OutOfMemoryError; AMD 6900 XT (16 GB), 32 GB RAM, 2.8 s Clip (16 Frames), Memory Profile 4, 480p

Hello,

sorry for bothering you with a newbee question. But I failed to find

a solution by looking up. Gemini and ChatGPT couldnt help me either. Even though I spend a lot of time with them :)

I cant generate any video with some resonable lenght. Even 2.8 s in 480p leads to an OutOfMemory Error when I use Wan2.1, Text2video 1.3B and memory profile 4. I trried several other configurations, deleted WAN2GP and installed it new, but without success. Wan2GP-AMD recognizes the 6900 XT, but than torch tries to allocate to much VRAM...

Yes, very short clips with 1+s length are generated, but there is not much fun in them :)

It would be very nice, if you could lend me a helping hand.

My system: Ryzen 5 5600X, 32 GB Ram, AMD 6800 XT (16 GB), MAG B550 Tomahawk, Windows 11

Thank you very much

sokkel

App: Wan2GP - AMD (wan2gp-amd.git)
Repo: https://github.com/6Morpheus6/wan2gp-amd.git
Generated: 2026-08-30T17:39:33.724Z
Pinokio: 8.0.40
Platform: win32 x64
Node: v22.21.1

Summary

System

{
  "pinokio": {
    "version": "8.0.40",
    "node": "v22.21.1",
    "platform": "win32",
    "arch": "x64"
  },
  "hardware": {
    "gpu": "amd",
    "gpu_model": "amd radeon rx 6900 xt",
    "ram_gb": 32,
    "vram_gb": 16
  },
  "os": {
    "platform": "Windows",
    "distro": "Microsoft Windows 11 Pro",
    "release": "10.0.26200",
    "codename": "25H2",
    "kernel": "10.0.26200",
    "arch": "x64",
    "build": "26200",
    "servicepack": "0.0",
    "uefi": true
  },
  "system": {
    "manufacturer": "Micro-Star International Co., Ltd.",
    "model": "MS-7C91",
    "version": "2.0",
    "virtual": false
  },
  "cpu": {
    "manufacturer": "AMD",
    "brand": "Ryzen 5 5600X 6-Core Processor",
    "vendor": "AuthenticAMD",
    "family": "25",
    "model": "33",
    "stepping": "0",
    "revision": "8448",
    "speed": 3.7,
    "speedMin": 3.7,
    "speedMax": 3.7,
    "cores": 12,
    "physicalCores": 6,
    "processors": 1,
    "performanceCores": 12,
    "efficiencyCores": 0,
    "virtualization": false,
    "cache": {
      "l1d": 192,
      "l1i": 192,
      "l2": 3145728,
      "l3": 33554432
    }
  },
  "memory": {
    "total": 34282233856,
    "free": 25978597376,
    "used": 8303636480,
    "active": 8303632384,
    "available": 25978601472,
    "buffers": 0,
    "cached": 0,
    "slab": 0,
    "buffcache": 0,
    "swaptotal": 16777216000,
    "swapused": 470810624,
    "swapfree": 16306405376
  },
  "gpus": [
    {
      "model": "amd radeon rx 6900 xt"
    }
  ],
  "graphics": {
    "controllers": [
      {
        "vendor": "Advanced Micro Devices, Inc.",
        "model": "AMD Radeon RX 6900 XT",
        "bus": "PCI",
        "vram": 16368,
        "vramDynamic": true
      }
    ],
    "displays": [
      {
        "model": "GO7Q4G",
        "main": true,
        "builtin": false,
        "connection": "DP",
        "currentResX": 2560,
        "currentResY": 1440,
        "resolutionX": 2560,
        "resolutionY": 1440,
        "pixelDepth": 32,
        "currentRefreshRate": 120
      }
    ]
  }
}

Logs

logs/api/start.js/1788111318580

Source: api / start.js
Lines: 86 total, last 86 included

[api shell.run]
Microsoft Windows [Version 10.0.26200.9168]
(c) Microsoft Corporation. Alle Rechte vorbehalten.

C:\pinokio\api\wan2gp-amd.git\app>conda_hook & conda deactivate & conda deactivate & conda deactivate & conda activate base & C:\pinokio\api\wan2gp-amd.git\app\env\Scripts\activate C:\pinokio\api\wan2gp-amd.git\app\env & bluefairy-activate.cmd && python wgp.py
[OK] bluefairy active: uv/npm/bun protection enabled
[GGUF][llama.cpp CUDA] kernels unavailable, using fallback
Loaded plugin: Motion Designer (from motion_designer)
* Running on local URL:  http://localhost:7860
* To create a public link, set `share=True` in `launch()`.
Loading Model 'ckpts\wan2.1_text2video_1.3B_mbf16.safetensors' ...
Loading Text Encoder 'ckpts\umt5-xxl\models_t5_umt5-xxl-enc-quanto_int8.safetensors' ...
************ Memory Management for the GPU Poor (mmgp 3.7.14) by DeepBeepMeep ************
Pinning data of 'transformer' to reserved RAM
The whole model was pinned to reserved RAM: 11 large blocks spread across 2706.90 MB
Hooked to model 'transformer' (WanModel)
Async loading plan for model 'transformer' : base size of 0.76 MB will be preloaded with a 88.58 MB async circular shuttle
Hooked to model 'vae' (WanVAE_)
Hooked to model 'text_encoder' (T5Encoder)
Async loading plan for model 'text_encoder' : base size of 2003.01 MB will be preloaded with a 184.10 MB async circular shuttle
  0%|                                                                                                                                                                                                        | 0/30 [00:00<?, ?steps/s]C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\utils\_device.py:122: UserWarning: 1Torch was not compiled with memory efficient attention. (Triggered internally at B:\src\torch\aten\src\ATen\native\transformers\hip\sdp_utils.cpp:862.)
  return func(*args, **kwargs)
  0%|                                                                                                                                                                                                        | 0/30 [00:16<?, ?steps/s]
Traceback (most recent call last):
  File "C:\pinokio\api\wan2gp-amd.git\app\wgp.py", line 7861, in generate_media
    samples = wan_model.generate(
              ^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\models\wan\any2video.py", line 1727, in generate
    noise_pred = denoise_fn(latents)
                 ^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\models\wan\any2video.py", line 1634, in denoise_with_cfg_fn
    ret_values = trans( **gen_args , **kwargs)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\nn\modules\module.py", line 1778, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\nn\modules\module.py", line 1789, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\mmgp\offload.py", line 3359, in check_change_module
    return previous_method(*args, **kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\models\wan\modules\model.py", line 2036, in forward
    x_list[i] = block(x, context = context, hints= hints, audio_scale= audio_scale, multitalk_audio = multitalk_audio, multitalk_masks =multitalk_masks, e= e0,  motion_vec = motion_vec, lynx_ip_embeds= lynx_ip_embeds, lynx_ref_buffer = lynx_ref_buffer, sub_x_no =i,  **block_kwargs)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\nn\modules\module.py", line 1778, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\nn\modules\module.py", line 1789, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\mmgp\offload.py", line 3337, in check_load_into_GPU_needed_other
    return previous_method(*args, **kwargs) # other
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\models\wan\modules\model.py", line 653, in forward
    y, x_ref_attn_map = self.self_attn( xlist, grid_sizes, freqs, block_mask = block_mask, ref_target_masks = multitalk_masks, ref_images_count = ref_images_count, standin_phase= standin_phase, lynx_ref_buffer = lynx_ref_buffer, lynx_ref_scale = lynx_ref_scale, sub_x_no = sub_x_no)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\nn\modules\module.py", line 1778, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\nn\modules\module.py", line 1789, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\mmgp\offload.py", line 3337, in check_load_into_GPU_needed_other
    return previous_method(*args, **kwargs) # other
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\models\wan\modules\model.py", line 385, in forward
    x = pay_attention( qkv_list, recycle_q=True)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\_dynamo\eval_frame.py", line 1297, in _fn
    return fn(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\shared\attention.py", line 531, in pay_attention
    x = sdpa_wrapper(qkv_list, lq, attention_mask=attention_mask, causal=causal)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\_dynamo\eval_frame.py", line 1297, in _fn
    return fn(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\shared\attention.py", line 233, in sdpa_wrapper
    o = F.scaled_dot_product_attention(q, k, v, attn_mask=attention_mask, is_causal=causal).transpose(1,2)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\pinokio\api\wan2gp-amd.git\app\env\Lib\site-packages\torch\utils\_device.py", line 122, in __torch_function__
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.48 GiB. GPU 0 has a total capacity of 15.98 GiB of which 14.33 GiB is free. Of the allocated memory 1.30 GiB is allocated by PyTorch, and 28.04 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://docs.pytorch.org/docs/stable/notes/cuda.html#optimizing-memory-usage-with-pytorch-cuda-alloc-conf)
Error Queue autosaved successfully to error_queue.zip