RuntimeError: CUDA error: HIPBLAS_STATUS_NOT_SUPPORTED when calling HIPBLAS_STATUS_NOT_SUPPORTED

[KEDI103]

I tried to run SD 3 but I run into this I also installed amdsmi for be sure its not causing by it.
I got Radeon VII gfx_906
using
pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.1
in ubuntu
Also my VRAM getting up to %80

heres terminal

To see the GUI go to: http://127.0.0.1:8188
got prompt
model_type FLOW
Using split attention in VAE
Using split attention in VAE
no CLIP/text encoder weights in checkpoint, the text encoder model will not be loaded.
clip missing: ['text_projection.weight']
Requested to load SD3ClipModel
Loading 1 new model

rocblaslt warning: No paths matched /media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/lib/hipblaslt/library/*gfx906*co. Make sure that HIPBLASLT_TENSILE_LIBPATH is set correctly.
!!! Exception during processing!!! CUDA error: HIPBLAS_STATUS_NOT_SUPPORTED when calling `HIPBLAS_STATUS_NOT_SUPPORTED`
Traceback (most recent call last):
  File "/media/b_cansin/ai/ai/ComfyUI-master/execution.py", line 151, in recursive_execute
    output_data, output_ui = get_output_data(obj, input_data_all)
  File "/media/b_cansin/ai/ai/ComfyUI-master/execution.py", line 81, in get_output_data
    return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
  File "/media/b_cansin/ai/ai/ComfyUI-master/execution.py", line 74, in map_node_over_list
    results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy_extras/nodes_sd3.py", line 79, in encode
    cond, pooled = clip.encode_from_tokens(tokens, return_pooled=True)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/sd.py", line 142, in encode_from_tokens
    cond, pooled = self.cond_stage_model.encode_token_weights(tokens)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/sd3_clip.py", line 104, in encode_token_weights
    lg_out, l_pooled = self.clip_l.encode_token_weights(token_weight_pairs_l)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/sd1_clip.py", line 40, in encode_token_weights
    out, pooled = self.encode(to_encode)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/sd1_clip.py", line 201, in encode
    return self(tokens)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/sd1_clip.py", line 180, in forward
    outputs = self.transformer(tokens, attention_mask, intermediate_output=self.layer_idx, final_layer_norm_intermediate=self.layer_norm_hidden_state)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/clip_model.py", line 134, in forward
    x = self.text_model(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/clip_model.py", line 109, in forward
    x, i = self.encoder(x, mask=mask, intermediate_output=intermediate_output)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/clip_model.py", line 68, in forward
    x = l(x, mask, optimized_attention)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/clip_model.py", line 49, in forward
    x += self.self_attn(self.layer_norm1(x), mask, optimized_attention)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/clip_model.py", line 16, in forward
    q = self.q_proj(x)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/ops.py", line 50, in forward
    return self.forward_comfy_cast_weights(*args, **kwargs)
  File "/media/b_cansin/ai/ai/ComfyUI-master/comfy/ops.py", line 46, in forward_comfy_cast_weights
    return torch.nn.functional.linear(input, weight, bias)
RuntimeError: CUDA error: HIPBLAS_STATUS_NOT_SUPPORTED when calling `HIPBLAS_STATUS_NOT_SUPPORTED`

Prompt executed in 144.77 seconds

[cosmiconions]

Hello there! I seem to be getting the same error with my 6700 10 GB (non-XT). I too am running Ubuntu on version 22.04, paired with a Ryzen 5600x and 32 GB of RAM. So far, I've attempted to reinstall Torch and Rocm to ensure I have the right version, but the issue still persists. I also seem to be getting that same "rocblaslt warning: No paths matched" warning, except with the 1030 gpu code. I'm fairly new to comfy, and this issue is persistent through the rest of the models as well, but they do seem to work on automatic/sdnext.

[AngryLoki]

For some reason latest versions of Pytorch insist on using majestic hipBLASLt library, which in ROCm 6.1 is supported only by MI250X and MI300X (insert <Spanish Laughing Guy>).

To skip loading of this library, use TORCH_BLAS_PREFER_HIPBLASLT=0 environment variable.
Note that this variable is supported only since pytorch 2.4.0-rc1 (not in 2.3.1), so install nightly build with

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.1 --upgrade

[KEDI103]

For some reason latest versions of Pytorch insist on using majestic hipBLASLt library, which in ROCm 6.1 is supported only by MI250X and MI300X (insert <Spanish Laughing Guy>).

To skip loading of this library, use TORCH_BLAS_PREFER_HIPBLASLT=0 environment variable. Note that this variable is supported only since pytorch 2.4.0-rc1 (not in 2.3.1), so install nightly build with

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.1 --upgrade

With this it fixed problem. Thank you for your help.

cd "/media/b_cansin/ai/ai/ComfyUI-master"
source venv/bin/activate
TORCH_BLAS_PREFER_HIPBLASLT=0 python main.py