Kneeling down to ask the expert to come up with Mocha's E5m2 quantification model, 30
#34
by
wanghaohao
- opened
Kneeling down to ask the expert to come up with Mocha's E5m2 quantification model, 30
30 series graphics card cannot use E4m3
30 series graphics card cannot use E4m3
It can, the issue is just when using torch.compile with it, and that issue has been resolved in latest Triton versions: https://github.com/woct0rdho/triton-windows/releases/tag/v3.5.0-windows.post21
感谢大佬！
wanghaohao
changed discussion status to
closed
wanghaohao
changed discussion status to
open
大佬，我更新了还是不行，是不是卡的事。是3080ti 20gb版本
ComfyUI Error Report
Error Details
- Node ID: 314
- Node Type: WanVideoSampler
- Exception Type: torch._inductor.exc.InductorError
- Exception Message: CompilationError: at 1:0:
def triton_poi_fused__to_copy_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
^
ValueError("type fp8e4nv not supported in this architecture. The supported fp8 dtypes are ('fp8e4b15', 'fp8e5')")
Set TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS="+dynamo"
Stack Trace
File "D:\comfyui WAN2.2\ComfyUI\execution.py", line 496, 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, hidden_inputs=hidden_inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\ComfyUI\execution.py", line 315, 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, hidden_inputs=hidden_inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\ComfyUI\execution.py", line 289, in _async_map_node_over_list
await process_inputs(input_dict, i)
File "D:\comfyui WAN2.2\ComfyUI\execution.py", line 277, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "D:\comfyui WAN2.2\ComfyUI\custom_nodes\ComfyUI-WanVideoWrapper\nodes_sampler.py", line 3004, in process
raise e
File "D:\comfyui WAN2.2\ComfyUI\custom_nodes\ComfyUI-WanVideoWrapper\nodes_sampler.py", line 2889, in process
noise_pred, noise_pred_ovi, self.cache_state = predict_with_cfg(
^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\ComfyUI\custom_nodes\ComfyUI-WanVideoWrapper\nodes_sampler.py", line 1526, in predict_with_cfg
raise e
File "D:\comfyui WAN2.2\ComfyUI\custom_nodes\ComfyUI-WanVideoWrapper\nodes_sampler.py", line 1397, in predict_with_cfg
noise_pred_cond, noise_pred_ovi, cache_state_cond = transformer(
^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\ComfyUI\custom_nodes\ComfyUI-WanVideoWrapper\wanvideo\modules\model.py", line 2845, in forward
x, x_ip, lynx_ref_feature, x_ovi = block(x, x_ip=x_ip, lynx_ref_feature=lynx_ref_feature, x_ovi=x_ovi, **kwargs) #run block
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_dynamo\eval_frame.py", line 414, in __call__
return super().__call__(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_dynamo\eval_frame.py", line 845, in compile_wrapper
raise e.remove_dynamo_frames() from None # see TORCHDYNAMO_VERBOSE=1
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\compile_fx.py", line 990, in _compile_fx_inner
raise InductorError(e, currentframe()).with_traceback(
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\compile_fx.py", line 974, in _compile_fx_inner
mb_compiled_graph = fx_codegen_and_compile(
^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\compile_fx.py", line 1695, in fx_codegen_and_compile
return scheme.codegen_and_compile(gm, example_inputs, inputs_to_check, graph_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\compile_fx.py", line 1505, in codegen_and_compile
compiled_module = graph.compile_to_module()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\graph.py", line 2319, in compile_to_module
return self._compile_to_module()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\graph.py", line 2329, in _compile_to_module
mod = self._compile_to_module_lines(wrapper_code)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\graph.py", line 2397, in _compile_to_module_lines
mod = PyCodeCache.load_by_key_path(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\codecache.py", line 3548, in load_by_key_path
mod = _reload_python_module(key, path, set_sys_modules=in_toplevel)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\runtime\compile_tasks.py", line 33, in _reload_python_module
exec(code, mod.__dict__, mod.__dict__)
File "C:\Users\XIANNV\AppData\Local\Temp\torchinductor_XIANNV\hw\chw33gu6hts6mskcbvdesadieemv5nrgkae4ezxshioxefatqobw.py", line 192, in <module>
triton_poi_fused__to_copy_2 = async_compile.triton('triton_poi_fused__to_copy_2', '''
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\async_compile.py", line 500, in triton
kernel.precompile(
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\runtime\triton_heuristics.py", line 448, in precompile
self._precompile_worker()
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\runtime\triton_heuristics.py", line 470, in _precompile_worker
compile_results.append(self._precompile_config(c))
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\torch\_inductor\runtime\triton_heuristics.py", line 778, in _precompile_config
binary = triton.compile(*compile_args, **compile_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\triton\compiler\compiler.py", line 278, in compile
module = src.make_ir(options, codegen_fns, module_map, context)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\comfyui WAN2.2\python_embeded\Lib\site-packages\triton\compiler\compiler.py", line 81, in make_ir
return ast_to_ttir(self.fn, self, context=context, options=options, codegen_fns=codegen_fns,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "ast.py", line 407, in visit
File "ast.py", line 415, in generic_visit