ATI HD 4870 Persistant Crashing

By hittman
Sep 29, 2008
  1. So I recently built this new computer... and the following problems haven't presented themselves until recently:

    I'm trying to play Valve games like Half-Life 2, Day of Defeat:Source, Counter-Strike:Source, TeamFortress, etc. My computer will load game and I can play for about 2 mins then it crashes the application. I'm also getting periodic reboots when just sitting on desktop web-surfing and whatnot.

    Here's what I've tried so far:
    Clean install of factory ATI Drivers & Catalyst CC that came with card.
    Tried the new updated drivers from Visiontek, ATI, etc.
    Let CCC adjust the settings manually, no over-clocking etc.

    I'm kinda noob to these sorts of things, but I can follow any directions you give.

    (my system specs are under my username FYI)

    Removed the long paragraph of junk :)
  2. EXCellR8

    EXCellR8 The Conservative Posts: 1,835

    My best guess would be that heat is causing the system to crash. The Radeon 4800 series video cards are notorious for running hot. The GPU core itself can function in temps exceeding 100c but the rest of the system may fail due to the overall ambient temp. The best way to prevent this card from overheating is to either create a custom CCC profile that defines a fan speed rating greater than 25% or you can use RivaTuner to adjust the fan settings (recommended). You should see a significant drop in core temp as well as overall system temp. If you do find that temperature is a factor in system stability, you could take the card apart and apply some better thermal compound to the core. I applied some OCZ FreeZe eXtreme to my 4870 and the idle temps dropped 10 degrees.
  3. hittman

    hittman TS Rookie Topic Starter


    So I've gotten the application installed and I think I've found the correct area to modify the fan speed:

    You suggested above 25% which was the default. Wasn't sure what range I was shooting for... so I tried 35%. Seems to have dropped my temp from 80C down to the 60C range according to CCC. (while idle)

    I've since launched the applications that were crashing and I'm still having the same issue, without fail about 2-5mins and I get a system crash/reboot. Maybe I need to set the fan to a higher percentage? -or- Perhaps I'm experiencing a different issue all together?

    Any suggestions are welcome, I'll try just about anything at this point.

  4. mailpup

    mailpup TS Special Forces Posts: 6,964   +355

    Maybe it's a RAM problem. Some motherboards default RAM voltage is lower than what higher performance RAM requires. Check the voltage settings in the BIOS against the requirements of your RAM.
  5. k.jacko

    k.jacko TS Rookie Posts: 493

    I take it you've plugged both 6-pin power molexes into your 4870?
    Maybe when it starts running at max its requiring more voltage than your psu can give it.
  6. Ultiweap

    Ultiweap TS Enthusiast Posts: 606

    Ok I don't think it is the graphic card as the card you have must play the game you listed easily withtout any problem. So i'm turning to your RAM and Processor.

    And also i'm thinking that it can also be a problem with you PSU. So can you ask me what PSU are you using? Asking that in case your PSU can't give all the power needed for the full performance of the card.

    Also some test that you can do in meantime do a test for your RAM.
    Tutorial: How to use Memtest
  7. EXCellR8

    EXCellR8 The Conservative Posts: 1,835

    You adjusted the fan speed correctly; if heat generated by the video card was the issue, increasing the speed would have probably done the trick. I usually set the fan to ~60% before I start gaming, which usually keeps the temps down below 55c during load. Sure, it sounds like a dust buster but the card stays relatively cool. I'm gonna have to go with these guys on the problem being caused by something else. It's a really good chance that your PSU, if under-powering the card, is crashing the system.
  8. Ultiweap

    Ultiweap TS Enthusiast Posts: 606

    Yeah it is unbelievable that a PSU not giving the full power did restart the PC as normally the circuit inside the card should have been burnt in normal case of even just shut down and maybe the PSU won't power on once more but sometimes it happen that the PSU just restart when using a high voltage for a peripheral that it can't support.
  9. hittman

    hittman TS Rookie Topic Starter

    So the PSU is made by Ultra (550W), and I have both 6 pin power connectors plugged into the card. This is the exact model of the PSU I'm using:

    It states that it's "SLI Ready" so I would assume running a single card shouldn't be a problem...

    As for the memory, I did have some trouble getting the computer to load the OS when I first put it together. The factory settings in the BIOS of my motherboard was setting the RAM to something like 400MHz. I set the RAM to 1066MHz as suggested by another forum and it booted fine after that. ( aside from the constant crashing I now get =P )

    As for running a check on the memory, I haven't done so yet. I will attempt to download and install the software suggested when I get home from work.

    Thanks for all the help, and I will let you know what happens.
  10. hittman

    hittman TS Rookie Topic Starter


    Thought maybe some stats on what the ASUS PC Probe was reporting might be helpful as far as the PSU is concerned.

    Ranges from....
    1.26V on Vcore
    11.84V to 11.90V on +12V
    5.11V on +5V
    3.23V to 3.25V on +3.3V

    Also another note, the RAM is set at 2.1V as suggested by the sticker on the stick itself. Still trying to get that MemTest onto a bootable CD, I don't have any burning software loaded as of yet, so it's making it difficult to do.
  11. Ultiweap

    Ultiweap TS Enthusiast Posts: 606

    Ok now I see. I don't think it is you PSU then as it is completely ok for supporting the card so make the test of the RAM with memtest.

    For burning software I will suggest this one easy and small one:
    Magic ISO
  12. hittman

    hittman TS Rookie Topic Starter

    Memtest86+ Results

    Results of the Memtest86+ utility after 3.5hours:

    WallTime: 3:32:56
    Cached: 4095M
    RsvdMem: e820-Std
    Cache: on
    ECC: off
    Test: Std
    Pass: 4
    Errors: 1734144
    ECC Errs: 0

    Error Confidence Value: 118
    Lowest Error Address: 0000010d2d0 - 1.8MB
    Highest Error Address: 0012fffbff8 - 4863.6MB
    Bits in Error Mask: b109a1ff
    Bits in Error: - Total: 17 Min:1 Max:8 Avg:1
    Max Contiguous Errors: 1
    Errors Per Memory Slot

    Tests 0,1,2,3,4,6,7,9: No Errors
    Test 5: 32767 Errors
    Test 8: 26 Errors

    Trying to read up on what this all means... only thing I could find was that errors in tests 5 and 8 seem to point to bad memory. (which is what I experienced in my tests) So I'm thinking maybe bad RAM, but which sticks? Do I need to test one by one, or does this information also tell me which sticks are bad?
  13. nickc

    nickc TechSpot Paladin Posts: 923   +11

    u definitely have bad ram but the way to find which stick is to run the test on each stick individually, run at least 7 passes on each stick.
Topic Status:
Not open for further replies.

Similar Topics

Add New Comment

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...