Dain-App 1.0 [Nvidia Only]

Animation Interpolation powered by AI · By GRisk

Add to collection

tool Community Devlog

Dain-App 1.0 [Nvidia Only] community

CUDA call failed

A topic by cantcatto created Jan 20, 2020 Views: 6,536 Replies: 18

Viewing posts 1 to 10

cantcatto4 years ago

I keep getting this error whenever I try to render something.

Traceback (most recent call last):
File "my_design.py", line 79, in render
File "my_DAIN_class.py", line 469, in RenderVideo
File "my_DAIN_class.py", line 202, in interpolate
File "site-packages\torch\nn\modules\module.py", line 541, in __call__
File "networks\DAIN_slowmotion.py", line 148, in forward
File "networks\DAIN_slowmotion.py", line 212, in forward_flownets
File "site-packages\torch\nn\modules\module.py", line 541, in __call__
File "PWCNet\PWCNet.py", line 220, in forward
File "site-packages\torch\nn\modules\module.py", line 541, in __call__
File "PWCNet\correlation_package_pytorch1_0\correlation.py", line 59, in forward
File "PWCNet\correlation_package_pytorch1_0\correlation.py", line 27, in forward
RuntimeError: CUDA call failed (correlation_forward_cuda at correlation_cuda.cc:80)
(no backtrace available)

My GPU supports CUDA, and I have 8gb of ram so its definitely not a memory issue,

Any solutions?

GRiskDeveloper4 years ago

Hey there, what graphic card you have?

William Tumeo4 years ago

Same error here.
My gpu is low end tho: Nvidia GeForce MX110.

Framerate Index: 0

C:\Users\tumeo\Downloads\DAIN_APP Alpha\torch\nn\functional.py:2494: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details.
"See the documentation of nn.Upsample for details.".format(mode))
..\torch\csrc\autograd\python_function.cpp:622: UserWarning: Legacy autograd function with non-static forward method is deprecated and will be removed in 1.3. Please use new-style autograd function with static forward method. (Example: https://pytorch.org/docs/stable/autograd.html#torch.autograd.Function)
error in correlation_forward_cuda_kernel: no kernel image is available for execution on the device
Traceback (most recent call last):
File "my_design.py", line 79, in render
File "my_DAIN_class.py", line 469, in RenderVideo
File "my_DAIN_class.py", line 202, in interpolate
File "site-packages\torch\nn\modules\module.py", line 541, in __call__
File "networks\DAIN_slowmotion.py", line 148, in forward
File "networks\DAIN_slowmotion.py", line 212, in forward_flownets
File "site-packages\torch\nn\modules\module.py", line 541, in __call__
File "PWCNet\PWCNet.py", line 220, in forward
File "site-packages\torch\nn\modules\module.py", line 541, in __call__
File "PWCNet\correlation_package_pytorch1_0\correlation.py", line 59, in forward
File "PWCNet\correlation_package_pytorch1_0\correlation.py", line 27, in forward
RuntimeError: CUDA call failed (correlation_forward_cuda at correlation_cuda.cc:80)
(no backtrace available)

Chippo4 years ago

Same exact error as you guys, and I'm scratching my head as to what the issue is!
My 940MX is compatible with CUDA and has the proper NVIDIA toolkit installed.

Help would be greatly appreciated!

GRiskDeveloper4 years ago(+1)

So far this is the bug that most people have reported. I'm still trying to find a fix for it. I'm not sure if i'm will be able to fix it in the next version, but still is one of my top priorities.

If you can, please post your operation system and if it is 32 or 64 bits

Chippo4 years ago(+1)

Windows 10 on a 64-bit operating system.
Intel i7 7th Gen
8GB RAM

Throwing in some extra details just in case.

GRiskDeveloper4 years ago(+1)

Thanks, any extra info is helpful.

William Tumeo4 years ago(+1)

Windows 10 x64
Intel i5-8265U
8GB of RAM
Nvidia GeForce MX110 - 2GB

Fours4414 years ago (1 edit)

Windows 10 64x

AMD Ryzen 5 1600

16GB DDR4 1197.1 MHz Ram

GeForce 750 GTX Ti 2GB

cantcatto4 years ago

OS: Windows 10 Home 64-bit

CPU: Intel i5-8265U

GPU: nVidia MX130

Arkas4 years ago

OS: Windows 10 Pro 64-bit (Installed on high-speed SSD)

CPU: Intel i7-960 (First Gen i7)

GPU: Nvidia MSI 980 - 4 GB Video Memory (Factory Overclocked)

RAM: 24 GB - 1066 (1600 Underclocked for faster timings)

Antivirus: Windows Defender (No Third Party Protection installed)

Additional:

MSI Afterburner (Running and monitoring graphics card - Fails regardless if running or not)
Running GeForce Driver: 441.87
Graphics Card Default Mode Set to - Optimal Power (Nvidia Control Panel)
Chrome was open with Hardware acceleration on (Not sure if other software might interfere with run-time execution)

Hope a detailed run-down might give some insight into any possible conflict.

CnoH4 years ago (3 edits) (+1)

I changed my os to Windows10 x64, and instal the cuda 10.0 and the cudnn,

and I used the Dain-App Alpha 0.21, it works fine now.

-------

OS: Windows 7 x64

CPU: Intel i5-4440

GPU: Nvidia GeForce GTX 960

I used the Dain-App Alpha 0.2, and got this CUDA called failed

cantcatto4 years ago

Could it be that I have CUDA 10.2 installed? Should I uninstall it and install an older version? (It says on the github that it uses CUDA 9.0)

William Tumeo4 years ago

I tested it without CUDA SDK/Runtimes and with CUDA 9.2. Same error.

Emppixel4 years ago

Just happened upon this tool, HAVE NVIDIA card, HAVE CUDA, and got the exact same error

William Tumeo4 years ago

I think this is probably something related to the gpu architecture.
I managed to test it using the following architectures:
- Maxwell: doesn't work
- Turing: doesn't work
- Pascal: works

mattrified4 years ago (1 edit)

Ah. That would make sense. I believe I have Maxwell as well and am getting the same error.

NVIDIA GeForce GTX 980 Ti

Operation system -- Windows 10, 64-Bit

bitsydoge4 years ago

I tryed with a GTX 760 and got error too.

I'll try with a RTX 2070 Super tommorrow with the same computer to see if it's a GPU problem

William Tumeo4 years ago

I tried the version 0.21 and worked for Maxwell gpu (did not tested on Turing yet).

itch.io

Dain-App 1.0 [Nvidia Only]

CUDA call failed