Backend GPU activation troubles - C++ or JavaCPP
tomMEM
New Altair Community Member
Hello, I try to activate the GPU backend at a Windows11 AMD with RTX3060 GPU Notebook.
I installed cuda 10.1 and cudnn 7.6. The cudart64_101.dll I am not sure where to place.
deviceQuery.exe gives: CUDA Driver = CUDART, CUDA Driver Version = 11.4, CUDA Runtime Version = 10.1, NumDevs = 1, Device0 = NVIDIA GeForce RTX 3060 Laptop GPU. Result = PASS
Running the RM batch, the log indicates "Versions of org.bytedeco:javacpp:1.5.3 and org.bytedeco:cuda:10.1-7.6-1.5.2 do not match" and later cudaGetSymbolAddress(...) failed; Error code: [13].
JavaCPP1.5.3 seems not to support 10.1 - at least according to bytedeco/javacpp documentation. Also not sure if C++ version from Visual Studio19 might create some issues.
Looking for some advice.
Thank u, T
I installed cuda 10.1 and cudnn 7.6. The cudart64_101.dll I am not sure where to place.
deviceQuery.exe gives: CUDA Driver = CUDART, CUDA Driver Version = 11.4, CUDA Runtime Version = 10.1, NumDevs = 1, Device0 = NVIDIA GeForce RTX 3060 Laptop GPU. Result = PASS
Running the RM batch, the log indicates "Versions of org.bytedeco:javacpp:1.5.3 and org.bytedeco:cuda:10.1-7.6-1.5.2 do not match" and later cudaGetSymbolAddress(...) failed; Error code: [13].
JavaCPP1.5.3 seems not to support 10.1 - at least according to bytedeco/javacpp documentation. Also not sure if C++ version from Visual Studio19 might create some issues.
Looking for some advice.
Thank u, T
0
Best Answer
-
Hello, after uninstallation of Cuda10.1 and all Nvidia items, as well uninstallation of Rapidminer and removal of .javaCPP .rapidminer under users, a fresh installation of Cuda11.2 (cuda_11.2.0_460.89_win10.), Rapidminer and updated ND4j, Deeplearning extensions the GPU backend can be activated and is functional. Now only CPU backend is not functional, but I do not mind.
Thank you for the updates.
Best
T0
Answers
-
Hello, after uninstallation of Cuda10.1 and all Nvidia items, as well uninstallation of Rapidminer and removal of .javaCPP .rapidminer under users, a fresh installation of Cuda11.2 (cuda_11.2.0_460.89_win10.), Rapidminer and updated ND4j, Deeplearning extensions the GPU backend can be activated and is functional. Now only CPU backend is not functional, but I do not mind.
Thank you for the updates.
Best
T0