Could not load symbol cublasGetSmCountTarget from cublas64_11.dll. Error code 127

[xor0x]

Hi to all, I have the same problem here. Windows 11, python 3.10, and CUDA 11.3

d:\TalgilProjects\Console-Detection\venv\lib\site-packages\torchvision\models_utils.py:252: UserWarning: Accessing the model URLs via the internal dictionary of the module is deprecated since 0.13 and will be removed in 0.15. Please access them via the appropriate Weights Enum instead. warnings.warn( d:\TalgilProjects\Console-Detection\venv\lib\site-packages\torchvision\models_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead. warnings.warn( d:\TalgilProjects\Console-Detection\venv\lib\site-packages\torchvision\models_utils.py:223: UserWarning: Arguments other than a weight enum or None for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing weights=None. warnings.warn(msg) Could not load symbol cublasGetSmCountTarget from cublas64_11.dll. Error code 127

[rkcosmos]

can you try the last version (1.6.2)?

[ZZQ908268856]

Which toolkit version should be updated?

[Sharrnah]

Getting also the error Could not load symbol cublasGetSmCountTarget from cublas64_11.dll. Error code 127

Windows 10, python 3.10, CUDA 11.8

no User Warnings. but i might be missing a verbose option..

recognition seems to work. Not sure if it is using GPU though.

[IbenBru]

Have you ever found a solution for this?

[lacorreia65]

I'm also getting the same error "Could not load symbol cublasGetSmCountTarget from cublas64_11.dll. Error code 127"

I am not sure how critical is it though.

Using Windows 10, Python 3.10.9, CUDAToolkit 11.3, torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1

[liye-yang]

I'm also getting the same error in another work, but the training seems to still work?

[Hepta-Col]

Hi, I am also getting the same error message. Did anyone solve this?

[rmast]

Seen the same, with a fresh built torch 1.9.1. I'll continue using newer torch-versions.

[magicse]

I'm also getting the same error "Could not load symbol cublasGetSmCountTarget from cublas64_11.dll. Error code 127"

I am not sure how critical is it though.

Using Windows 10, Python 3.10.9, CUDAToolkit 11.3, torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1

https://github.com/NVIDIA-AI-IOT/torch2trt/issues/698#issuecomment-1681603715