Open ahmedaboghaly opened 5 years ago
Old Fermi core, requires CUDA 8, this means you must compile your own with CUDA Toolkit 8.0GA2 which also means you must have MSVC 14.00 (older legacy version too) but that can be installed by the Individual Packages section of MSVS2017 Installer.
Release is no longer compiled for CUDA 8 since around when RTX cards / CUDA 10 came out, therefore how it doesn't work.
I am using xmr-stak-v2.10.4-win64 - Windows 10 but it's can not mining by GPU, it's using the CPU only
My GPU type is Nvidia Quadro k6000 y cuda 10.1 and I have installed Visual Studio in its latest version.
I can get visual studio 2017 to compile 2.10.4 and work with CUDA 8 cards by following the compilation instructions. This also works for Quadro 2000 and other Fermi based cards like the Tesla M2050 and GTX 580 and the K6000 .
xmr-stak 2.10.5 b846a6f7c
Brought to you by fireice_uk and psychocrypt under GPLv3. Based on CPU mining code by wolf9466 (heavily optimized by fireice_uk). Based on NVIDIA mining code by KlausT and psychocrypt. Based on OpenCL mining code by wolf9466.
Configurable dev donation level is set to 0.0%
Upcoming xmr-stak-gui is sponsored by:
|___/ https://ryo-currency.com |___/
[2019-07-02 00:19:29] : Mining coin: cryptonight_turtle [2019-07-02 00:19:29] : Compiling code and initializing GPUs. This will take a while... [2019-07-02 00:19:30] : Device 0 work size 8 / 32. [2019-07-02 00:19:30] : OpenCL device 0 - Load precompiled code from file C:\Users\arg5\AppData\Local/.openclcache/7de8eac8a1602582946f826b032febefd78de7c51208b1ac1abeefbbc48aff07.openclbin [2019-07-02 00:19:30] : Device 0 work size 8 / 32. [2019-07-02 00:19:31] : OpenCL device 0 - Load precompiled code from file C:\Users\arg5\AppData\Local/.openclcache/7de8eac8a1602582946f826b032febefd78de7c51208b1ac1abeefbbc48aff07.openclbin [2019-07-02 00:19:31] : Starting AMD GPU (OpenCL) thread 0, no affinity. [2019-07-02 00:19:31] : Starting AMD GPU (OpenCL) thread 1, no affinity. [2019-07-02 00:19:31] : NVIDIA: try to load library 'xmrstak_cuda_backend' NVIDIA: found 1 potential device's [2019-07-02 00:19:31] : Starting NVIDIA GPU thread 0, no affinity. CUDA [9.1/8.0] GPU#0, device architecture 20: "GeForce GTX 580"... device init succeeded [2019-07-02 00:19:31] : NVIDIA: use library 'xmrstak_cuda_backend' [2019-07-02 00:19:31] : Starting 2x thread, affinity: 1. [2019-07-02 00:19:31] : hwloc: memory pinned [2019-07-02 00:19:31] : Starting 2x thread, affinity: 2. [2019-07-02 00:19:31] : hwloc: memory pinned [2019-07-02 00:19:31] : Starting 2x thread, affinity: 3. [2019-07-02 00:19:31] : hwloc: memory pinned [2019-07-02 00:19:31] : Fast-connecting to arqma.herominers.com:10321 pool ... [2019-07-02 00:19:31] : Pool arqma.herominers.com:10321 connected. Logging in... [2019-07-02 00:19:31] : Difficulty changed. Now: 240009. [2019-07-02 00:19:31] : Pool logged in. [2019-07-02 00:19:31] : Switch to assembler version for 'intel_avx' cpu's [2019-07-02 00:19:31] : Switch to assembler version for 'intel_avx' cpu's [2019-07-02 00:19:31] : Switch to assembler version for 'intel_avx' cpu's [CUDA] Error gpu 0: <C:/Users/arg5/xmr-stak-2.10.5/xmrstak/backend/nvidia/nvcc_code/cuda_core.cu>:909
Then it drops to command prompt or exits.
I specifically compiled this version with arch=20 in the cmake options. As a matter of fact, I used the exact same options when I built the 2.10.5 version as when I did the 2.10.4 version. cmake .. -DCUDA_ARCH=20 -DCMAKE_LINK_STATIC=ON -DMICROHTTPD_ENABLE=OFF -DCPU_ENABLE=ON -DOpenCL_ENABLE=ON -DXMR-STAK_LARGEGRID=OFF -DXMR-STAK_THREADS=64 -DOpenSSL_ENABLE=OFF -DHWLOC_ENABLE=ON
So something changed with the release of VS 2019 or the code in cuda_core.cu ?
fixed in #2467
I am using xmr-stak-v2.10.4-win64 but it's can not mining by GPU, it's using the CPU only
NOT : GPU type is Nvidia Quadro 2000