Cuda Out of Memory
#23
by
xings19
- opened
my code:
pipeline = StableDiffusionUpscalePipeline.from_pretrained(model_id,revision="fp16",torch_dtype=torch.float16)
pipeline = pipeline.to("cuda")
upscaled_image = pipeline(prompt=prompt, image=in_img).images[0]
Error:
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.50 GiB (GPU 0; 23.69 GiB total capacity; 3.62 GiB already allocated; 18.78 GiB free; 3.92 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
in_img size is (320,768)
If you have low GPU RAM available, make sure to add a pipe.enable_attention_slicing() after sending it to cuda for less VRAM usage (to the cost of speed)