CUDNN version 90.5 is not compatible with 9.5

Hello,
I updated my environment to CUDA 12.6, OpenCV 4.10.0, and cuDNN 9.5.0. Now, in my Rust project, I can’t use cuDNN, and I get the following error:

Video probe: {Width: 1920px | Height: 1080px | FPS: 30}
CUDA is available: true, 1
Opencv Version: 4.10.0
Model format: darknet

[ WARN:0@28.535] global init.hpp:32 checkVersions cuDNN reports version 90.5 which is not compatible with the version 9.5 with which OpenCV was built

Is OpenCV 4.10.0 not compatible with cuDNN 9.5.0?
My OpenCV build had no issues. What am I doing wrong?

I’ll leave the details of my machine configuration.

Nvida

$ nvidia-smi
Wed Oct 16 22:12:49 2024       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 560.35.03              Driver Version: 560.35.03      CUDA Version: 12.6     |
|-----------------------------------------+------------------------+----------------------+
| 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        Off |   00000000:05:00.0 Off |                  N/A |
|  0%   47C    P8             14W /  170W |    1248MiB /  12288MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A     10642      C   python3.8                                    1114MiB |
|    0   N/A  N/A     12349      G   /usr/lib/xorg/Xorg                              4MiB |
|    0   N/A  N/A   1260046      C   target/debug/app                              106MiB |
+-----------------------------------------------------------------------------------------+

nvcc

$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Thu_Sep_12_02:18:05_PDT_2024
Cuda compilation tools, release 12.6, V12.6.77
Build cuda_12.6.r12.6/compiler.34841621_0

cudnn_version.h

$ cat /usr/include/cudnn_version.h | grep CUDNN_MAJOR -A 2
#define CUDNN_MAJOR 9
#define CUDNN_MINOR 5
#define CUDNN_PATCHLEVEL 0
--
#define CUDNN_VERSION (CUDNN_MAJOR * 10000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

bug resolved

1 Like