Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Cuda Versions

Download as txt, pdf, or txt
Download as txt, pdf, or txt
You are on page 1of 3

brrsoftwares@worker-3:~/gpu-examples$ nvidia-smi

Mon Oct 30 10:51:25 2023


+----------------------------------------------------------------------------------
-----+
| NVIDIA-SMI 535.113.01 Driver Version: 535.113.01 CUDA Version: 12.2
|
|-----------------------------------------+----------------------
+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr.
ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util
Compute M. |
| | |
MIG M. |
|
=========================================+======================+==================
====|
| 0 NVIDIA GeForce RTX 3060 On | 00000000:01:00.0 Off |
N/A |
| 0% 44C P8 20W / 170W | 2MiB / 12288MiB | 0%
Default |
| | |
N/A |
+-----------------------------------------+----------------------
+----------------------+

+----------------------------------------------------------------------------------
-----+
| Processes:
|
| GPU GI CI PID Type Process name GPU
Memory |
| ID ID Usage
|
|
===================================================================================
====|
| No running processes found
|
+----------------------------------------------------------------------------------
-----+

nvidia-smi -pm 1

To know what CUDA and cuDNN versions are compatible with your GPU 3060 for Ubuntu
22.04, you can refer to the following table:

CUDA Version cuDNN Version Supported NVIDIA Hardware Linux, Windows


12.0 8.9.5 GeForce RTX 3060 Yes, >=525.60.13, >=527.41
11.8 8.9.5 GeForce RTX 3060 Yes, >= 450.80.02, >=452.39
11.7 8.6.0 GeForce RTX 3060 Yes, >= 460.39
11.6 8.6.0 GeForce RTX 3060 Yes, >= 455.78.01
11.5 8.2.0 GeForce RTX 3060 Yes, >= 450.80.02
11.4 8.2.0 GeForce RTX 3060 Yes, >= 440.100
11.3 8.0.5 GeForce RTX 3060 Yes, >= 440.100
11.2 8.0.5 GeForce RTX 3060 Yes, >= 440.100
11.1 7.6.5 GeForce RTX 3060 Yes, >= 440.100
11.0 7.6.5 GeForce RTX 3060 No
It is recommended to use the latest versions of CUDA and cuDNN for the best
performance. However, you may need to use an older version of CUDA or cuDNN if you
are using a specific machine learning framework or library that has specific
requirements.

To check the version of CUDA installed on your system, run the following command:

nvcc --version
To check the version of cuDNN installed on your system, run the following command:

cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR


If you do not have CUDA or cuDNN installed, you can install them using the
following commands:

sudo apt install nvidia-cuda-toolkit


sudo apt install cuda-toolkit-11-8
sudo apt install nvidia-cudnn-toolkit-11-8
Note that you may need to reboot your system after installing CUDA and cuDNN.

To download CUDA 12.0 and cuDNN 8.9.5 for GeForce RTX 3060 on Ubuntu 22.04, you can
follow these steps:

Download the NVIDIA driver.

You can download the NVIDIA driver from the NVIDIA website. Make sure to select the
driver that is compatible with your GPU and operating system. For the RTX 3060 on
Ubuntu 22.04, you will need the driver version 525.60.13 or later.

Download the CUDA Toolkit.

You can download the CUDA Toolkit from the NVIDIA Developer website. Make sure to
select the Toolkit version that is compatible with your driver version. For driver
version 525.60.13 or later, you will need the CUDA Toolkit version 12.0.

Download cuDNN.

You can download cuDNN from the NVIDIA Developer website. Make sure to select the
cuDNN version that is compatible with your CUDA Toolkit version. For CUDA Toolkit
version 12.0, you will need cuDNN version 8.9.5.

Install the NVIDIA driver.

To install the NVIDIA driver, run the following command:

sudo apt install nvidia-driver-525


Install the CUDA Toolkit.

To install the CUDA Toolkit, run the following command:

sudo apt install cuda-toolkit-11-8


Install cuDNN.

To install cuDNN, run the following commands:

sudo cp cudnn-linux-x86_64-8.9.5.25_cuda12-archive.tar.xz /tmp


cd /tmp
sudo tar xvf cudnn-linux-x86_64-8.9.5.25_cuda12-archive.tar.xz
sudo cp cuda/include/cudnn.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
Verify the installation.

To verify that CUDA and cuDNN are installed correctly, run the following commands:

nvcc --version
cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR
You should see the following output:

nvcc: NVIDIA Compiler Driver, version 12.0.0


CUDNN_MAJOR = 8
If you see this output, then CUDA and cuDNN are installed correctly.

You might also like