Description
I don't know what you guys did with this update. This used to work for me. Windows 11, 5800X 96GB DDR, 4090 24GB VRAM. It used to work fine. Now it's totally broken.
F:\AI\LatentSync\latentsync\models\attention.py:273: UserWarning: CuDNN attention has been runtime disabled. (Triggered internally at C:\actions-runner_work\pytorch\pytorch\pytorch\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:578.)
hidden_states = F.scaled_dot_product_attention(query, key, value, attn_mask=attention_mask)
Sample frames: 16: 0%| | 0/20 [00:00<?, ?it/s]
Doing inference...: 0%| | 0/15 [00:01<?, ?it/s]
Error during processing: No available kernel. Aborting execution.
Traceback (most recent call last):
File "F:\AI\LatentSync\gradio_app.py", line 45, in process_video
result = main(
File "F:\AI\LatentSync\scripts\inference.py", line 82, in main
pipeline(
File "F:\AI\LatentSync\env\lib\site-packages\torch\utils_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "F:\AI\LatentSync\latentsync\pipelines\lipsync_pipeline.py", line 413, in call
noise_pred = self.denoising_unet(
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "F:\AI\LatentSync\latentsync\models\unet.py", line 401, in forward
sample, res_samples = downsample_block(
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "F:\AI\LatentSync\latentsync\models\unet_blocks.py", line 394, in forward
hidden_states = attn(hidden_states, encoder_hidden_states=encoder_hidden_states).sample
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "F:\AI\LatentSync\latentsync\models\attention.py", line 104, in forward
hidden_states = block(
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "F:\AI\LatentSync\latentsync\models\attention.py", line 182, in forward
hidden_states = self.attn1(norm_hidden_states, attention_mask=attention_mask) + hidden_states
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "F:\AI\LatentSync\env\lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "F:\AI\LatentSync\latentsync\models\attention.py", line 273, in forward
hidden_states = F.scaled_dot_product_attention(query, key, value, attn_mask=attention_mask)
RuntimeError: No available kernel. Aborting execution.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "F:\AI\LatentSync\env\lib\site-packages\gradio\queueing.py", line 625, in process_events
response = await route_utils.call_process_api(
File "F:\AI\LatentSync\env\lib\site-packages\gradio\route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
File "F:\AI\LatentSync\env\lib\site-packages\gradio\blocks.py", line 2220, in process_api
result = await self.call_function(
File "F:\AI\LatentSync\env\lib\site-packages\gradio\blocks.py", line 1731, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
File "F:\AI\LatentSync\env\lib\site-packages\anyio\to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
File "F:\AI\LatentSync\env\lib\site-packages\anyio_backends_asyncio.py", line 2470, in run_sync_in_worker_thread
return await future
File "F:\AI\LatentSync\env\lib\site-packages\anyio_backends_asyncio.py", line 967, in run
result = context.run(func, *args)
File "F:\AI\LatentSync\env\lib\site-packages\gradio\utils.py", line 894, in wrapper
response = f(*args, **kwargs)
File "F:\AI\LatentSync\gradio_app.py", line 53, in process_video
raise gr.Error(f"Error during processing: {str(e)}")
gradio.exceptions.Error: 'Error during processing: No available kernel. Aborting execution.'