MiniMax H3 Comfy kitchen attention error: "quant_qk_per_thread_int8: Q/K base pointers and B/H/N strides must preserve 4-element alignment"

[genericgod]

Custom Node Testing

• I have tried disabling custom nodes and the issue persists (see how to disable custom nodes if you need help)

Expected Behavior

Faster inference with comfy kitchen attention.

Actual Behavior

Error during SamplerCustomAdvanced with any int8 minimax model. I haven't tested any other model yet.
Sage-Attention works fine.

Steps to Reproduce

Environment:

[INFO] pytorch version: 2.9.1+cu130
[INFO] Device: cuda:0 NVIDIA GeForce RTX 3060 : cudaMallocAsync
[INFO] Python version: 3.11.8 (tags/v3.11.8:db85d51, Feb  6 2024, 22:03:32) [MSC v.1937 64 bit (AMD64)]
[INFO] ComfyUI version: 0.31.0
[INFO] comfy-aimdo version: 0.4.13
[INFO] comfy-kitchen version: 0.2.30
[INFO] comfyui-frontend-package version: 1.48.7
[INFO] comfyui-workflow-templates version: 0.11.37
[INFO] comfyui-embedded-docs version: 0.5.9
[INFO] comfy-kitchen version: 0.2.30
[INFO] comfy-aimdo version: 0.4.13

1. Run with python.exe -s ComfyUI\main.py --listen --enable-cors-header --preview-method auto --fast fp16_accumulation cublas_ops --reserve-vram 1 --fp8_e4m3fn-text-enc --cache-classic --disable-all-custom-nodes
2. Load attached workflow.
   2.1 attention-error.json
3. Load any MiniMax H3 int8 model.
4. Start workflow.
5. See error.

Debug Logs

Node threw an error during execution.


# ComfyUI Error Report
## Error Details
- **Node ID:** 125
- **Node Type:** SamplerCustomAdvanced
- **Exception Type:** RuntimeError
- **Exception Message:** RuntimeError: quant_qk_per_thread_int8: Q/K base pointers and B/H/N strides must preserve 4-element alignment

## Stack Trace

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 545, 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 "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 344, 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 "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 318, in _async_map_node_over_list
    await process_inputs(input_dict, i)

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 306, in process_inputs
    result = f(**inputs)
             ^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
    return method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1990, in EXECUTE_NORMALIZED
    to_return = cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy_extras\nodes_custom_sampler.py", line 1063, in execute
    samples = guider.sample(noise.generate_noise(latent), latent_image, sampler, sigmas, denoise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=noise.seed)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1335, in sample
    output = executor.execute(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1262, in outer_sample
    output = self.inner_sample(noise, latent_image, device, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1237, in inner_sample
    samples = executor.execute(self, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1005, in sample
    samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\utils\_contextlib.py", line 120, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\sampling.py", line 1551, in sample_er_sde
    denoised = model(x, sigmas[i] * s_in, **extra_args)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 640, in __call__
    out = self.inner_model(x, sigma, model_options=model_options, seed=seed)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1208, in __call__
    return self.outer_predict_noise(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1215, in outer_predict_noise
    ).execute(x, timestep, model_options, seed)
      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1218, 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 "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 620, in sampling_function
    out = calc_cond_batch(model, conds, x, timestep, model_options)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 211, in calc_cond_batch
    return _calc_cond_batch_outer(model, conds, x_in, timestep, model_options)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 219, in _calc_cond_batch_outer
    return executor.execute(model, conds, x_in, timestep, model_options)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 335, in _calc_cond_batch
    output = model.apply_model(input_x, timestep_, **c).chunk(batch_chunks)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\model_base.py", line 203, in apply_model
    return comfy.patcher_extension.WrapperExecutor.new_class_executor(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\model_base.py", line 247, in _apply_model
    model_output = self.diffusion_model(xc, t, context=context, control=control, transformer_options=transformer_options, **extra_conds)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 502, in forward
    out = comfy.patcher_extension.WrapperExecutor.new_class_executor(
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 638, in _forward
    h = block(h, t_emb, mod_segments, rope_freqs, transformer_options=transformer_options)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 261, in forward
    x = _mod_gate(x, gate_msa, self.attn(h, rope_freqs=rope_freqs, transformer_options=transformer_options), mod_segments)
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 172, in forward
    out = optimized_attention(q, k, v, self.heads, mask=None, skip_reshape=True, transformer_options=transformer_options)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 191, in wrapper
    return optimized_attention_override.container_function(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 628, in _attention_comfy_kitchen_int8_containers
    quantized = comfy_kitchen.prequantize_int8_attention(
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\sage_attention.py", line 322, in prequantize_int8_attention
    _cuda_backend._C.sage_sdpa_quantize(


## System Information
- **ComfyUI Version:** 0.31.0
- **Arguments:** ComfyUI\main.py --listen --enable-cors-header --preview-method auto --fast fp16_accumulation cublas_ops --reserve-vram 1 --fp8_e4m3fn-text-enc --cache-classic --disable-all-custom-nodes
- **OS:** win32
- **Python Version:** 3.11.8 (tags/v3.11.8:db85d51, Feb  6 2024, 22:03:32) [MSC v.1937 64 bit (AMD64)]
- **Embedded Python:** true
- **PyTorch Version:** 2.9.1+cu130
## Devices

- **Name:** cuda:0 NVIDIA GeForce RTX 3060 : cudaMallocAsync
  - **Type:** cuda
  - **VRAM Total:** 12884377600
  - **VRAM Free:** 11789139968
  - **Torch VRAM Total:** 0
  - **Torch VRAM Free:** 0

## Logs

2026-08-11T13:45:26.604634 - �[32m[INFO]�[0m Adding extra search path checkpoints E:\AI\ComfyUI-models-SSD\checkpoints
2026-08-11T13:45:26.604634 - �[32m[INFO]�[0m Adding extra search path text_encoders E:\AI\ComfyUI-models-SSD\text_encoders
2026-08-11T13:45:26.604634 - �[32m[INFO]�[0m Adding extra search path diffusion_models E:\AI\ComfyUI-models-SSD\diffusion_models
2026-08-11T13:45:26.604634 - �[32m[INFO]�[0m Adding extra search path diffusion_models E:\AI\ComfyUI-models-SSD\unet
2026-08-11T13:45:26.604634 - �[32m[INFO]�[0m Adding extra search path vae E:\AI\ComfyUI-models-SSD\vae
2026-08-11T13:45:59.243447 - �[32m[INFO]�[0m Found comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['adaln', 'apply_rope', 'apply_rope1', 'apply_rope1_', 'apply_rope_', 'apply_rope_split_half', 'apply_rope_split_half1', 'apply_rope_split_half1_', 'apply_rope_split_half_', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'int8_linear', 'na3d', 'quantize_and_rotate_rowwise', 'quantize_int8_rowwise', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'rms_adaln', 'rms_rope', 'rms_rope1', 'rms_rope1_', 'rms_rope_', 'rms_rope_split_half', 'rms_rope_split_half1', 'rms_rope_split_half1_', 'rms_rope_split_half_', 'w4a8_int8_linear']}
2026-08-11T13:45:59.243447 - �[32m[INFO]�[0m Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['adaln', 'apply_rope', 'apply_rope1', 'apply_rope1_', 'apply_rope_', 'apply_rope_split_half', 'apply_rope_split_half1', 'apply_rope_split_half1_', 'apply_rope_split_half_', 'convrot_w4a4_linear', 'dequantize_convrot_w4a4_weight', 'dequantize_int8_convrot_weight', 'dequantize_int8_convrot_weight_dtype', 'dequantize_int8_embedding', 'dequantize_int8_simple', 'dequantize_int8_simple_dtype', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'dequantize_w4a8_int8_weight', 'gemv_awq_w4a16', 'int8_linear', 'na3d', 'prepare_int4_weight_for_int8_linear', 'quantize_and_rotate_rowwise', 'quantize_convrot_w4a4_weight', 'quantize_int8_convrot_weight', 'quantize_int8_rowwise', 'quantize_int8_tensorwise', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'quantize_svdquant_w4a4', 'quantize_w4a8_int8_weight', 'rms_adaln', 'rms_rope', 'rms_rope1', 'rms_rope1_', 'rms_rope_', 'rms_rope_split_half', 'rms_rope_split_half1', 'rms_rope_split_half1_', 'rms_rope_split_half_', 'rotate_int8_convrot_weight', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4', 'scaled_mm_svdquant_w4a4', 'stochastic_rounding_fp8', 'w4a8_int8_linear']}
2026-08-11T13:45:59.243447 - �[32m[INFO]�[0m Found comfy_kitchen backend cuda: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['adaln', 'apply_rope', 'apply_rope1', 'apply_rope1_', 'apply_rope_', 'apply_rope_split_half', 'apply_rope_split_half1', 'apply_rope_split_half1_', 'apply_rope_split_half_', 'convrot_w4a4_linear', 'dequantize_convrot_w4a4_weight', 'dequantize_int8_convrot_weight', 'dequantize_int8_convrot_weight_dtype', 'dequantize_int8_simple', 'dequantize_int8_simple_dtype', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'dequantize_w4a8_int8_weight', 'gemv_awq_w4a16', 'int8_linear', 'na3d', 'prepare_int4_weight_for_int8_linear', 'quantize_and_rotate_rowwise', 'quantize_convrot_w4a4_weight', 'quantize_int8_convrot_weight', 'quantize_int8_rowwise', 'quantize_int8_tensorwise', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'quantize_svdquant_w4a4', 'quantize_w4a8_int8_weight', 'rms_adaln', 'rms_rope', 'rms_rope1', 'rms_rope1_', 'rms_rope_', 'rms_rope_split_half', 'rms_rope_split_half1', 'rms_rope_split_half1_', 'rms_rope_split_half_', 'rotate_int8_convrot_weight', 'scaled_mm_nvfp4', 'scaled_mm_svdquant_w4a4', 'stochastic_rounding_fp8', 'w4a8_int8_linear']}
2026-08-11T13:45:59.243447 - �[32m[INFO]�[0m Found comfy_kitchen backend hip: {'available': False, 'disabled': False, 'unavailable_reason': 'PyTorch ROCm/HIP runtime not available', 'capabilities': []}
2026-08-11T13:45:59.445136 - �[32m[INFO]�[0m Checkpoint files will always be loaded safely.
2026-08-11T13:45:59.656826 - �[32m[INFO]�[0m Total VRAM 12288 MB, total RAM 32689 MB
2026-08-11T13:45:59.656826 - �[32m[INFO]�[0m pytorch version: 2.9.1+cu130
2026-08-11T13:46:03.455244 - �[32m[INFO]�[0m Enabled fp16 accumulation.
2026-08-11T13:46:03.455244 - �[32m[INFO]�[0m Set vram state to: NORMAL_VRAM
2026-08-11T13:46:03.455244 - �[32m[INFO]�[0m Device: cuda:0 NVIDIA GeForce RTX 3060 : cudaMallocAsync
2026-08-11T13:46:03.472350 - �[32m[INFO]�[0m Using async weight offloading with 2 streams
2026-08-11T13:46:03.472350 - �[32m[INFO]�[0m Enabled pinned memory 13075.0
2026-08-11T13:46:04.293399 - �[32m[INFO]�[0m working around nvidia conv3d memory bug.
2026-08-11T13:46:04.317944 - �[32m[INFO]�[0m Using pytorch attention
2026-08-11T13:46:44.337968 - �[32m[INFO]�[0m aimdo: src-win/cuda-detour.c:38:INFO:aimdo_setup_hooks: installing 6 hooks
2026-08-11T13:46:44.365703 - �[32m[INFO]�[0m aimdo: src/control.c:262:INFO:comfy-aimdo NVML pressure enabled
2026-08-11T13:46:44.382009 - �[32m[INFO]�[0m aimdo: src-win/shmem-detect.c:80:INFO:comfy-aimdo WDDM adapter match: NVIDIA GeForce RTX 3060 runtime_luid=00000000:0000e624 dxgi_luid=00000000:0000e624
2026-08-11T13:46:44.382009 - �[32m[INFO]�[0m aimdo: src/control.c:277:INFO:comfy-aimdo inited for GPU: NVIDIA GeForce RTX 3060 (VRAM: 12287 MB)
2026-08-11T13:46:44.382009 - �[32m[INFO]�[0m DynamicVRAM support detected and enabled
2026-08-11T13:46:44.382009 - �[32m[INFO]�[0m Python version: 3.11.8 (tags/v3.11.8:db85d51, Feb  6 2024, 22:03:32) [MSC v.1937 64 bit (AMD64)]
2026-08-11T13:46:44.383518 - �[32m[INFO]�[0m ComfyUI version: 0.31.0
2026-08-11T13:46:44.408980 - �[32m[INFO]�[0m comfy-aimdo version: 0.4.13
2026-08-11T13:46:44.443844 - �[32m[INFO]�[0m comfy-kitchen version: 0.2.30
2026-08-11T13:46:44.574473 - �[32m[INFO]�[0m comfyui-frontend-package version: 1.48.7
2026-08-11T13:46:44.574473 - �[32m[INFO]�[0m comfyui-workflow-templates version: 0.11.37
2026-08-11T13:46:44.574473 - �[32m[INFO]�[0m comfyui-embedded-docs version: 0.5.9
2026-08-11T13:46:44.574473 - �[32m[INFO]�[0m comfy-kitchen version: 0.2.30
2026-08-11T13:46:44.574473 - �[32m[INFO]�[0m comfy-aimdo version: 0.4.13
2026-08-11T13:46:44.598468 - �[32m[INFO]�[0m [Prompt Server] web root: D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static
2026-08-11T13:46:44.599469 - �[32m[INFO]�[0m Asset seeder disabled
2026-08-11T13:46:53.170816 - �[32m[INFO]�[0m No OpenGL_accelerate module loaded: No module named 'OpenGL_accelerate'
2026-08-11T13:46:57.715842 - �[32m[INFO]�[0m Skipping loading of custom nodes
2026-08-11T13:46:57.757749 - �[32m[INFO]�[0m Context impl SQLiteImpl.
2026-08-11T13:46:57.757749 - �[32m[INFO]�[0m Will assume non-transactional DDL.
2026-08-11T13:46:58.358222 - �[32m[INFO]�[0m Starting server

2026-08-11T13:46:58.359232 - �[32m[INFO]�[0m To see the GUI go to: http://0.0.0.0:8188
2026-08-11T13:46:58.359232 - �[32m[INFO]�[0m To see the GUI go to: http://[::]:8188
2026-08-11T13:52:12.524005 - �[1m�[31m[ERROR]�[0m Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 84, in _run
  File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] Eine vorhandene Verbindung wurde vom Remotehost geschlossen
2026-08-11T13:52:12.574476 - �[1m�[31m[ERROR]�[0m Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
  File "asyncio\events.py", line 84, in _run
  File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] Eine vorhandene Verbindung wurde vom Remotehost geschlossen
2026-08-11T13:55:26.572698 - �[32m[INFO]�[0m got prompt
2026-08-11T13:55:26.693070 - �[32m[INFO]�[0m VAE load device: cuda:0, offload device: cpu, dtype: torch.float32
2026-08-11T13:55:26.896550 - �[32m[INFO]�[0m Found quantization metadata version 1
2026-08-11T13:55:27.053147 - �[32m[INFO]�[0m VAE load device: cuda:0, offload device: cpu, dtype: torch.float16
2026-08-11T13:55:27.080318 - �[32m[INFO]�[0m Found quantization metadata version 1
2026-08-11T13:55:27.081321 - �[32m[INFO]�[0m Using MixedPrecisionOps for text encoder
2026-08-11T13:55:27.396780 - �[32m[INFO]�[0m CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float8_e4m3fn
2026-08-11T13:55:27.401332 - �[32m[INFO]�[0m Requested to load MiniMaxH3TEModel_
2026-08-11T13:55:27.560699 - �[32m[INFO]�[0m Model MiniMaxH3TEModel_ prepared for dynamic VRAM loading. 14257MB Staged. 0 patches attached. Force pre-loaded 310 weights: 1272 KB.
2026-08-11T13:55:39.698704 - �[32m[INFO]�[0m Found quantization metadata version 1
2026-08-11T13:55:39.698704 - �[32m[INFO]�[0m Detected mixed precision quantization
2026-08-11T13:55:39.699801 - �[32m[INFO]�[0m Using mixed precision operations
2026-08-11T13:55:39.699801 - �[32m[INFO]�[0m Native ops: convrot_w4a4, int8_tensorwise, asym_w4a8_int8 , emulated ops: float8_e4m3fn, nvfp4, mxfp8, float8_e5m2
2026-08-11T13:55:39.722467 - �[32m[INFO]�[0m model weight dtype torch.bfloat16, manual cast: torch.bfloat16
2026-08-11T13:55:39.726482 - �[32m[INFO]�[0m model_type FLOW_AV
2026-08-11T13:55:40.996129 - �[32m[INFO]�[0m Requested to load MiniMaxH3
2026-08-11T13:55:40.997129 - �[32m[INFO]�[0m 0 models unloaded.
2026-08-11T13:55:41.179234 - �[32m[INFO]�[0m Model MiniMaxH3 prepared for dynamic VRAM loading. 19995MB Staged. 0 patches attached. Force pre-loaded 210 weights: 1175 KB.
2026-08-11T13:55:47.596995 - 
  0%|                                                               | 0/20 [00:00<?, ?it/s,   Model Initializing ...  ]2026-08-11T13:55:49.734274 - 
  0%|                                                               | 0/20 [00:02<?, ?it/s,   Model Initializing ...  ]2026-08-11T13:55:49.734274 - 
2026-08-11T13:55:49.862317 - �[1m�[31m[ERROR]�[0m !!! Exception during processing !!! quant_qk_per_thread_int8: Q/K base pointers and B/H/N strides must preserve 4-element alignment
2026-08-11T13:55:50.066434 - �[1m�[31m[ERROR]�[0m Traceback (most recent call last):
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 545, 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 "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 344, 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 "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 318, in _async_map_node_over_list
    await process_inputs(input_dict, i)
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\execution.py", line 306, in process_inputs
    result = f(**inputs)
             ^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
    return method(locked_class, **inputs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy_api\latest\_io.py", line 1990, in EXECUTE_NORMALIZED
    to_return = cls.execute(*args, **kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy_extras\nodes_custom_sampler.py", line 1063, in execute
    samples = guider.sample(noise.generate_noise(latent), latent_image, sampler, sigmas, denoise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=noise.seed)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1335, in sample
    output = executor.execute(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1262, in outer_sample
    output = self.inner_sample(noise, latent_image, device, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1237, in inner_sample
    samples = executor.execute(self, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1005, in sample
    samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\utils\_contextlib.py", line 120, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\sampling.py", line 1551, in sample_er_sde
    denoised = model(x, sigmas[i] * s_in, **extra_args)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 640, in __call__
    out = self.inner_model(x, sigma, model_options=model_options, seed=seed)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1208, in __call__
    return self.outer_predict_noise(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1215, in outer_predict_noise
    ).execute(x, timestep, model_options, seed)
      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 1218, 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 "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 620, in sampling_function
    out = calc_cond_batch(model, conds, x, timestep, model_options)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 211, in calc_cond_batch
    return _calc_cond_batch_outer(model, conds, x_in, timestep, model_options)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 219, in _calc_cond_batch_outer
    return executor.execute(model, conds, x_in, timestep, model_options)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 335, in _calc_cond_batch
    output = model.apply_model(input_x, timestep_, **c).chunk(batch_chunks)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\model_base.py", line 203, in apply_model
    return comfy.patcher_extension.WrapperExecutor.new_class_executor(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\model_base.py", line 247, in _apply_model
    model_output = self.diffusion_model(xc, t, context=context, control=control, transformer_options=transformer_options, **extra_conds)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 502, in forward
    out = comfy.patcher_extension.WrapperExecutor.new_class_executor(
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\patcher_extension.py", line 113, in execute
    return self.original(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 638, in _forward
    h = block(h, t_emb, mod_segments, rope_freqs, transformer_options=transformer_options)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 261, in forward
    x = _mod_gate(x, gate_msa, self.attn(h, rope_freqs=rope_freqs, transformer_options=transformer_options), mod_segments)
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\minimax\model.py", line 172, in forward
    out = optimized_attention(q, k, v, self.heads, mask=None, skip_reshape=True, transformer_options=transformer_options)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 191, in wrapper
    return optimized_attention_override.container_function(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 628, in _attention_comfy_kitchen_int8_containers
    quantized = comfy_kitchen.prequantize_int8_attention(
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfy_kitchen\sage_attention.py", line 322, in prequantize_int8_attention
    _cuda_backend._C.sage_sdpa_quantize(
RuntimeError: quant_qk_per_thread_int8: Q/K base pointers and B/H/N strides must preserve 4-element alignment

2026-08-11T13:55:50.070526 - �[32m[INFO]�[0m �[32mPrompt executed in 23.49 seconds�[0m


## Attached Workflow
Please make sure that workflow does not contain any sensitive information such as API keys or passwords.

{"id":"8ab855ce-ac2c-40a5-8edd-2b5089c90335","revision":0,"last_node_id":205,"last_link_id":446,"nodes":[{"id":115,"type":"ResolutionSelector","pos":[-1420.6647250655953,5649.535905628488],"size":[270,170],"flags":{},"order":0,"mode":0,"showAdvanced":false,"inputs":[{"localized_name":"aspect_ratio","name":"aspect_ratio","type":"COMBO","widget":{"name":"aspect_ratio"},"link":null},{"localized_name":"megapixels","name":"megapixels","type":"FLOAT","widget":{"name":"megapixels"},"link":null},{"localized_name":"multiple","name":"multiple","type":"INT","widget":{"name":"multiple"},"link":null}],"outputs":[{"localized_name":"width","name":"width","type":"INT","links":[442]},{"localized_name":"height","name":"height","type":"INT","links":[443]}],"title":"Resolution Selector (Size)","properties":{"Node name for S&R":"ResolutionSelector","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":["4:3 (Standard)",0.4,32],"color":"#322","bgcolor":"#533"},{"id":119,"type":"VAELoader","pos":[-1490,5270],"size":[640,86],"flags":{},"order":1,"mode":0,"inputs":[{"localized_name":"vae_name","name":"vae_name","type":"COMBO","widget":{"name":"vae_name"},"link":null}],"outputs":[{"localized_name":"VAE","name":"VAE","type":"VAE","links":[251,441]}],"properties":{"Node name for S&R":"VAELoader","cnr_id":"comfy-core","ver":"0.30.0","models":[{"name":"minimax_h3_video_vae_fp16.safetensors","url":"https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_fp16.safetensors","directory":"vae"}]},"widgets_values":["minimax_h3_video_vae_int8_convrot.safetensors"]},{"id":120,"type":"VAELoader","pos":[-1490,5400],"size":[650,86],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"vae_name","name":"vae_name","type":"COMBO","widget":{"name":"vae_name"},"link":null}],"outputs":[{"localized_name":"VAE","name":"VAE","type":"VAE","links":[250]}],"properties":{"Node name for S&R":"VAELoader","cnr_id":"comfy-core","ver":"0.30.0","models":[{"name":"minimax_h3_audio_vae_fp32.safetensors","url":"https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors","directory":"vae"}]},"widgets_values":["minimax_h3_audio_vae_fp32.safetensors"]},{"id":121,"type":"VAEDecodeAudio","pos":[512.5041178301644,5211.4206472358255],"size":[230,78],"flags":{"collapsed":false},"order":14,"mode":0,"inputs":[{"localized_name":"samples","name":"samples","type":"LATENT","link":396},{"localized_name":"vae","name":"vae","type":"VAE","link":250}],"outputs":[{"localized_name":"AUDIO","name":"AUDIO","type":"AUDIO","links":[438]}],"properties":{"Node name for S&R":"VAEDecodeAudio","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":[]},{"id":122,"type":"VAEDecode","pos":[522.2553531792138,5029.6628233585125],"size":[230,78],"flags":{"collapsed":false},"order":15,"mode":0,"inputs":[{"localized_name":"samples","name":"samples","type":"LATENT","link":397},{"localized_name":"vae","name":"vae","type":"VAE","link":251}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[437]}],"properties":{"Node name for S&R":"VAEDecode","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":[]},{"id":123,"type":"KSamplerSelect","pos":[-267.3353823726477,5291.535217156722],"size":[370,86],"flags":{},"order":3,"mode":0,"inputs":[{"localized_name":"sampler_name","name":"sampler_name","type":"COMBO","widget":{"name":"sampler_name"},"link":null}],"outputs":[{"localized_name":"SAMPLER","name":"SAMPLER","type":"SAMPLER","links":[411]}],"properties":{"Node name for S&R":"KSamplerSelect","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":["er_sde"]},{"id":124,"type":"BasicScheduler","pos":[-274.2301952457128,5410.550243889142],"size":[370,142],"flags":{},"order":10,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":426},{"localized_name":"scheduler","name":"scheduler","type":"COMBO","widget":{"name":"scheduler"},"link":null},{"localized_name":"steps","name":"steps","type":"INT","widget":{"name":"steps"},"link":null},{"localized_name":"denoise","name":"denoise","type":"FLOAT","widget":{"name":"denoise"},"link":null}],"outputs":[{"localized_name":"SIGMAS","name":"SIGMAS","type":"SIGMAS","links":[256]}],"properties":{"Node name for S&R":"BasicScheduler","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":["beta",20,1]},{"id":125,"type":"SamplerCustomAdvanced","pos":[147.5894839652561,5118.947730004133],"size":[230,327.2166748046875],"flags":{},"order":13,"mode":0,"inputs":[{"localized_name":"noise","name":"noise","type":"NOISE","link":253},{"localized_name":"guider","name":"guider","type":"GUIDER","link":254},{"localized_name":"sampler","name":"sampler","type":"SAMPLER","link":411},{"localized_name":"sigmas","name":"sigmas","type":"SIGMAS","link":256},{"localized_name":"latent_image","name":"latent_image","type":"LATENT","link":445}],"outputs":[{"localized_name":"output","name":"output","type":"LATENT","links":[396,397]},{"localized_name":"denoised_output","name":"denoised_output","type":"LATENT","links":[]}],"properties":{"Node name for S&R":"SamplerCustomAdvanced","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":[]},{"id":126,"type":"BasicGuider","pos":[-262.410516034744,5088.947730004133],"size":[360,78],"flags":{},"order":12,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":425},{"localized_name":"conditioning","name":"conditioning","type":"CONDITIONING","link":446}],"outputs":[{"localized_name":"GUIDER","name":"GUIDER","type":"GUIDER","links":[254]}],"properties":{"Node name for S&R":"BasicGuider","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":[]},{"id":127,"type":"UNETLoader","pos":[-1490,4885.176177082825],"size":[640,138],"flags":{},"order":4,"mode":0,"showAdvanced":true,"inputs":[{"localized_name":"unet_name","name":"unet_name","type":"COMBO","widget":{"name":"unet_name"},"link":null},{"localized_name":"weight_dtype","name":"weight_dtype","type":"COMBO","widget":{"name":"weight_dtype"},"link":null}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[434]}],"properties":{"Node name for S&R":"UNETLoader","cnr_id":"comfy-core","ver":"0.30.0","models":[{"name":"minimax_h3_ref2va_pruned_int8_convrot.safetensors","url":"https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors","directory":"diffusion_models"}]},"widgets_values":["minimax_h3_fl2va_pruned_int8_convrot.safetensors","default"]},{"id":128,"type":"CLIPLoader","pos":[-1490,5090],"size":[640,138],"flags":{},"order":5,"mode":0,"inputs":[{"localized_name":"clip_name","name":"clip_name","type":"COMBO","widget":{"name":"clip_name"},"link":null},{"localized_name":"type","name":"type","type":"COMBO","widget":{"name":"type"},"link":null},{"localized_name":"device","name":"device","shape":7,"type":"COMBO","widget":{"name":"device"},"link":null}],"outputs":[{"localized_name":"CLIP","name":"CLIP","type":"CLIP","links":[440]}],"properties":{"Node name for S&R":"CLIPLoader","cnr_id":"comfy-core","ver":"0.30.0","models":[{"name":"qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors","url":"https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors","directory":"text_encoders"}]},"widgets_values":["qwen3vl_32b_minimax_h3_int4_convrot.safetensors","minimax","default"]},{"id":129,"type":"RandomNoise","pos":[-116.14198579900382,4931.612200494712],"size":[360,90],"flags":{},"order":6,"mode":0,"inputs":[{"localized_name":"noise_seed","name":"noise_seed","type":"INT","widget":{"name":"noise_seed"},"link":null}],"outputs":[{"localized_name":"NOISE","name":"NOISE","type":"NOISE","links":[253]}],"properties":{"Node name for S&R":"RandomNoise","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":[7241536727808,"fixed"]},{"id":131,"type":"ComfyMathExpression","pos":[-1436.314415999868,6016.055658375906],"size":[225,166],"flags":{"collapsed":false},"order":9,"mode":0,"inputs":[{"label":"a","localized_name":"values.a","name":"values.a","type":"FLOAT,INT,BOOLEAN","link":261},{"label":"b","localized_name":"values.b","name":"values.b","shape":7,"type":"FLOAT,INT,BOOLEAN","link":null},{"localized_name":"expression","name":"expression","type":"STRING","widget":{"name":"expression"},"link":null}],"outputs":[{"localized_name":"FLOAT","name":"FLOAT","type":"FLOAT","links":[]},{"localized_name":"INT","name":"INT","type":"INT","links":[444]},{"localized_name":"BOOL","name":"BOOL","type":"BOOLEAN","links":[]}],"properties":{"Node name for S&R":"ComfyMathExpression","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":["max(5, round(a * 24)) + (5 - (max(5, round(a * 24)) % 17)) % 17"]},{"id":132,"type":"PrimitiveFloat","pos":[-1420.6647250655953,5869.535905628488],"size":[270,86],"flags":{},"order":7,"mode":0,"inputs":[{"localized_name":"value","name":"value","type":"FLOAT","widget":{"name":"value"},"link":null}],"outputs":[{"localized_name":"FLOAT","name":"FLOAT","type":"FLOAT","links":[261]}],"title":"Float (Duration)","properties":{"Node name for S&R":"PrimitiveFloat","cnr_id":"comfy-core","ver":"0.30.0"},"widgets_values":[10],"color":"#322","bgcolor":"#533"},{"id":200,"type":"ModelAttentionBackend","pos":[-673.8750283282184,4593.761268738543],"size":[270,86],"flags":{},"order":8,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":434},{"localized_name":"attention","name":"attention","type":"COMBO","widget":{"name":"attention"},"link":null}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[425,426]}],"properties":{"Node name for S&R":"ModelAttentionBackend","cnr_id":"comfy-core","ver":"0.31.0"},"widgets_values":["comfy kitchen attention"]},{"id":203,"type":"CreateVideo","pos":[838.0353844697504,4963.420768298481],"size":[270,134],"flags":{},"order":16,"mode":0,"inputs":[{"localized_name":"images","name":"images","type":"IMAGE","link":437},{"localized_name":"audio","name":"audio","shape":7,"type":"AUDIO","link":438},{"localized_name":"fps","name":"fps","type":"FLOAT","widget":{"name":"fps"},"link":null},{"localized_name":"bit_depth","name":"bit_depth","shape":7,"type":"INT","widget":{"name":"bit_depth"},"link":null}],"outputs":[{"localized_name":"VIDEO","name":"VIDEO","type":"VIDEO","links":[439]}],"properties":{"Node name for S&R":"CreateVideo"},"widgets_values":[24,8]},{"id":204,"type":"SaveVideo","pos":[842.7421051012527,5184.952597823082],"size":[270,150],"flags":{},"order":17,"mode":0,"inputs":[{"localized_name":"video","name":"video","type":"VIDEO","link":439},{"localized_name":"filename_prefix","name":"filename_prefix","type":"STRING","widget":{"name":"filename_prefix"},"link":null},{"localized_name":"format","name":"format","type":"COMBO","widget":{"name":"format"},"link":null},{"localized_name":"codec","name":"codec","type":"COMFY_DYNAMICCOMBO_V3","widget":{"name":"codec"},"link":null}],"outputs":[{"localized_name":"video","name":"video","type":"VIDEO","links":null}],"properties":{},"widgets_values":["video/ComfyUI","auto","auto"]},{"id":205,"type":"MiniMaxH3ImageToVideo","pos":[-758.3783692302057,4946.162260317085],"size":[400,270],"flags":{},"order":11,"mode":0,"inputs":[{"localized_name":"clip","name":"clip","type":"CLIP","link":440},{"localized_name":"vae","name":"vae","type":"VAE","link":441},{"localized_name":"first_frame","name":"first_frame","shape":7,"type":"IMAGE","link":null},{"localized_name":"last_frame","name":"last_frame","shape":7,"type":"IMAGE","link":null},{"localized_name":"prompt","name":"prompt","type":"STRING","widget":{"name":"prompt"},"link":null},{"localized_name":"width","name":"width","type":"INT","widget":{"name":"width"},"link":442},{"localized_name":"height","name":"height","type":"INT","widget":{"name":"height"},"link":443},{"localized_name":"length","name":"length","type":"INT","widget":{"name":"length"},"link":444}],"outputs":[{"localized_name":"positive","name":"positive","type":"CONDITIONING","links":[446]},{"localized_name":"LATENT","name":"LATENT","type":"LATENT","links":[445]}],"properties":{"Node name for S&R":"MiniMaxH3ImageToVideo"},"widgets_values":["Test",1344,768,124]}],"links":[[250,120,0,121,1,"VAE"],[251,119,0,122,1,"VAE"],[253,129,0,125,0,"NOISE"],[254,126,0,125,1,"GUIDER"],[256,124,0,125,3,"SIGMAS"],[261,132,0,131,0,"FLOAT"],[396,125,0,121,0,"LATENT"],[397,125,0,122,0,"LATENT"],[411,123,0,125,2,"SAMPLER"],[425,200,0,126,0,"MODEL"],[426,200,0,124,0,"MODEL"],[434,127,0,200,0,"MODEL"],[437,122,0,203,0,"IMAGE"],[438,121,0,203,1,"AUDIO"],[439,203,0,204,0,"VIDEO"],[440,128,0,205,0,"CLIP"],[441,119,0,205,1,"VAE"],[442,115,0,205,5,"INT"],[443,115,1,205,6,"INT"],[444,131,1,205,7,"INT"],[445,205,1,125,4,"LATENT"],[446,205,0,126,1,"CONDITIONING"]],"groups":[{"id":1,"title":"Models","bounding":[-1520,4840,700,670],"color":"#3f789e","flags":{}},{"id":2,"title":"Sampling","bounding":[-282.410516034744,4848.947730004133,690,670],"color":"#3f789e","flags":{}},{"id":3,"title":"Conditioning","bounding":[-790,4840,480,670],"color":"#3f789e","flags":{}},{"id":4,"title":"Decoding and create video","bounding":[440,4840,670,670],"color":"#3f789e","flags":{}},{"id":5,"title":"User Inputs","bounding":[-1460.664725065595,5579.535905628488,397.93012595518485,643.2924544023099],"color":"#3f789e","flags":{}}],"config":{},"extra":{"ds":{"scale":0.5559917313492355,"offset":[1841.4003857605705,-4154.9372755112]},"frontendVersion":"1.48.6","VHS_latentpreview":true,"VHS_latentpreviewrate":8,"VHS_MetadataImage":true,"VHS_KeepIntermediate":true},"version":0.4}

Other

No response

[enlyth]

Same for me on torch 2.9.0+cu130, here's a vibe fix in the meantime:

comfy\ldm\modules\attention.py

def _comfy_kitchen_int8_inputs(q, k, v, heads, mask, skip_reshape, enable_gqa):
    dim_head = q.shape[-1] if skip_reshape else q.shape[-1] // heads
    b = q.shape[0]
    if not skip_reshape:
        q, k, v = _reshape_qkv_to_heads(q, k, v, b, heads, dim_head, enable_gqa, expand_kv=False)
        q, k, v = map(lambda t: t.transpose(1, 2), (q, k, v))

    # add this if statement:
    if b == 1:
        q, k, v = (t if t.is_contiguous() else t.contiguous() for t in (q, k, v))

    if mask is not None:
        if mask.ndim == 2:
            mask = mask.unsqueeze(0)
        if mask.ndim == 3:
            mask = mask.unsqueeze(1)

    return q, k, v, mask, b, dim_head

[dhavalhirdhav]

Same for me on torch 2.9.0+cu130, here's a vibe fix in the meantime:

comfy\ldm\modules\attention.py

def _comfy_kitchen_int8_inputs(q, k, v, heads, mask, skip_reshape, enable_gqa):
    dim_head = q.shape[-1] if skip_reshape else q.shape[-1] // heads
    b = q.shape[0]
    if not skip_reshape:
        q, k, v = _reshape_qkv_to_heads(q, k, v, b, heads, dim_head, enable_gqa, expand_kv=False)
        q, k, v = map(lambda t: t.transpose(1, 2), (q, k, v))

    # add this if statement:
    if b == 1:
        q, k, v = (t if t.is_contiguous() else t.contiguous() for t in (q, k, v))

    if mask is not None:
        if mask.ndim == 2:
            mask = mask.unsqueeze(0)
        if mask.ndim == 3:
            mask = mask.unsqueeze(1)

    return q, k, v, mask, b, dim_head

not working for me, even after updating the attention.py.. on NVIDIA RTX 3090

[danielgue2]

for me the same as well: 2.9.1+cu130