ROCm + RX 6600 XT: Anima extremely slow

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Your question

Hi, I'm testing the [wai-anima](https://civitai.com/models/2544636/wai-anima) model on an AMD RX 6600 XT 8GB (gfx1032) using the Windows ROCm build of ComfyUI.

System:

RX 6600 XT 8GB — gfx1032
Windows 11
ComfyUI 0.33.0
PyTorch 2.13.0+rocm10.1.0a20260822
ROCm 10.1.0a20260822

The model runs successfully with no errors, but the performance is extremely slow:

30 steps: 12 minutes 12 seconds
~24.4 seconds per step

Is ~24.4 sec/step normal for Anima on an RX 6600 XT with a standard workflow, or is this unusually slow for this GPU?

Is there any recommended setting, backend, or workaround that can significantly improve Anima performance on this GPU?

### Logs

```powershell
[INFO] setup plugin alembic.autogenerate.schemas
[INFO] setup plugin alembic.autogenerate.tables
[INFO] setup plugin alembic.autogenerate.types
[INFO] setup plugin alembic.autogenerate.constraints
[INFO] setup plugin alembic.autogenerate.defaults
[INFO] setup plugin alembic.autogenerate.comments
[INFO] setup plugin alembic.autogenerate.checkconstraint_byname
[WARNING]

To use the `--enable-manager` feature, the `comfyui-manager` package must be installed first.
command:
C:\comfyui-rocm\python_env\python.exe -m pip install -r C:\comfyui-rocm\manager_requirements.txt
or using uv:
uv pip install -r C:\comfyui-rocm\manager_requirements.txt

[START] Security scan
[DONE] Security scan
## ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2026-08-24 20:32:37.699
** Platform: Windows
** Python version: 3.12.9 (tags/v3.12.9:fdb8142, Feb 4 2025, 15:27:58) [MSC v.1942 64 bit (AMD64)]
** Python executable: C:\comfyui-rocm\python_env\python.exe
** ComfyUI Path: C:\comfyui-rocm
** ComfyUI Base Folder Path: C:\comfyui-rocm
** User directory: C:\comfyui-rocm\user
** ComfyUI-Manager config path: C:\comfyui-rocm\user\__manager\config.ini
** Log path: C:\comfyui-rocm\user\comfyui.log
[INFO]
Prestartup times for custom nodes:
[INFO] 4.3 seconds: C:\comfyui-rocm\custom_nodes\ComfyUI-Manager
[INFO]
[INFO] Found comfy_kitchen backend cuda: {'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_', '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', '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_svdquant_w4a4', 'stochastic_rounding_fp8', 'w4a8_int8_linear']}
[INFO] 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']}
[INFO] 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']}
[INFO] Found comfy_kitchen backend hip: {'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_', 'dequantize_convrot_w4a4_weight', 'dequantize_int8_convrot_weight_dtype', 'dequantize_int8_simple_dtype', 'dequantize_per_tensor_fp8', 'dequantize_w4a8_int8_weight', 'gemv_awq_w4a16', 'quantize_and_rotate_rowwise', 'quantize_convrot_w4a4_weight', 'quantize_int8_convrot_weight', 'quantize_int8_rowwise', 'quantize_int8_tensorwise', '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_', 'stochastic_rounding_fp8']}
[INFO] Checkpoint files will always be loaded safely.
[INFO] Total VRAM 8176 MB, total RAM 32671 MB
[INFO] pytorch version: 2.13.0+rocm10.1.0a20260822
[INFO] Set: torch.backends.cudnn.enabled = False for better AMD performance.
[INFO] AMD arch: gfx1032
[INFO] ROCm version: (7, 16)
[INFO] Set vram state to: NORMAL_VRAM
[INFO] Disabling smart memory management
[INFO] Device: cuda:0 AMD Radeon RX 6600 XT : native
[INFO] Using async weight offloading with 2 streams
[SageAttention] Loaded 0 autotune configs from disk cache
[aiter] Windows: CK and HIP ops are not available. Triton ops only.
[INFO] Using sub quadratic optimization for attention, if you have memory or speed issues try using: --use-split-cross-attention
[INFO] aimdo: src-win/cuda-detour.c:38:INFO:aimdo_setup_hooks: installing 6 hooks
[INFO] aimdo: src-win/shmem-detect.c:80:INFO:comfy-aimdo WDDM adapter match: AMD Radeon RX 6600 XT runtime_luid=00000000:00010556 dxgi_luid=00000000:00010556
[INFO] aimdo: src/control.c:277:INFO:comfy-aimdo inited for GPU: AMD Radeon RX 6600 XT (VRAM: 8176 MB)
[INFO] DynamicVRAM support detected and enabled
[INFO] Python version: 3.12.9 (tags/v3.12.9:fdb8142, Feb 4 2025, 15:27:58) [MSC v.1942 64 bit (AMD64)]
[INFO] ComfyUI version: 0.33.0
[INFO] comfy-aimdo version: 0.4.13
[INFO] comfy-kitchen version: 0.2.31
[INFO] comfyui-frontend-package version: 1.49.6
[INFO] comfyui-workflow-templates version: 0.11.44
[INFO] comfyui-embedded-docs version: 0.5.10
[INFO] comfy-kitchen version: 0.2.31
[INFO] comfy-aimdo version: 0.4.13
[INFO] [Prompt Server] web root: C:\comfyui-rocm\python_env\Lib\site-packages\comfyui_frontend_package\static
[INFO] Asset seeder disabled
[INFO] No OpenGL_accelerate module loaded: No module named 'OpenGL_accelerate'
[CFZ MIOpen] Startup: no saved config at C:\comfyui-rocm\custom_nodes\CFZ-Caching\miopen_config.json — skipped
[INFO] [comfyui-int8-fast-rocm] detected gfx1032 (not RDNA3/RDNA4) -- patching, this arch is assumed to lack working INT8 GEMM until confirmed otherwise -- patched registry.get_implementation('int8_linear') and the W4A8 eager backend's direct int8_linear reference -> rocm_int8_linear (Triton GEMM)
[INFO] ### Loading: ComfyUI-Manager (V3.41)
[INFO] [ComfyUI-Manager] network_mode: public
[INFO] [ComfyUI-Manager] ComfyUI per-queue preview override detected (PR #11261). Manager's preview method feature is disabled. Use ComfyUI's --preview-method CLI option or 'Settings > Execution > Live preview method'.
[INFO] ### ComfyUI Revision: 1 [ef32a575] | Released on '2026-08-23'
[INFO]
Import times for custom nodes:
[INFO] 0.0 seconds: C:\comfyui-rocm\custom_nodes\CFZ-SwitchMenu
[INFO] 0.0 seconds: C:\comfyui-rocm\custom_nodes\CFZ-Caching
[INFO] 0.0 seconds: C:\comfyui-rocm\custom_nodes\websocket_image_save.py
[INFO] 0.0 seconds: C:\comfyui-rocm\custom_nodes\ComfyUI-INT8-Fast-ROCM
[INFO] 0.3 seconds: C:\comfyui-rocm\custom_nodes\ComfyUI-Manager
[INFO] 0.8 seconds: C:\comfyui-rocm\custom_nodes\ComfyUI-HFRemoteVae
[INFO]
[INFO] Context impl SQLiteImpl.
[INFO] Will assume non-transactional DDL.
[INFO] Using LRU cache
[INFO] Starting server

[INFO] To see the GUI go to: http://127.0.0.1:8188
[INFO] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json
[INFO] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json
[INFO] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json
[INFO] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json
[INFO] [ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json
[WARNING] [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[WARNING] [DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/groupNode.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[CFZ Load] No cache files found
[CFZ Load] No cache files found
[WARNING] [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/buttonGroup.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[WARNING] [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/button.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[INFO] got prompt
[INFO] Using split attention in VAE
[INFO] Using split attention in VAE
[INFO] VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
[INFO] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
[INFO] Requested to load AnimaTEModel_
[INFO] Model AnimaTEModel_ prepared for dynamic VRAM loading. 1136MB Staged. 0 patches attached. Force pre-loaded 113 weights: 128 KB.
[INFO] Model AnimaTEModel_ prepared for dynamic VRAM loading. 1136MB Staged. 0 patches attached. Force pre-loaded 113 weights: 128 KB.
[INFO] model weight dtype torch.bfloat16, manual cast: None
[INFO] model_type FLOW
[INFO] Requested to load Anima
[INFO] Model Anima prepared for dynamic VRAM loading. 3988MB Staged. 0 patches attached. Force pre-loaded 156 weights: 77 KB.
23%|███████████████████▎ | 7/30 [02:49<09:20, 24.37s/it]FETCH ComfyRegistry Data [DONE]
[INFO] [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes
FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE]
[INFO] [ComfyUI-Manager] All startup tasks have been completed.
100%|██████████████████████████████████████████████████████████████████████████████████| 30/30 [12:12<00:00, 24.41s/it]
[INFO] Requested to load WanVAE
[INFO] Model WanVAE prepared for dynamic VRAM loading. 241MB Staged. 0 patches attached. Force pre-loaded 60 weights: 61 KB.
[INFO] Prompt executed in 00:12:34
```

### Other

_No response_

Start by reproducing the reported Anima workflow on the Windows ROCm build with the listed RX 6600 XT, ComfyUI, PyTorch, and ROCm versions. Compare the observed 30-step runtime with the standard workflow and review the startup log’s backend, attention, offloading, and GPU entries. Done means determining whether the timing is expected and documenting a verified setting, backend, or workaround if one exists.

Written by the indexing model from the issue text.