update the comfyui lastest
use this branch: https://github.com/kijai/ComfyUI/tree/mxfp8
still not work:
and this is my computer info:
Total VRAM 32607 MB, total RAM 196370 MB
pytorch version: 2.9.1+cu130
Set vram state to: NORMAL_VRAM
Disabling smart memory management
Device: cuda:0 NVIDIA GeForce RTX 5090 : native
Using async weight offloading with 2 streams
Enabled pinned memory 88366.0
working around nvidia conv3d memory bug.
Using pytorch attention
aimdo: src-win/cuda-detour.c:77:INFO:aimdo_setup_hooks: found driver at 00007FFDA7FB0000, installing 4 hooks
aimdo: src-win/cuda-detour.c:61:DEBUG:install_hook_entrys: hooks successfully installed
aimdo: src/control.c:66:INFO:comfy-aimdo inited for GPU: NVIDIA GeForce RTX 5090 (VRAM: 32606 MB)
DynamicVRAM support detected and enabled
Python version: 3.11.9 | packaged by Anaconda, Inc. | (main, Apr 19 2024, 16:40:41) [MSC v.1916 64 bit (AMD64)]
ComfyUI version: 0.17.0
comfy-aimdo version: 0.2.8
comfy-kitchen version: 0.2.8

and the error:
Requested to load LTXAV
Model LTXAV prepared for dynamic VRAM loading. 22917MB Staged. 1660 patches attached.
Patching torch settings: torch.backends.cuda.matmul.allow_fp16_accumulation = True
0%| | 0/8 [00:00<?, ?it/s, Model Initializing ... ]
Patching torch settings: torch.backends.cuda.matmul.allow_fp16_accumulation = False
!!! Exception during processing !!! float8 types are not supported by dlpack
Traceback (most recent call last):
File "F:\ComfyUI\ComfyUI\execution.py", line 524, in execute
output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\execution.py", line 333, in get_output_data
return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\execution.py", line 307, in async_map_node_over_list
await process_inputs(input_dict, i)
File "F:\ComfyUI\ComfyUI\execution.py", line 295, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy_api\internal_init.py", line 149, in wrapped_func
return method(locked_class, **inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy_api\latest_io.py", line 1764, in EXECUTE_NORMALIZED
to_return = cls.execute(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy_extras\nodes_custom_sampler.py", line 769, in execute
samples = comfy.sample.sample_custom(model, noise, cfg, sampler, sigmas, positive, negative, latent_image, noise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=noise_seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\sample.py", line 71, in sample_custom
samples = comfy.samplers.sample(model, noise, positive, negative, cfg, model.load_device, sampler, sigmas, model_options=model.model_options, latent_image=latent_image, denoise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 1069, in sample
return cfg_guider.sample(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 1051, in sample
output = executor.execute(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.wrappers[self.idx](self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\custom_nodes\ComfyUI-KJNodes\nodes\ltxv_nodes.py", line 878, in call
out = executor(noise, latent_image, sampler, sigmas, denoise_mask, combined_callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 105, in call
return new_executor.execute(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 112, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 995, in outer_sample
output = self.inner_sample(noise, latent_image, device, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 981, in inner_sample
samples = executor.execute(self, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 112, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 751, in sample
samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\utils_contextlib.py", line 120, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\k_diffusion\sampling.py", line 1015, in sample_lcm
denoised = model(x, sigmas[i] * s_in, **extra_args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 400, in call
out = self.inner_model(x, sigma, model_options=model_options, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 954, in call
return self.outer_predict_noise(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 961, in outer_predict_noise
).execute(x, timestep, model_options, seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 112, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 964, 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 "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 380, in sampling_function
out = calc_cond_batch(model, conds, x, timestep, model_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 205, in calc_cond_batch
return _calc_cond_batch_outer(model, conds, x_in, timestep, model_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 213, in _calc_cond_batch_outer
return executor.execute(model, conds, x_in, timestep, model_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 112, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\samplers.py", line 325, in calc_cond_batch
output = model.apply_model(input_x, timestep, **c).chunk(batch_chunks)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\model_base.py", line 170, in apply_model
return comfy.patcher_extension.WrapperExecutor.new_class_executor(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 112, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\model_base.py", line 209, in _apply_model
model_output = self.diffusion_model(xc, t, context=context, control=control, transformer_options=transformer_options, **extra_conds)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ldm\lightricks\av_model.py", line 1020, in forward
return super().forward(
^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ldm\lightricks\model.py", line 879, in forward
return comfy.patcher_extension.WrapperExecutor.new_class_executor(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 112, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ldm\lightricks\model.py", line 932, in _forward
x = self._process_transformer_blocks(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ldm\lightricks\av_model.py", line 929, in _process_transformer_blocks
vx, ax = block(
^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ldm\lightricks\av_model.py", line 276, in forward
attn1_out = self.attn1(norm_vx, pe=v_pe, mask=self_attention_mask, transformer_options=transformer_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ldm\lightricks\model.py", line 401, in forward
q = self.to_q(x)
^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch\nn\modules\module.py", line 1786, in call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\ops.py", line 922, in forward
input = QuantizedTensor.from_float(input_reshaped, self.layout_type, scale=scale)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\comfy_kitchen\tensor\base.py", line 234, in from_float
qdata, params = get_layout_class(layout_cls).quantize(tensor, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\ComfyUI\ComfyUI\comfy\quant_ops.py", line 114, in quantize
qdata, block_scale = ck.quantize_mxfp8(tensor, pad_32x=needs_padding)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\comfy_kitchen_init.py", line 186, in quantize_mxfp8
return torch.ops.comfy_kitchen.quantize_mxfp8(x, pad_32x)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch_ops.py", line 1255, in call
return self._op(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch_library\custom_ops.py", line 343, in backend_impl
result = self._backend_fns[device_type](*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch_compile.py", line 53, in inner
return disable_fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch_dynamo\eval_frame.py", line 1044, in _fn
return fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch_library\custom_ops.py", line 376, in wrapped_fn
return fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\comfy_kitchen\backends\eager\quantization.py", line 570, in op_quantize_mxfp8
return impl(**kwargs)
^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\comfy_kitchen\backends\cuda_init.py", line 298, in quantize_mxfp8
wrap_for_dlpack(qx),
^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\comfy_kitchen\backends\cuda_init.py", line 136, in _wrap_for_dlpack
return tensor.dlpack(stream=-1)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\env\python311b\Lib\site-packages\torch_tensor.py", line 1782, in dlpack
return _C._to_dlpack(self, dl_device=dl_device, copy=copy)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
BufferError: float8 types are not supported by dlpack

Might be work in progress since its not yet added to comfy ;-)

EDIT:
Ah, I see you already posted at Comfy also https://github.com/Comfy-Org/ComfyUI/pull/12907

And almost double speed.. . 😲 all i want for xmas is a new GPU ;-) haha

And almost double speed.. . 😲 all i want for xmas is a new GPU ;-) haha

mxfp8 speed and memory usage should be similar to fp8 isn't? 🤔