torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 11.00 GiB total capacity; 10.15 GiB already allocated; 0 bytes free; 10.33 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
From looking at this issue it seems like the model needs ~16gb VRAM to run, my GTX 1080 ti only has 11gb with very little else consuming it.
Sun Oct 8 13:38:46 2023
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 530.50 Driver Version: 531.79 CUDA Version: 12.1 |
|-----------------------------------------+----------------------+----------------------+
| 0 NVIDIA GeForce GTX 1080 Ti On | 00000000:01:00.0 On | N/A |
| 19% 58C P0 65W / 280W| 644MiB / 11264MiB | 0% Default |
As suggested when googling the issue, I have tried reducing the batch size.
python gligen_inference.py --batch_size 1
But I am still running out of memory, does anyone have any ideas or is what I am attempting not possible?
From looking at this issue it seems like the model needs ~16gb VRAM to run, my GTX 1080 ti only has 11gb with very little else consuming it.
As suggested when googling the issue, I have tried reducing the batch size.
python gligen_inference.py --batch_size 1
But I am still running out of memory, does anyone have any ideas or is what I am attempting not possible?