0

It's been a while since I have this problem with my system. Whenever I use programs (except games) that somehow use Hardware Acceleration such as Browsers, Video Converters, Photoshop etc, my screen goes black and refreshes again and in the notification area I get this message: ATI Catalyst control center stopped responding and has successfully recovered. As I've googled the net it appears that this problem is common for AMD graphic-card users.

This is what I've done

  1. Checked for any high electrical current on my case
  2. Cleaned every part of the computer Hardware
  3. Installed Windows once again
  4. Updated Graphic Card driver and any other driver to the latest version
  5. Checked my ram for any issue with Mem-test. nothing found on ram
  6. Updated my Bios to the latest version and reset all clocks to the default. Nothing happened so I continued overclocking.

My system specs are below

  • Asus-Z87-Plus
  • Core i5 4570
  • WD 1TB black
  • ATI 7850 Sapphire
  • 8GIG DDR3 @1600 compatible with motherboard
  • 850W power Supply
  • Windows 8.1 pro x64

I'd appreciate any help or suggestion which can lead to solve this problem not only for me but for so many people having this problem.

0

1 Answer 1

0

Disclaimer: It certainly can be software, it could be an issue with dirty power, you could find that there are manufacturer drivers out there that are not the latest, but are tested to be more reliable by users. It could be the power connections are not as well connected as they could be, and that it might not be seated well in the socket. I'm listing things you can test, and ways you can "improve" the situation IF it is the usual cooling issue.

The AMD card is more likely to hit this gpu recover which is a hard fail usually like a few bits of ram total wrong, or processing going awry, usually in high temperatures (probably ram from most of the designs).

Continued overclocking No! :-)
Run furmark test the gpu processor, then test the ram with an artifact testing, always watch the temperatures when doing the hard testing like that. By testing the processing of it "more" with the renders, and testing the ram of it "more" with the artifacts testing, you could see if one aspect of it causes the recover problem. Watch the temperatures and see what temperature it more likely happens in.

Crank up the fan till your ears bleed (100%), and test again, use the furmark or play a game with the fans going full bore.

Try to keep the temperatures displayed to the user below 80*C and see if it still does it.
Sure the GPU is fully capable of surviving some 100*C temperatures, and working fine (so they say) I just don't find that to be true myself, and the rams are often hotter than the GPU in different manufactured designs.
Clean out the dust.

Stop the automated stepping, and create specific profiles for the clock rates. I use "Green" as a Gpu 300- mem300 for desktop and 2d browser, and idling. I use "low" 600 - 600 for watching HD video, minor games that are mostly 2d, and editing and other low GPU difficulty tasks. I use a Stock profile called "Med" I manualy created for games, and I do have a "High" Overclock profile for those tough to render games.

Profiles are then easily accessed via the Catalyst notification tray icon (right click). For me it has always been easier and better to snap to the exact speed I desire, vrses the auto. Some people would find it a pain. My "green" 300-300 setting is lower than the software will even set it.
Potentially fan speed can be locked into manual profiles also, but I am using speedfan and advanced fan control (graphing) to get that automatic my way.

As a last effort when you have nothing to lose and no warranty, and are ready to throw it out, get ram pads, and thermal interface goop like as5 and dissassemble the whole thing, clean it out, change the thermal interfaces so it is new and fresh and clean, and reassemble more carefully than they do when pumping them out on a production line.

Mine works :-) But I breech the warranty right after I test it to be operational, I always use some fan graphing program, like afterburner, atitool (obsolete mostly), or this time it is speedfan, to change the temperature profile. I do not need it cooler than 60* and I do not want it hotter than 80*. I get it in that zone as best as possible.

I find drivers with the least stability complaints and everything goes perfect for years. I avoid updating drivers when it is working properly, unless the update states that it solves specific issues with a game I am having that specific issue with. I would also update if they would give us better control of the thermal profiles without 3rd party software.

The Bclk adjustments on the motherboard over 100 increase the PCI bus speed, I tried for a long time to get minor bclk tweaked just a bit higher like 102 & 105, because it would Push the Video and my raid a little harder. On my motherboard an Asus Hero, changing the bclk never did anything useful and everytime created instabilities that did not show up right away.

8
  • first thank you for your detailed answer, really appreciate it. i gotta say my graphic card stays as cool as possible even when the clock is 1200MHZ while base clock is 860 and memory clock is something around 1420 with some over voltage. with those spec it passes every time i run Furmark, i use 2x 220v fans for it. the problem doesn't happen on high system load but on low system load. sometimes as low as when a software starts to get to know the graphic card. and sometimes on Auto cad 3D. not a big deal for that card. i use MSI after burner.
    – user326006
    Jul 10, 2014 at 20:36
  • 1420 was to high for the memory, I saw other people doing that , and wondering why it fails in games, even though it worked in testing. I figure the max is more like 1300, from research and my own testing, they already push the memory stock. Try locking the stepping. Yea the problem occurs when doing the most mundane tasks to playing the hardest games, it was happening to me before. After making all these changes I have not encounted the problem again. I added one more to the bottom.
    – Psycogeek
    Jul 11, 2014 at 13:19
  • well I've been playing games for a long time on 1400 and nothing happened. it appears that your problem which caused this error for you was temperature issue and with your idea it solved. but i really don't think the same about my situation. my card's temp doesn't go further than 70.C in ANYway. when i set my mobo to auto it sets the BCLK to 102 so i leave it unchanged. AMD's website has an article about this and one of their suspicions is about Power supply. i don't see any reason why power failure should occur when i'm browsing the web with Hardware acceleration turned on!
    – user326006
    Jul 11, 2014 at 13:40
  • If you had a complete power failure , then the whole computer would blip out. If the curcuits for the 12v driving seperate "rails" powering the video card , which can be seperate curcuits blipped out it is possible to knock out the video card. I doubt a recovery happening during simple desktop actions is from using Too Much power. Why would you control the overclock of your video card, without controlling the entire base frequency by which it operates at?
    – Psycogeek
    Jul 11, 2014 at 15:16
  • i need to check the connectors on my power now. and what do you mean by your last question? if you're talking about bclk, i don't change it because it will make my computer unstable and lead to data loss on hard drive. isn't that right?
    – user326006
    Jul 11, 2014 at 21:53

You must log in to answer this question.