Unabled cuda with unminable in config.json and screen not responding.

[gkapa12]

I am using xmrig-6.15.2 for mining with nvidia-gpu. Installed cude and all dependencies and enabled cude in config.json. but screen just not responding. when I mine on cpu, it works fine. below i have attached screenshot.

system: Ubuntu 18

after this line, screen not responding and gpu load is also almost 0%.

[Spudz76]

I would expect problems using less than toolkit 10.2 for anything newer than Pascal

If windows get the 11_4 dll

If Linux get the newest CUDA toolkit and rebuild against that instead of 10.0

[sunbearc22]

I encountered a similar issue even after compiling xmrig-cuda with CUDA 11.4 in Ubuntu 20.04. How do I overcome this issue?

* ABOUT        XMRig/6.16.2 gcc/9.4.0
* LIBS         libuv/1.34.2 OpenSSL/1.1.1f hwloc/2.1.0
   ...
 * CUDA         11.4/11.4/6.15.1
 * NVML         11.470.86/470.86 press e for health report
 * CUDA GPU     #0 65:00.0 NVIDIA GeForce GTX TITAN 875/3004 MHz smx:14 arch:35 mem:5608/6075 MB

Following instructions in https://xmrig.com/docs/miner/build/ubuntu, I had run the Basic Build commands for xmrig:

1. sudo apt-get install git build-essential cmake libuv1-dev libssl-dev libhwloc-dev
2. git clone https://github.com/xmrig/xmrig.git
3. mkdir xmrig/build && cd xmrig/build
4. cmake ..
5. make -j$(nproc)

and for CUDA:

1. git clone https://github.com/xmrig/xmrig-cuda.git
2. mkdir xmrig-cuda/build && cd xmrig-cuda/build
3. cmake .. -DCUDA_LIB=/usr/local/cuda11.4/lib64/stubs/libcuda.so -DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda11.4
4. make -j$(nproc)

In the config.json file,

    "cuda": {
        "enabled": true,
        "loader": "/home/user/test/xmrig-cuda/build/libxmrig-cuda.so",
        "nvml": true,
        "astrobwt": [
            {
                "index": 0,
                "threads": 32,
                "blocks": 8,
                "bfactor": 0,
                "bsleep": 0,
                "affinity": -1
            }
        ],

In directory /home/user/test/xmrig/build/, I ran this cmd $ ./xmrig -c config.json

[sunbearc22]

It compiled with some issues... Are they the cause for the nothing happens issue that I previously mentioned?

$ make -j$(nproc)
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/RandomX/monero/xmrig-cu_generated_randomx_monero.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/RandomX/arqma/xmrig-cu_generated_randomx_arqma.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/RandomX/keva/xmrig-cu_generated_randomx_keva.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/KawPow/raven/xmrig-cu_generated_KawPow.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/xmrig-cu_generated_cuda_core.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/xmrig-cu_generated_cuda_extra.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/RandomX/wownero/xmrig-cu_generated_randomx_wownero.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/RandomX/xmrig-cu_generated_randomx.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/AstroBWT/dero/xmrig-cu_generated_AstroBWT.cu.o
[ 58%] Building NVCC (Device) object CMakeFiles/xmrig-cu.dir/src/RandomX/graft/xmrig-cu_generated_randomx_graft.cu.o
/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/RandomX/randomx_cuda.hpp(319): warning: pointless comparison of unsigned integer with zero

/home/user/test/xmrig-cuda/src/cuda_extra.cu: In function ‘void cryptonight_extra_cpu_prepare(nvid_ctx*, uint32_t, const xmrig_cuda::Algorithm&)’:
/home/user/test/xmrig-cuda/src/cuda_extra.cu:415:170: warning: passing argument 5 to restrict-qualified parameter aliases with argument 6 [-Wrestrict]
  415 |         CUDA_CHECK_KERNEL(ctx->device_id, cryptonight_extra_gpu_prepare<Algorithm::CN_0><<<grid, block >>>(wsize, ctx->d_input, ctx->inputlen, startNonce,
      |                                                                                                                                                                          ^                            
/home/user/test/xmrig-cuda/src/cuda_extra.cu: In function ‘int cuda_get_deviceinfo(nvid_ctx*)’:
/home/user/test/xmrig-cuda/src/cuda_extra.cu:571:26: warning: comparison of integer expressions of different signedness: ‘int’ and ‘const size_t’ {aka ‘const long unsigned int’} [-Wsign-compare]
  571 |         if (ctx->device_blocks > max_blocks) {
      |         ~~~~~~~~~~~~~~~~~^~~~~~~~~~~~
Scanning dependencies of target xmrig-cu
[ 64%] Building CXX object CMakeFiles/xmrig-cu.dir/src/KawPow/raven/CudaKawPow_gen.cpp.o
[ 70%] Linking CXX static library libxmrig-cu.a
[ 70%] Built target xmrig-cu
Scanning dependencies of target xmrig-cuda
[ 94%] Building C object CMakeFiles/xmrig-cuda.dir/src/crypto/cn/c_blake256.c.o
[ 94%] Building CXX object CMakeFiles/xmrig-cuda.dir/src/xmrig-cuda.cpp.o
[ 94%] Building CXX object CMakeFiles/xmrig-cuda.dir/src/crypto/common/Algorithm.cpp.o
[ 94%] Building CXX object CMakeFiles/xmrig-cuda.dir/src/CudaCryptonightR_gen.cpp.o
[100%] Linking CXX shared library libxmrig-cuda.so
[100%] Built target xmrig-cuda

[Spudz76]

No, those are normal. I think I submitted patches for some of the warnings but merges haven't happened here in a long time.

[sunbearc22]

@Spudz76 In this case, can you explain why XMRIG hangs upon using GPU? I am quite certain the GPU is not fired up. Mining via CPU works. How do I troubleshoot this issue?

[Spudz76]

I believe astrobwt vs Kepler (arch 30, 32, 35, 37...) has problems in general.

See also #129 where I think I fixed it by updating the bundled CUB files. The related merge request at #130 seemed to fix it but like most Pull Requests over here it's been awaiting merge for months.

[sunbearc22]

Thanks. For now, it seems xmrig is stable for mining with CPU but not with GPU. Hope things get resolved here. Meanwhile, I have managed to use Pheonix miner in place of xmrig miner when GPU is involved.

[Spudz76]

I have found T-Rex to be superior (in stability at least) plus it only supports CUDA so they aren't wasting focus on OpenCL.