Ethereum Community Forum
Home› Mining
Vega Finally Unleashed! Fine-tuned performance and power consumption in a single chassis.

Strange problem occuring on GPU3

Rigboy786Rigboy786 Member Posts: 45 ✭
edited February 2018 in Mining
I have 2x R9 290 and 2x R9 290x so effectively 4 of the same GPUs, to make my life easier.
I have not done any overclocking yet as I want to seem them run stable first and then I will tweak slowly.

GPU0 t=67C f=65%, GPU1 t=71C f=65%, GPU2 t=66C f=65%, GPU3 t=76C f=97%
ETH: GPU0 26.836 Mh/s, GPU1 26.831 Mh/s, GPU2 26.762 Mh/s, GPU3 26.668 Mh/s

AFter a while I will see GPU3 drop off to 0.00 Mh/s and then in about 1-2 minutes the system will reboot and all will be fine, until the next time the GPU stops hashing again. Sometimes it will last 10minute, sometimes 10 hours.

I opened all GPUs, re-applied thermal paste and I cleaned all the GPU's, coolers and fans... everything is like new.

Can anyone share some ideas what could be causing 1 GPU to stop hashing?
· Share on TwitterShare on Facebook

Comments

  • headshot155headshot155 Member Posts: 158 ✭✭
    I have 13 identical cards and I get a very varied performance from them with exactly the same settings. Each manufacturer have tolerances in the manufacturing process and performance scopes for the memory they purchase - so what is in scope for normal GPU work will be different to mining as the stresses you put on the card are quite different. Your current stock settings are stressing GPU3 enough for it to stop mining. I think that stock settings for cards are way more stressful for cards relative to overclocking if done correctly.

    Wow, it must be noisy with a your GPU fan at 97%. People seem very fearful of overclocking, but I think its a misnomer. To get a GPU to work best you should be underclocking the core and perhaps slightly overclocking the RAM, and also undervolting. In a gaming scenario this would lead to a significant degradation in performance while for mining it means better performance while producing less heat and consuming less power.

    In underclocking and undervolting you will find the fan speeds drop significantly and so will the stress on the cards.

    Start reading into changing bios memory straps etc - this website https://mining.help is a great place to begin
    · Share on TwitterShare on Facebook
  • lablettlablett Member Posts: 333 ✭✭
    Download HWINFO and see if you have any GPU errors. Also compare the voltage/watt values shown in HWINFO. Also look at the graphs in GPU-Z to see how the card is behaving compared to the rest. I would also check the speed the PCI is running on each card. For me GPU3 is running too hot and I hope HWINFO will give you an indication. Like the above post I would look to reduce the voltage,etc. When I have a card that give me a problem I just drop the memory speed until it is stable. This will give a short term fix. Also can you improve the airflow around GPU3 ?
    · Share on TwitterShare on Facebook
Sign In or Register to comment.