I’ve installed NVIDIA Driver 460.56 and CUDA 11.1 for machine learning. But there was a driver breakdown when I run my code for the first time. I think it’s because that CUDA 11.1 is not stable. So should I choose 11.0 or 11.1?
The question title and the question body have a mismatch. RTX 3060 uses the Ampere architecture, which requires CUDA 11.x. CUDA 11.2 in particular fixed various issues discovered in earlier versions of CUDA 11.x, so unless other requirements preclude you from using it, I would suggest installing that.
I have not idea what you mean by “there was a driver breakdown”. You would want to use the latest production driver package available for your hardware. The NVIDIA download page should point you to the appropriate one if once specify your GPU. I believe 460 family drivers are still up-to-date. Personally I am running with 461.40 in Win10 right now.
it seems that RTX3060 desktop is not support 461.40 ，notebook suport ，desktop supports 461.72
1 Like
also wait for 3060 support cuda11 sdk
Now sure what you are waiting for. The latest CUDA version certainly supports the RTX 3060:
As for drivers for Windows 10, the NVIDIA download page at present directs to version 461.92 (dated March 16, 2021) for the Desktop RTX 3060 and version 461.72 (dated February 25, 2021) for the Laptop RTX 3060.
Hi,
I have following below specs of my laptop can anyone help me to redirect which cudnn and cuda toolkit will work, I am tired of this installation, i tried with 11 and 12 version but nothing worked as gpu is still not getting detected.
Laptop specs:
os: windows 11
GPU: RTX 3060 (16 GB)
Sorry for this naive question, very new to Data science.
Can you make sure, how Cuda&CuDNN install in RTX360 on Ubuntu Server 20.04?
I want the exact type of version that can’t generate errors during and after completion. OR when I would run Pytorch, that system shouldn’t give the Uncompatible Version issues.
Build the procedure of setup installation pipeline.
NVIDIA RTX Virtual Workstation (vWS) on CSP Market CUDA Programming and Performance CUDA Setup and Installation Linux