WARNING: NVIDIA Insufficient driver!

[jsbranco]

Hi,

I recently built the v2.0 version and although I am able to run it on CPU, I can't seem to have it running on my GTX 1080.

It shows "Nvidia Insufficient Driver", and switches to CPU. Although, I have CUDA 9.0 installed and my GTX is recognised. Has anyone seen this? ./xmr-stak

[2017-12-08 08:31:27] : Start mining: MONERO
WARNING: NVIDIA Insufficient driver!
[2017-12-08 08:31:27] : WARNING: backend NVIDIA disabled.
[2017-12-08 08:31:27] : Starting single thread, no affinity.
[2017-12-08 08:31:27] : Fast-connecting to pool.supportxmr.com:7777 pool ...
[2017-12-08 08:31:27] : MEMORY ALLOC FAILED: mlock failed
[2017-12-08 08:31:27] : Pool pool.supportxmr.com:7777 connected. Logging in...
[2017-12-08 08:31:27] : Difficulty changed. Now: 25000.
[2017-12-08 08:31:27] : Pool logged in.

nvidia-smi

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 387.12                 Driver Version: 387.12                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 1080    Off  | 00000000:0E:00.0  On |                  N/A |
|  0%   41C    P8    13W / 210W |    701MiB /  8105MiB |      1%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      1165      G   /usr/lib/xorg/Xorg                           404MiB |
|    0      7023      G   compiz                                       173MiB |
|    0      7503      G   ...-token=A4DF6F97357101C23AE30692EEAD51E2   120MiB |
+-----------------------------------------------------------------------------+

nvcc --version

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2017 NVIDIA Corporation
Built on Fri_Sep__1_21:08:03_CDT_2017
Cuda compilation tools, release 9.0, V9.0.176

[psychocrypt]

Could you please restall the newest cuda version and restart the system. It looks like the driver is not updated only the toolkit.

[renownedd]

I am having this same issue when trying to run nvidia and amd gpu at the same time, I have the most recent drivers. Will this miner work with both cards at once?

[JerichoJones]

@renownedd - You don't specify the OS. I run AMD/CPU/NVIDIA at the same time on Windows.

[renownedd]

It was something I had messed up. I figured it out thanks!

On Sun, Dec 10, 2017 at 11:41 AM, JerichoJones notifications@github.com wrote:

@renownedd https://github.com/renownedd - You don't specify the OS. I run AMD/CPU/NVIDIA at the same time on Windows.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/fireice-uk/xmr-stak/issues/401#issuecomment-350561015, or mute the thread https://github.com/notifications/unsubscribe-auth/Agyu-RQ1klQ8lfEB_KgVqIUvj1PBKibDks5s_AnBgaJpZM4Q6aP2 .

[chichicuervo]

Can you share what you did? I'm having this same problem.

My crazy setup: FreeBSD 11.1-RELEASE chrooted into a linux (debian jessie) jail. nvidia-driver: 384.90, cuda: 9.0.176-384.81-linux

I'm seriously hoping it's a config thing and not minor version number.

[JerichoJones]

@chichicuervo What do you get back when you run:

nvcc --version

[axpatito]

+1 on this ticket. @renownedd pls share how you fixed.

[renownedd]

I had my nvidia on a pcie riser and forgot to connect the pcie power cord from the psu to the gpu.... After plugging it in it worked. I have had a lot of issues with it tho. I have had a lot of luck doing it as follows: use ddu to uninstall drivers. only connect one card at a time. So amd or nvidia first then install drivers. then disconnect that gpu and connect the other one install the drivers. then plug both gpus in and it works for me. I have also sometimes had to have a display connected to both gpus at the same time to get them to work properly. Otherwise programs like radeon settings wattman or geforce experience may not open because they dont detect a display connected to the gpu.

[JerichoJones]

@renownedd Thanks for this!

I have also sometimes had to have a display connected to both gpus at the same time to get them to work properly.

Dongles might help here.

[Maxattax97]

Having the same issue on OpenSUSE Tumbleweed with kernel 4.17.3-1 x64, running XMR-Stak from master branch (commit b3f79de as of now).

Output of nvidia-smi:

Mon Jul  9 21:21:13 2018       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.67                 Driver Version: 390.67                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 960     Off  | 00000000:01:00.0  On |                  N/A |
|  0%   55C    P0    29W / 160W |    672MiB /  4040MiB |     12%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      1478      G   /usr/bin/X                                   283MiB |
|    0      1800      G   /usr/bin/kwin_x11                            176MiB |
|    0      1858      G   /usr/bin/krunner                               1MiB |
|    0      1866      G   /usr/bin/plasmashell                          48MiB |
|    0      2141      G   /usr/bin/amarok                                6MiB |
|    0      2533      G   ...-token=1DFA3534D5CB04D316789FD9F95FBE9D    21MiB |
|    0      2731      G   ...ax/.local/share/Steam/ubuntu12_32/steam    75MiB |
|    0      2748      G   /usr/bin/akonadi_archivemail_agent             2MiB |
|    0      2780      G   /usr/bin/akonadi_mailfilter_agent              2MiB |
|    0      2793      G   /usr/bin/akonadi_sendlater_agent               2MiB |
|    0     25909      G   ...passed-by-fd --v8-snapshot-passed-by-fd    46MiB |
+-----------------------------------------------------------------------------+

Output of nvcc --version:

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Wed_Apr_11_23:16:29_CDT_2018
Cuda compilation tools, release 9.2, V9.2.88

CMake configuration:

export CC=/usr/bin/gcc-7
export CXX=/usr/bin/g++-7
cmake .. -DCMAKE_BUILD_TYPE=Release -DMICROHTTPD_ENABLE=ON -DOpenSSL_ENABLE=ON -DCPU_ENABLE=ON -DHWLOC_ENABLE=ON -DCUDA_ENABLE=ON -DCUDA_COMPILER=nvcc -DOpenCL_ENABLE=OFF -DCUDA_ARCH=52

Output of ldd libxmrstak_cuda_backend.so

        linux-vdso.so.1 (0x00007ffc461e4000)
        libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f97615d1000)
        libdl.so.2 => /lib64/libdl.so.2 (0x00007f97613cd000)
        librt.so.1 => /lib64/librt.so.1 (0x00007f97611c5000)
        libmicrohttpd.so.12 => /usr/lib64/libmicrohttpd.so.12 (0x00007f9760fa7000)
        libssl.so.1.1 => /usr/lib64/libssl.so.1.1 (0x00007f9760d3b000)
        libcrypto.so.1.1 => /usr/lib64/libcrypto.so.1.1 (0x00007f97608b1000)
        libhwloc.so.5 => /usr/lib64/libhwloc.so.5 (0x00007f9760671000)
        libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00007f97602e8000)
        libm.so.6 => /lib64/libm.so.6 (0x00007f975ff55000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f975fd3d000)
        libc.so.6 => /lib64/libc.so.6 (0x00007f975f97d000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f9761dee000)
        libgnutls.so.30 => /usr/lib64/libgnutls.so.30 (0x00007f975f61a000)
        libz.so.1 => /lib64/libz.so.1 (0x00007f975f403000)
        libnuma.so.1 => /usr/lib64/libnuma.so.1 (0x00007f975f1f7000)
        libpciaccess.so.0 => /usr/lib64/libpciaccess.so.0 (0x00007f975efed000)
        libxml2.so.2 => /usr/lib64/libxml2.so.2 (0x00007f975ec85000)
        libp11-kit.so.0 => /usr/lib64/libp11-kit.so.0 (0x00007f975e957000)
        libidn2.so.0 => /usr/lib64/libidn2.so.0 (0x00007f975e73a000)
        libunistring.so.2 => /usr/lib64/libunistring.so.2 (0x00007f975e3b8000)
        libtasn1.so.6 => /usr/lib64/libtasn1.so.6 (0x00007f975e1a5000)
        libnettle.so.6 => /usr/lib64/libnettle.so.6 (0x00007f975df6c000)
        libhogweed.so.4 => /usr/lib64/libhogweed.so.4 (0x00007f975dd36000)
        libgmp.so.10 => /usr/lib64/libgmp.so.10 (0x00007f975da9f000)
        liblzma.so.5 => /usr/lib64/liblzma.so.5 (0x00007f975d86b000)
        libffi.so.7 => /usr/lib64/libffi.so.7 (0x00007f975d661000)

Related issues: #953 #777 #502 #545 #928

From what I can tell, it seems like most success has come from installing and compiling with CUDA 8. I think I've tried this in the past with great failure, but I might look into it again. Is CUDA 9.x still not supported? My drivers seem to be up to date.