Video generation stops due to CUDA error: misaligned address

Using the 1-click installer package on Windows 11. No modifications to the installation. Running run.bat via Powershell as administrator. Browser is Edge.

Generating a video using default settings. Working well up until 2.3 seconds generated, then stopped processing, and the preview video disappeared. The following stacktrace appeared in Powershell. There is no other feedback in the browser to indicate a problem - video generation just "hangs" indefinitely.

```
Traceback (most recent call last):
File "F:\framepack_cu126_torch26\webui\demo_gradio.py", line 290, in worker
current_pixels = vae_decode(real_history_latents[:, :, :section_latent_frames], vae).cpu()
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\utils\_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "F:\framepack_cu126_torch26\webui\diffusers_helper\hunyuan.py", line 98, in vae_decode
image = vae.decode(latents.to(device=vae.device, dtype=vae.dtype)).sample
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\utils\accelerate_utils.py", line 46, in wrapper
return method(self, *args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 868, in decode
decoded = self._decode(z).sample
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 836, in _decode
return self._temporal_tiled_decode(z, return_dict=return_dict)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 1052, in _temporal_tiled_decode
decoded = self.tiled_decode(tile, return_dict=True).sample
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 984, in tiled_decode
decoded = self.decoder(tile)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 618, in forward
hidden_states = up_block(hidden_states)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 404, in forward
hidden_states = resnet(hidden_states)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 176, in forward
hidden_states = self.conv1(hidden_states)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\diffusers\models\autoencoders\autoencoder_kl_hunyuan_video.py", line 79, in forward
return self.conv(hidden_states)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\conv.py", line 725, in forward
return self._conv_forward(input, self.weight, self.bias)
File "F:\framepack_cu126_torch26\system\python\lib\site-packages\torch\nn\modules\conv.py", line 720, in _conv_forward
return F.conv3d(
RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED
Error in listener thread: CUDA error: misaligned address
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

```

Start at webui/demo_gradio.py in worker and webui/diffusers_helper/hunyuan.py in vae_decode, then follow the reported diffusers and PyTorch Hunyuan video VAE stack. Reproduce with run.bat on the default video-generation settings and investigate the cuDNN and CUDA misaligned-address errors; done means generation no longer hangs or stops at roughly 2.3 seconds.

Written by the indexing model from the issue text.