xmr-stak-nvidia
xmr-stak-nvidia copied to clipboard
fireice-uk
"No Cuda Device Found"
Hi,
Compile went fine, but it can't seem to find GPU? -
`XMR-Stak-NVIDIA mining software, NVIDIA Version. NVIDIA mining code was written by KlausT and psychocrypt. Brought to you by fireice_uk under GPLv3.
Configurable dev donation level is set to 1 %
You can use following keys to display reports 'h' - hashrate 'r' - results 'c' - connection
[2017-03-24 00:27:58] : Connecting to pool pool.supportxmr.com:3333 ... No CUDA device found! [2017-03-24 00:27:58] : Setup failed for GPU 0. Exitting. ` This would be Linux x86_64, Nvidia GT760. Any thoughts?
Best,
How did you fix it? I'm seeing a similar issue. I don't even get to the "Connecting.." message. When I run it, it just says "No CUDA device found!" and exits. I have an NVIDIA card..
@fireice-uk I wrote issue https://github.com/fireice-uk/xmr-stak-nvidia/issues/39 because the miner failed to recommend a configuration.
I've downloaded and run the most recent release (found at https://github.com/fireice-uk/xmr-stak-nvidia/releases). After the initial run, it gave me suggested copy/past settings and I entered those in the config.
However, upon attempting to run it after that I am getting a similar error as prometx, which is as follows: [2017-06-06 17:16:57] : Connecting to <mypoolhere> ... [2017-06-06 17:16:57] : Connected. Logging in... Unable to query number of CUDA devices! [2017-06-06 17:16:57] : Setup failed for GPU 0. Exitting.
prior to trying xmr-stak-nvidia, ccminer was working fine recognizing the device and hashing. System: Win7 64bit, Nvidia GeForce GT 650M. I'll mention that Intel HD Graphics are also onboard in the event that causes a conflict or trouble detecting for some odd reason.
Thanks for the speedy reply. Suggested config output was:
"gpu_threads_conf" : [ { "index" : 0, "threads" : 64, "blocks" : 6, "bfactor" : 6, "bsleep" : 25, "affine_to_cpu" : false, }, ],
Update: Spoke too soon, accidentally ran cpu miner, not gpu. If an email was triggered for the comment, please disregard. Problem persists.
Uninstall/Clean reinstall of graphics drivers got me past initial problem. Connects to pool, changes difficulty, detects new block, then crashes graphics driver and quits application. Error thrown: GPU 0: unknown error C:/Users/xxxxxxx/xxxxxx/Visual Studio 2015/Projects/xmr-stak-nvidia/xmr-stak-nvidia/nvcc_code/cuda_extra.cu line 246
Is there a particular Visual Studio Redistribution I need installed? I verified that VS 2015 was installed on the machine. If anyone has insight about what is causing this error, I will be grateful for it.
Windows is killing all long running actions on the gpu, with bfactor 8 it should run.
I changed to bfactor 8 and had no luck, so I kept tweaking the numbers in the config and still nothing. After a number of attempts, I got this error again: "Unable to query number of CUDA devices! [2017-06-06 17:16:57] : Setup failed for GPU 0. Exitting." Not sure what the deal is because other system reports show the drivers are installed and working.
After another uninstall/reinstall of the graphics drivers and another system restart, I tried bfactor 8 again and continually reduced the threads while testing each time afterward until I got it hashing without crashing but it's only been running for 15 min. or so. I noticed threads can be set higher when cpu miner is not running at the same time. Setting true/false for "affine_to_cpu" didn't seem to make a difference in stability, max threads or hashrate. The current max threads setting seems to be 35 without crashing, but it provides a very low hashrate.
I'll keep testing when/if I get more time for this. Thanks for the support.
On a side note, is there a way to increase the donation amount using a full release download, or would I have to modify this setting before doing a binary build and compile?
So the "Setup failed for GPU 0. Exitting" error returns every so often. I'm beginning to suspect a driver issue. I'll continue testing in spare time and update if I find a long-term stable solution in case it helps anyone else in future with a similar problem.
I have a switchable nvidia gt950m alongwith an Intel HD gpu. I am getting the same error No CUDA device found! Is there a fix for it?
This solution worked for me. I think your GPU card doesn't support some of the new versions of drivers.
Delete and purge all Nvidia and CUDA installations.
sudo apt purge nvidia*
sudo apt purge cuda*
Install Nvidia 375. (Older than 400 series)
sudo apt install nvidia-375
Download and install CUDA 8.0 GA1. (Older than 10 and 9 series) Select "yes" to everything except Nvidia driver installation, because we already installed nvidia-375 driver! https://developer.nvidia.com/cuda-80-download-archive
Restart and it will work.
