R9 290x ran great for a while, but now -

edited January 2014 in Hardware support

Mining ftc with 2 - R9 290x Win 7 Home  64 bit. cgminer 3.0.1


Overclocked yesterday, +15% for power and clock speed, ran at 95C for a number of hours. Obtained kH/s of 780+ Both cards were over 700 kH/s
Used these parameters loading cgminer
cgminer --scrypt -o ftc.give-me-coins.com:3336 -O user.miner:pass --thread-concurrency 8192 -w 256 -I 18

I shut the system down completely for about 4 hours last night. Ran cgminer again overnight Used defaults -- no overclocking. 8 hours. It ran very slowly in comparison to the previous performance. 12 - 16 kH/s according to cgminer.

Today, shut down to try to troubleshoot. Unable to run cgminer using  --thread-concurrency 8192 -w 256 -I 18 Error message about thread 0, killing thread 0. Card driver stops and then restarts.

Can anybody help?

Comments

  • Maybe diagnostics could be run on the cards to determine whether they have been damaged 

  • Several things.

    First, 95C is too hot!!   You should not do sustained operation above 80C.  Use --auto-fan, and set your --temp-target at 80, --temp-overheat at 85, and --temp-cutoff at 90.


    Second, use a newer version of cgminer.  The latest version that still supports GPUs is 3.7.2.

    Lastly, Delete any xxxx.bin files that cgminer created in the directory where you started, and let it re-create them from scratch using the parameters you have selected.
  • I'm getting around 880 KH/s with a R9 290x using:


    "worksize" : "512"
    "intensity" : "20"
    "thread-concurrency" : "32765"
    "gpu-engine" : "922"
    "gpu-memclock" : "1500"
  • Maybe you shod try smos-linux from smos-linux.org
    I'm using it for 280x and works fine

  • You might want to have a look at this thread for info on heat management, fan management, and gpu clock management.

    https://forum.give-me-coins.com/discussion/666/heat-death-how-to-kill-your-gpu-in-less-than-a-year#Item_7

    I have a couple of r9 270's working fine with the config shown in that thread.  I use TC = 0 and let cgminer figure out how to max out the card.  It seems to do OK.  I agree that 95 C is too hot.  See the temperature parameters in the config file I quoted in the other thread.

    Sincerely,

    Ron

  • I just posted a similar situation about my 5750 now being stuck in low level mode. Sounds like your card is too. Open catalyst / overdrive and see what clock speed it's running at. I made a Bamt USB drive and even in Linux my card is stuck at the "low level" or idle speed. I too would like to know how to get it out of that mode.

    In my case it was working just fine then stopped so I don't think drivers became an issue and after a test in Bamt with the same issue I'm certain it's not the drivers, OS or CGminer version. Question is does a damaged card revert to Idle mode and somehow lock itself at that speed?

  • If running Windows, you could try MSI Afterburner:

    http://event.msi.com/vga/afterburner/download.htm

    or possibly ASUS GPU Tweak:

    http://support.asus.com/download.aspx?SLanguage=en&m=gpu tweak&os=30

    to reset the card to the default clocks then reboot.

    If you're running Linux, I posted some instructions here about resetting cards:

    https://forum.give-me-coins.com/discussion/708/hd-7990-shutdown-pc-when-second-gpu-intensity-is-more-than-12#Item_15

    Sincerely,

    Ron



  • edited January 2014


Sign In or Register to comment.