OOM on low VRAM

[ivantan-ys]

I'm getting OOM on Low VRAM (GT1660Ti), 6GB - any solutions?

[Extraltodeus]

A better GPU would be a good one hey!

edit: no but really, it depends on what you're using. I can't really help you if that's all the informations you can bring.

[ghostsquad]

I have a 12GB 3060, using "--normalvram" and I've found using SDXL and a depth controlmap, and with AutomaticCFG enable, I run out of memory. Turn off AutomaticCFG (and close and reopen ComfyUI), and I don't get OOM errors anymore.

!!! Exception during processing!!! Allocation on device 0 would exceed allowed memory. (out of memory)
Currently allocated     : 11.26 GiB
Requested               : 31.50 MiB
Device limit            : 12.00 GiB
Free (according to CUDA): 0 bytes
PyTorch limit (set by user-supplied memory fraction)
                        : 17179869184.00 GiB
Traceback (most recent call last):
  File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute
    output_data, output_ui = get_output_data(obj, input_data_all)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 81, in get_output_data
    return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-0246\utils.py", line 381, in new_func
    res_value = old_func(*final_args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in map_node_over_list
    results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\efficiency-nodes-comfyui\efficiency_nodes.py", line 2206, in sample_adv
    return super().sample(model, noise_seed, steps, cfg, sampler_name, scheduler, positive, negative,
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\efficiency-nodes-comfyui\efficiency_nodes.py", line 713, in sample
    samples, images, gifs, preview = process_latent_image(model, seed, steps, cfg, sampler_name, scheduler,
                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\efficiency-nodes-comfyui\efficiency_nodes.py", line 537, in process_latent_image
    samples = KSamplerAdvanced().sample(model, add_noise, seed, steps, cfg, sampler_name, scheduler,
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\nodes.py", line 1383, in sample
    return common_ksampler(model, noise_seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise, disable_noise=disable_noise, start_step=start_at_step, last_step=end_at_step, force_full_denoise=force_full_denoise)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\nodes.py", line 1319, in common_ksampler
    samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-prompt-control\prompt_control\hijack.py", line 39, in pc_sample
    r = cb(orig_sampler, *args, **kwargs)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-prompt-control\prompt_control\node_lora.py", line 125, in sampler_cb
    s = orig_sampler(*args, **kwargs)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack\modules\impact\sample_error_enhancer.py", line 22, in informative_sample
    raise e
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Impact-Pack\modules\impact\sample_error_enhancer.py", line 9, in informative_sample
    return original_sample(*args, **kwargs)  # This code helps interpret error messages that occur within exceptions but does not have any impact on other operations.
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Advanced-ControlNet\adv_control\control_reference.py", line 47, in refcn_sample
    return orig_comfy_sample(model, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\sample.py", line 37, in sample
    samples = sampler.sample(noise, positive, negative, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-prompt-control\prompt_control\hijack.py", line 92, in sample
    return super().sample(
           ^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 761, in sample
    return sample(self.model, noise, positive, negative, cfg, self.device, sampler, sigmas, self.model_options, latent_image=latent_image, denoise_mask=denoise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 663, in sample
    return cfg_guider.sample(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 650, in sample
    output = self.inner_sample(noise, latent_image, device, sampler, sigmas, denoise_mask, callback, disable_pbar, seed)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 629, in inner_sample
    samples = sampler.sample(self, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 534, in sample
    samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\sampling.py", line 706, in sample_dpmpp_3m_sde_gpu
    return sample_dpmpp_3m_sde(model, x, sigmas, extra_args=extra_args, callback=callback, disable=disable, eta=eta, s_noise=s_noise, noise_sampler=noise_sampler)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\sampling.py", line 663, in sample_dpmpp_3m_sde
    denoised = model(x, sigmas[i] * s_in, **extra_args)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 272, in __call__
    out = self.inner_model(x, sigma, model_options=model_options, seed=seed)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 616, in __call__
    return self.predict_noise(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 619, in predict_noise
    return sampling_function(self.inner_model, x, timestep, self.conds.get("negative", None), self.conds.get("positive", None), self.cfg, model_options=model_options, seed=seed)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-AutomaticCFG\nodes.py", line 39, in sampling_function_patched
    out = comfy.samplers.calc_cond_batch(model, conds, x, timestep, model_options)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 192, in calc_cond_batch
    c['control'] = control.get_control(input_x, timestep_, c, len(cond_or_uncond))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Advanced-ControlNet\adv_control\utils.py", line 536, in get_control_inject
    return self.get_control_advanced(x_noisy, t, cond, batched_number)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Advanced-ControlNet\adv_control\control.py", line 32, in get_control_advanced
    return self.sliding_get_control(x_noisy, t, cond, batched_number)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Advanced-ControlNet\adv_control\control.py", line 78, in sliding_get_control
    control = self.control_model(x=x_noisy.to(dtype), hint=self.cond_hint, timesteps=timestep.float(), context=context.to(dtype), y=y)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1511, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1520, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\cldm\cldm.py", line 289, in forward
    guided_hint = self.input_hint_block(hint, emb, context)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1511, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1520, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\diffusionmodules\openaimodel.py", line 60, in forward
    return forward_timestep_embed(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\diffusionmodules\openaimodel.py", line 50, in forward_timestep_embed
    x = layer(x)
        ^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1511, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1520, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ops.py", line 66, in forward
    return super().forward(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\conv.py", line 460, in forward
    return self._conv_forward(input, self.weight, self.bias)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\conv.py", line 456, in _conv_forward
    return F.conv2d(input, weight, bias, self.stride,
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
torch.cuda.OutOfMemoryError: Allocation on device 0 would exceed allowed memory. (out of memory)
Currently allocated     : 11.26 GiB
Requested               : 31.50 MiB
Device limit            : 12.00 GiB
Free (according to CUDA): 0 bytes
PyTorch limit (set by user-supplied memory fraction)
                        : 17179869184.00 GiB

Prompt executed in 19.73 seconds

Unfortunately, it's not entirely clear that this specifically is the problem, except for the fact that disabling it means I don't OOM. I have to choose between using AutomaticCFG and using ControlNet it seems (when using SDXL). I'd expect that AutomaticCFG would not use any real memory.

I'm happy to provide any more information needed to debug this.

[Extraltodeus]

Do you run the latest version? Is this a huge resolution? Do you use the automatic CFG with a particular preset?

I do sample at 5760*1080 with 12Gb of VRAM and do not have any OOM.

[ghostsquad]

I'm apparently not on the absolute latest. Currently running this commit:

commit be20375d30c2c5d715dd09c11b3046b835b0366d (HEAD -> main, origin/main, origin/HEAD)
Author: Extraltodeus <extraltodeus@gmail.com>
Date:   Wed May 15 01:56:07 2024 +0200

    Update README.md

I'll update it again, looks like lots has changed in the last couple of weeks.

[klromans557]

I've also noticed after updating in the last day or so that my VRAM is still high after comfyui is done generating images. This issue stops when I remove the Warp Drive node. This memory issue seems to happen when I use any the basic auto cfg node as well as the preset node.

[Extraltodeus]

Sorry to answer so late!

There is a newer version that I have made which is much lighter and does not monkeypatch anything.

You can find it here

I just created a few workflows if you want to try them you may need pamparamm schedulers