO image. Burn that to a CD/DVD. Then boot from that disc. Let it run overnight. If it finds any errors, it will report them to the screen. This is assuming that it makes it overnight. If there are no errors when you get up in the morning, then its not likely to be the memory.

I am beginning to suspect the video card, but lets eliminate the video card best we can first.
 
Jun 26, 2015
35
0
4,530
So I ran it this morning, both sticks passed with no errors. Also, I left the card in as I was pressed for time before I ran some errors and the card frankly is a pain in the butt to remove. So I guess the memory is fine. I am about to go to work, when I get home I will use my 7 disc and try a clean install of that and see if the card works better on that OS. If not I guess I will send it back to Newegg.
 
Jun 26, 2015
35
0
4,530
Ok, I did a clean install of Windows 7 this afternoon. The card displayed everything fine throughout the entire installation, the entire time I used Windows 7 installing things...

..until I installed the newest driver for the R9 280X. Same artifacts on the screen, same "AMD driver stopped working but has recovered" message from Windows 7. I am about to try to use an earlier driver for the card and see what happens, though this didn't work on Windows 8.
 
Jun 26, 2015
35
0
4,530


I have begun the RMA process. Last night I uninstalled the card, removed all drivers, and hooked the HDMI cable up to the on board graphics. I installed Starcraft 2 and was able to play fine. The game suggested medium on most settings, I chose low just to be safe. I didn't see any issues.

However today I am seeing the same reddish artifacts on the screen as I was when I was using the graphics card. I am including 3 pictures, hopefully you can see the weird spots clearly, sorry about the dirty monitor. I'm just gonna link the URL, the pictures are pretty massive.

http://i.imgur.com/QvGvBAG.jpg - This is from a forum. The spots are all around "New Topic" and along the right side.

http://i.imgur.com/ycJ0Eyk.jpg - Same, maybe a better view of the right side.

http://i.imgur.com/BVNnL9U.jpg - Just the desktop.

They are really easy to see when the screen is on something dark. Also images. I have no idea what this could be.

Also, my GPU box says a 750W PSU is recommended. Mine is 700W. The build should only be pushing slightly above 400W according to PCPartPicker.

I am worried now that maybe I am going to mail off the GPU tomorrow and it's not really the issue.

 
Yeah, I checked 2 places that I on the power supply minimum, and both also say 750 minimum.

That power supply is an unranked power supply on the list they keep on power supplies here. Which normally means that there have been no reviews done by trusted reviewers. Which normally means the company selling the units does not want them to be reviewed. Which normally means they are less than high quality units. Which is not a good thing.

So I looked at the egg rating on Newegg. 3 eggs out of 5. That concerns me. Most of the time, a good unit with over 100 reviews will show 4 eggs or better. I will stop there, because this involves a whole lot of speculation.

If you want to get a 750w unit, I would like to see you get a high quality unit. II recommend Seasonic power supplies. I believe that they are the best made power supplies on the planet. Its is the only brand of power supply that I will buy anymore. So I went to Amazon, and asked it to show me Seasonic 750w power supplies. Here is a link to what came up:

http://www.amazon.com/s/ref=nb_sb_ss_c_0_13?url=search-alias%3Dcomputers&field-keywords=seasonic+750w&sprefix=seasonic+750w%2Ccomputers%2C283&rh=n%3A541966%2Ck%3Aseasonic+750w
 
Replace your power supply, not the GPU. The Thermaltake TR2 is the exact power supply that brought me to Tom's Hardware in the beginning. My very first post here is related to that power supply. I've literally seen at least 100 threads here on Tom's alone that were related to that power supply model and erratic behavior, failures and "ghost issues" are common indicators of the TR2 troubles.


There is always a chance of something else having been damaged by a cheap power supply, but considering your issues linger and are similar, even with the GPU card removed, is a good indicator that the common denominator is the power supply which can in fact not even come close to supplying it's rated power and has many other issues besides. It used to be a Tier 5 "replace immediately" unit, but they have released a new version of the TR2 so it dropped down into the unreviewed list.


Most models still being sold by TigerDirect and some other vendors are highly likely to still be the older model though. There is also no proof at all yet that the revised model is any better.

http://www.tomshardware.com/answers/id-2194316/likelihood-psu-failure.html
 
Honestly, it's much better than the TR2, but still a lower quality unit. I'd like to call it a piece of trash, due to the high failure rate, but it's not quite that bad. You can do MUCH better for the price, like this Super Flower built unit that ranks up there with some of the best units available:

PCPartPicker part list / Price breakdown by merchant

Power Supply: EVGA 750W 80+ Bronze Certified Semi-Modular ATX Power Supply ($73.99 @ SuperBiiz)
Total: $73.99
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2015-07-09 17:17 EDT-0400


But this unit is more than enough for your configuration and is a Seasonic built unit. Very good unit too:

PCPartPicker part list / Price breakdown by merchant

Power Supply: Antec Neo Eco 620W 80+ Bronze Certified ATX Power Supply ($64.99 @ SuperBiiz)
Total: $64.99
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2015-07-09 17:18 EDT-0400
 
We'll pray to the PSU gods for you. Don't get too depressed if there is still an issue. It's neigh unto impossible to figure these kinds of issues out with a PSU shaped object installed in place of an actual PSU anyway, so at least it will narrow it down greatly if it doesn't end up being the issue, which I think it is. Plus, at least you'll know you have a very good, solid, reliable unit installed that's going to last you another five to ten years. Don't forget to register with EVGA when you get it too.
 
Jun 26, 2015
35
0
4,530
Update... the new PSU did nothing. I have it installed and the 280X back in the machine and I am getting the same issues as before. The same red artifacts, the same crashes. Although, I could have swore the latest driver from before was December of 2014 but now it's telling me I have the latest driver from November. Not that it matters, I guess. Something is messed up anyway.

I will try installing the driver from the disc again and fiddling around with BIOS, but I am lost as to what the issue could be. The same artifacts appear even when the video card was not in the system. Idk.
 
I don't ever recommend running CCleaner or any other registry modification software. It's too easy to screw things up, plus, if you've just recently done a clean install it's doubtful that much is there anyhow. I'd run the DDU as outlined at the following link, then install the latest drivers for your OS and card as indicated on the AMD website. Link to all relevant software is at the following link in addition to instructions on what to do:

http://www.tomshardware.com/faq/id-2492424/laymans-simple-guide-solving-graphics-card-issues.html
 
I use CCleaner on my registry now after almost any uninstall of software. They just leave way too much junk in the registry when you uninstall things.

But I also create a restore point before doing those things just to be safe. So far, I have never had to use one of those. I'll keep knocking on wood too. :)
 
I like the DDU much better for graphics card related cleaning as it only focuses on known files and entries that are related to the selected card family and doesn't bother other stuff in the registry that may be unrelated unbeknownst to the novice or even intermediate user. Safer for them and they don't tend to be as likely to mess anything up.
 
Jun 26, 2015
35
0
4,530
I am trying the DDU right now.

When I uninstalled the current driver it reverted back to a 2013 driver for the card and it gave me numerous "driver has stopped working" errors, but the machine never crashed. It crashes constantly with the latest drivers installed.

With a generic driver installed the machine works fine with no errors or crashes. It only seems to be when the drivers are installed that artifacts and crashes occur. Also I had PCIE set as the default graphics in BIOS even when the card was out of the syystem, Idk if maybe that's why I saw artifacts even using on board graphics.

Catalyst installing now. Another weird thing, the display would not go full screen with AMD drivers installed. If I have to RMA thus card I might go with Nvidia even if I have to downgrade a bit.

Catalyst almost done now. Ok so now it's giving me a driver from June 22 of 2015? Bizarre.

Ok so the artifacts disappeared, I loaded up Starcraft II and saw no abnormalities, the display was in full screen, and now the crashes have started up again. Just crashed for the 2nd time. I am rebooting into safe mode so I can maybe check the event viewer log without the PC crashing.

Ok so there are a lot of errors about the Intel HD Graphics Service or something stopping pretty much with every crash. I have disabled the service, let's see what happens.

Aaaaaaand all the artifacts are back. And right as I am about to take a picture at the SCII launcher, the machine crashes. :(

It really seems like a software issue as opposed to hardware. I have no idea why I'm only just now getting a June 2015 driver from AMD when it's always been Nov 2014 before now. I guess I will try a clean OS install tomorrow.
 
I'm not sure why you're bouncing from OS to OS. What version of windows is currently installed? Is it 32 or 64 bit? Did you disable the integrated graphics in the bios, if there is an option to do so there, and delete the display adapter for the integrated graphics in device manager? If not, I would do so.