error when run cmd "python -m vsrife"
Usulyre opened this issue · 24 comments
Got this error when run cmd "python -m vsrife"
C:\Windows\System32>python -m vsrife
Traceback (most recent call last):
File "", line 189, in run_module_as_main
File "", line 148, in get_module_details
File "", line 112, in get_module_details
File "C:\Program Files\Python311\Lib\site-packages\vsrife_init.py", line 13, in
from torch_tensorrt.fx import compile
File "C:\Program Files\Python311\Lib\site-packages\torch_tensorrt\fx_init.py", line 1, in
from .converters import * # noqa: F403 F401
^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\Python311\Lib\site-packages\torch_tensorrt\fx\converters_init.py", line 2, in
import tensorrt as trt
ModuleNotFoundError: No module named 'tensorrt'
C:\Windows\System32>
Also, where do I put the files i downloaded from after i uncompress them:
https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html
will tell you how to install the TensorRT.
Pretty much download the zip,
extract it,
set system environment PATH to the folder you just extracted or CUDA{version}/Bin if missing,
move DLL files from LIB to CUDA/{Version}/Bin,
and run a couple PIP installs. Goto Extracted folder/python and run
python.exe -m pip install tensorrt-*-cp3x-none-win_amd64.whl
After all that you need to run a couple PIP installs, these can be found in the extracted folder of TensorRT you just downloaded from nvidia.
python3 -m pip install \graphsurgeon\graphsurgeon-0.4.6-py2.py3-none-any.whl
python3 -m pip install \uff\uff-0.6.9-py2.py3-none-any.whl
python3 -m pip install \onnx_graphsurgeon\onnx_graphsurgeon-0.3.12-py2.py3-none-any.whl
Good luck!
https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html will tell you how to install the TensorRT.
Pretty much download the zip, extract it, set system environment PATH to the folder you just extracted or CUDA{version}/Bin if missing, move DLL files from LIB to CUDA/{Version}/Bin, and run a couple PIP installs. Goto Extracted folder/python and run python.exe -m pip install tensorrt-*-cp3x-none-win_amd64.whl
After all that you need to run a couple PIP installs, these can be found in the extracted folder of TensorRT you just downloaded from nvidia. python3 -m pip install \graphsurgeon\graphsurgeon-0.4.6-py2.py3-none-any.whl python3 -m pip install \uff\uff-0.6.9-py2.py3-none-any.whl python3 -m pip install \onnx_graphsurgeon\onnx_graphsurgeon-0.3.12-py2.py3-none-any.whl
Good luck!
Hi and thanks for your reply.
I got an error message when I ran that first cmdline. I have a GTX 1660.
ERROR: tensorrt-8.5.2.2-cp310-none-win_amd64.whl is not a supported wheel on this platform.
Hi and thanks for your reply.
I got an error message when I ran that first cmdline. I have a GTX 1660.
ERROR: tensorrt-8.5.2.2-cp310-none-win_amd64.whl is not a supported wheel on this platform.
Make sure you are using python 3.10 if you are typing 310. Also try updating PIP to newest version and make sure you have python 64 bit version.
More info on your possible problem. https://stackoverflow.com/a/39945920
Hi and thanks for your reply.
I got an error message when I ran that first cmdline. I have a GTX 1660.
ERROR: tensorrt-8.5.2.2-cp310-none-win_amd64.whl is not a supported wheel on this platform.Make sure you are using python 3.10 if you are typing 310. Also try updating PIP to newest version and make sure you have python 64 bit version.
More info on your possible problem. https://stackoverflow.com/a/39945920
I have python 64 bit 3.11
Does that matter?
If you have 3.11, you need to type 3.11 when typing the command
tensorrt-8.5.2.2-cp310-none-win_amd64.whl
should be
tensorrt-8.5.2.2-cp311-none-win_amd64.whl
If you have 3.11, you need to type 3.11 when typing the command
tensorrt-8.5.2.2-cp310-none-win_amd64.whl should be tensorrt-8.5.2.2-cp311-none-win_amd64.whl
when i run "python -m vsrife"
\Program Files\Python310\lib\site-packages\torch_tensorrt\fx\tracer\acc_tracer\acc_ops.py:840: UserWarning: Unable to import torchvision related libraries.: No module named 'torchvision'. Please install torchvision lib in order to lower stochastic_depth
when i run the script i get "RIFE: cuda is not available"
What does it say when you type nvcc --version. Should say if you have CUDA installed or not.
For Please install torchvision lib in order to lower stochastic_depth
Try pip install torchvision
Not trying to be mean, but if you can't figure out these simple problems your gonna have a bad time trying to run the more complicated tasks. Kinda seems like you don't even have CUDA install from nvidia even.
What does it say when you type nvcc --version. Should say if you have CUDA installed or not.
For Please install torchvision lib in order to lower stochastic_depth
Try pip install torchvision
Not trying to be mean, but if you can't figure out these simple problems your gonna have a bad time trying to run the more complicated tasks. Kinda seems like you don't even have CUDA install from nvidia even.
nvcc-version doesn't work as is?
I tried to follow this guide:
https://www.svp-team.com/forum/viewtopic.php?pid=81456#p81456
I still had the same issue with "RIFE: cuda is not available"
You need to download and install Nvidia Cuda toolkit. You might need to make a account with them first.
https://developer.nvidia.com/cuda-downloads
When installed you should be able to run and see this output.
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:41:10_Pacific_Daylight_Time_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0
Find a guide to install CUDA and make sure to add them to your system PATH or it wont find it when you try to run them. Also if you get stuck , try asking CHAT-GPT about how to install it. Good Luck!
I thought everything is included for the windows release:
"For ease of installation on Windows, you can download the 7z file on Releases which contains required runtime libraries and Python wheel file"
I thought everything is included for the windows release:
"For ease of installation on Windows, you can download the 7z file on Releases which contains required runtime libraries and Python wheel file"
NVIDIA CUDA is a separate thing and needs to be installed for many AI programs to work.
I thought everything is included for the windows release:
"For ease of installation on Windows, you can download the 7z file on Releases which contains required runtime libraries and Python wheel file"NVIDIA CUDA is a separate thing and needs to be installed for many AI programs to work.
Would cuda for python work?
pip install cuda-python
Would cuda for python work?
pip install cuda-python
Not sure, I installed mine with the package from NVIDIA website.
Would cuda for python work?
pip install cuda-pythonNot sure, I installed mine with the package from NVIDIA website.
Downloaded cudatoolkit v11.7
Not sure what from the custom install selection that I need keep check for cuda to be available for vs-rife.
I just selected cuda runtime libraries and nvcc and added to the path
Still not working.
When you run nvcc --version
Do you see?
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:41:10_Pacific_Daylight_Time_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0
If it says something like command not found, it is most likely not set in the path correctly.
https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#post-installation-actions
When you run nvcc --version
Do you see?
nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2022 NVIDIA Corporation Built on Wed_Sep_21_10:41:10_Pacific_Daylight_Time_2022 Cuda compilation tools, release 11.8, V11.8.89 Build cuda_11.8.r11.8/compiler.31833905_0
If it says something like command not found, it is most likely not set in the path correctly.
https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#post-installation-actions
I have to go to the actual folder where nvcc is to run that command
Ya, that means you did not add it to the PATH correctly. When it is added to the path it tells the system where to find NVCC when you run it.
step 5.3: Confirming that CUDA environment variables are set in Windows
For CUDA to function properly, you will need to ensure that CUDA environment variables are set in your PC‘s Path. To confirm, you will need to check the Environment Variables. This can be found by following these steps
Control Panel -> System and Security -> System-> Advanced System settings.
This will open up a window, click on the Advanced tab. as shown below
Image by author
https://miro.medium.com/v2/resize:fit:640/format:webp/1*a9zoQnDYS4vF8rofF7levw.png
https://miro.medium.com/v2/resize:fit:640/format:webp/1*7WOGbl9LAE4hoSjV-9Wehg.png
These paths are automatically added, but in case you can’t find them. Just type in C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\bin and C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\V11.2\libnvvp.
step 5.3: Confirming that CUDA environment variables are set in Windows
For CUDA to function properly, you will need to ensure that CUDA environment variables are set in your PC‘s Path. To confirm, you will need to check the Environment Variables. This can be found by following these steps
Control Panel -> System and Security -> System-> Advanced System settings.
This will open up a window, click on the Advanced tab. as shown below Image by author
https://miro.medium.com/v2/resize:fit:640/format:webp/1*a9zoQnDYS4vF8rofF7levw.png
https://miro.medium.com/v2/resize:fit:640/format:webp/1*7WOGbl9LAE4hoSjV-9Wehg.png
These paths are automatically added, but in case you can’t find them. Just type in C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\bin and C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\V11.2\libnvvp.
I dont know what libnvvp is for which i do not currently have.
nvcc works now but still getting the error of cuda not available.
By the way, this is part of the error when one looks at init.py of vsrife
if not torch.cuda.is_available():
raise vs.Error('RIFE: CUDA is not available')
Since I have 11.8 CUDA installed, mine is located at...
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\libnvvp
You can try adding these to the PATH if you installed them to the same location.
Also you can check them folders to see if that is where they are.
Since I have 11.8 CUDA installed, mine is located at...
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\libnvvp
You can try adding these to the PATH if you installed them to the same location. Also you can check them folders to see if that is where they are.
By the way, this is part of the error when one looks at init.py of vsrife
if not torch.cuda.is_available():
raise vs.Error('RIFE: CUDA is not available')
You need to run nvcc --version. It will tell you if its installed or not.
You need to run nvcc --version. It will tell you if its installed or not.
C:\Windows\System32>nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_May__3_19:00:59_Pacific_Daylight_Time_2022
Cuda compilation tools, release 11.7, V11.7.64
Build cuda_11.7.r11.7/compiler.31294372_0
Great, so that is installed correctly. The error you are getting I have no experience with. So looks like my help ends here. I wish you luck!