pytorch - 💡(How to fix) Fix [inductor] CPP kernel fusion AssertionError ((8, 7, 7), (56, 7)) when GroupNorm + scaled_dot_product_attention + bmm compiled on CPU [1 comments, 2 participants]

🐛 Describe the bug

TorchInductor crashes during CPP kernel fusion with AssertionError: ((8, 7, 7), (56, 7)) in codegen/cpp.py:fuse. The mismatch arises because GroupNorm's decomposition produces an as_strided tensor with layout [8, 7, 1] stride [7, 1, 56], which when fed into the scaled_dot_product_attention decomposition causes the softmax reduction buffers to be lowered with 2D loop ranges (56, 7) — incompatible with the upstream 3D pointwise kernels (8, 7, 7). The CPP fusion pass then hits an unguarded assert.

import torch
import torch.nn as nn
import torch.nn.functional as F

_m_t9128101 = nn.GroupNorm(1, 7).eval()
_m_t9128105 = nn.RReLU().eval()

torch.manual_seed(0)
t9128098 = torch.randn([8, 7, 7])

def model():
    t9128106 = torch.sigmoid(t9128098)
    t9128101 = _m_t9128101(t9128098)
    t9128104 = F.scaled_dot_product_attention(t9128101, t9128101, t9128101)
    t9128105 = _m_t9128105(t9128104)
    return torch.bmm(t9128106, t9128105)

# Eager: works fine
out_eager = model()
assert torch.isfinite(out_eager).all()

# Compiled: crashes
out_compiled = torch.compile(model, backend='inductor')()

Error logs

I0421 09:51:39.375000 2617416 torch/_dynamo/utils.py:1900] [0/0] ChromiumEventLogger initialized with id 7f69ab89-6e3d-4ac0-badf-e189bdfd2ff6 V0421 09:51:39.376000 2617416 torch/_dynamo/convert_frame.py:1790] [0/0] torchdynamo start compiling model /home/bugs/crash_3c1643d7_reporting.py:11, stack (elided 5 frames): V0421 09:51:39.376000 2617416 torch/_dynamo/convert_frame.py:1790] [0/0] File "/home/bugs/crash_3c1643d7_reporting.py", line 23, in <module> V0421 09:51:39.376000 2617416 torch/_dynamo/convert_frame.py:1790] [0/0] out_compiled = torch.compile(model, backend='inductor')() V0421 09:51:39.376000 2617416 torch/_dynamo/convert_frame.py:1790] [0/0] I0421 09:51:39.376000 2617416 torch/_dynamo/symbolic_convert.py:5320] [0/0] Step 1: torchdynamo start tracing model /home/bugs/crash_3c1643d7_reporting.py:11 I0421 09:51:39.377000 2617416 torch/fx/experimental/symbolic_shapes.py:3945] [0/0] create_env V0421 09:51:39.379000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/bugs/crash_3c1643d7_reporting.py:12 in model (model) V0421 09:51:39.379000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] t9128106 = torch.sigmoid(t9128098) V0421 09:51:39.380000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL torch [] V0421 09:51:39.380000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR sigmoid [LazyVariableTracker(unrealized: <class 'module'>)] V0421 09:51:39.415000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL t9128098 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>)] V0421 09:51:39.415000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 1 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>), LazyVariableTracker(unrealized: <class 'torch.Tensor'>)] V0421 09:51:39.421000 2617416 torch/_dynamo/variables/builder.py:4033] [0/0] wrap_to_fake G['t9128098'] (8, 7, 7) StatefulSymbolicContext(dynamic_sizes=[<DimDynamic.STATIC: 2>, <DimDynamic.STATIC: 2>, <DimDynamic.STATIC: 2>], dynamic_strides=[<DimDynamic.INFER_STRIDE: 4>, <DimDynamic.INFER_STRIDE: 4>, <DimDynamic.INFER_STRIDE: 4>], constraint_sizes=[None, None, None], constraint_strides=[None, None, None], specialize_on=[[], [], []], view_base_context=None, shape_ids=None, tensor_source=GlobalSource(global_name='t9128098'), shape_env_to_source_to_symbol_cache={}) <class 'torch.Tensor'> V0421 09:51:39.422000 2617416 torch/dynamo/output_graph.py:3602] [0/0] create_graph_input G_t9128098 G['t9128098'] FakeTensor(..., size=(8, 7, 7)) at debug_level 0 before=False V0421 09:51:39.427000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE STORE_FAST t9128106 [TensorVariable()] V0421 09:51:39.427000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/bugs/crash_3c1643d7_reporting.py:13 in model (model) V0421 09:51:39.427000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] t9128101 = _m_t9128101(t9128098) V0421 09:51:39.427000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL _m_t9128101 [] V0421 09:51:39.427000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL t9128098 [LazyVariableTracker(unrealized: <class 'torch.nn.modules.normalization.GroupNorm'>)] V0421 09:51:39.427000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 1 [LazyVariableTracker(unrealized: <class 'torch.nn.modules.normalization.GroupNorm'>), TensorVariable()] V0421 09:51:39.430000 2617416 torch/_dynamo/symbolic_convert.py:5824] [0/0] INLINING <code object forward at 0x7fac56264be0, file "/home/.venv/lib/python3.10/site-packages/torch/nn/modules/normalization.py", line 326>, inlined according trace_rules.lookup MOD_INLINELIST V0421 09:51:39.430000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/.venv/lib/python3.10/site-packages/torch/nn/modules/normalization.py:327 in forward (GroupNorm.forward) (inline depth: 1) V0421 09:51:39.430000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] return F.group_norm(input, self.num_groups, self.weight, self.bias, self.eps) V0421 09:51:39.434000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL F [] V0421 09:51:39.434000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR group_norm [LazyVariableTracker(unrealized: <class 'module'>)] V0421 09:51:39.434000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST input [LazyVariableTracker(unrealized: <class 'function'>)] V0421 09:51:39.434000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable()] V0421 09:51:39.434000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR num_groups [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), UnspecializedBuiltinNNModuleVariable(GroupNorm)] V0421 09:51:39.435000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>)] V0421 09:51:39.435000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR weight [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>), UnspecializedBuiltinNNModuleVariable(GroupNorm)] V0421 09:51:39.435000 2617416 torch/_dynamo/variables/builder.py:4033] [0/0] wrap_to_fake G['_m_t9128101']._parameters['weight'] (7,) StatefulSymbolicContext(dynamic_sizes=[<DimDynamic.STATIC: 2>], dynamic_strides=[<DimDynamic.INFER_STRIDE: 4>], constraint_sizes=[None], constraint_strides=[None], specialize_on=[[]], view_base_context=None, shape_ids=None, tensor_source=DictGetItemSource(base=UnspecializedParamBufferSource(base=UnspecializedBuiltinNNModuleSource(base=GlobalSource(global_name='_m_t9128101')), member='_parameters'), index='weight'), shape_env_to_source_to_symbol_cache={}) <class 'torch.nn.parameter.Parameter'> V0421 09:51:39.435000 2617416 torch/dynamo/output_graph.py:3602] [0/0] create_graph_input G_m_t9128101_parameters_weight G['_m_t9128101']._parameters['weight'] Parameter(FakeTensor(..., size=(7,), requires_grad=True)) at debug_level 0 before=False V0421 09:51:39.436000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>), LazyVariableTracker(realized: TensorVariable())] V0421 09:51:39.436000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR bias [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>), LazyVariableTracker(realized: TensorVariable()), UnspecializedBuiltinNNModuleVariable(GroupNorm)] V0421 09:51:39.436000 2617416 torch/_dynamo/variables/builder.py:4033] [0/0] wrap_to_fake G['_m_t9128101']._parameters['bias'] (7,) StatefulSymbolicContext(dynamic_sizes=[<DimDynamic.STATIC: 2>], dynamic_strides=[<DimDynamic.INFER_STRIDE: 4>], constraint_sizes=[None], constraint_strides=[None], specialize_on=[[]], view_base_context=None, shape_ids=None, tensor_source=DictGetItemSource(base=UnspecializedParamBufferSource(base=UnspecializedBuiltinNNModuleSource(base=GlobalSource(global_name='_m_t9128101')), member='_parameters'), index='bias'), shape_env_to_source_to_symbol_cache={}) <class 'torch.nn.parameter.Parameter'> V0421 09:51:39.436000 2617416 torch/dynamo/output_graph.py:3602] [0/0] create_graph_input G_m_t9128101_parameters_bias G['_m_t9128101']._parameters['bias'] Parameter(FakeTensor(..., size=(7,), requires_grad=True)) at debug_level 0 before=False V0421 09:51:39.437000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>), LazyVariableTracker(realized: TensorVariable()), LazyVariableTracker(realized: TensorVariable())] V0421 09:51:39.437000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR eps [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>), LazyVariableTracker(realized: TensorVariable()), LazyVariableTracker(realized: TensorVariable()), UnspecializedBuiltinNNModuleVariable(GroupNorm)] V0421 09:51:39.437000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 5 [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'int'>), LazyVariableTracker(realized: TensorVariable()), LazyVariableTracker(realized: TensorVariable()), LazyVariableTracker(unrealized: <class 'float'>)] V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE RETURN_VALUE None [TensorVariable()] V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:5915] [0/0] DONE INLINING <code object forward at 0x7fac56264be0, file "/home/.venv/lib/python3.10/site-packages/torch/nn/modules/normalization.py", line 326> V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE STORE_FAST t9128101 [TensorVariable()] V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/bugs/crash_3c1643d7_reporting.py:14 in model (model) V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] t9128104 = F.scaled_dot_product_attention(t9128101, t9128101, t9128101) V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL F [] V0421 09:51:39.442000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR scaled_dot_product_attention [LazyVariableTracker(unrealized: <class 'module'>)] V0421 09:51:39.443000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST t9128101 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>)] V0421 09:51:39.443000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST t9128101 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>), TensorVariable()] V0421 09:51:39.443000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST t9128101 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>), TensorVariable(), TensorVariable()] V0421 09:51:39.443000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 3 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>), TensorVariable(), TensorVariable(), TensorVariable()] V0421 09:51:39.456000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE STORE_FAST t9128104 [TensorVariable()] V0421 09:51:39.456000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/bugs/crash_3c1643d7_reporting.py:15 in model (model) V0421 09:51:39.456000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] t9128105 = _m_t9128105(t9128104) V0421 09:51:39.457000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL _m_t9128105 [] V0421 09:51:39.457000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST t9128104 [LazyVariableTracker(unrealized: <class 'torch.nn.modules.activation.RReLU'>)] V0421 09:51:39.457000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 1 [LazyVariableTracker(unrealized: <class 'torch.nn.modules.activation.RReLU'>), TensorVariable()] V0421 09:51:39.457000 2617416 torch/_dynamo/symbolic_convert.py:5824] [0/0] INLINING <code object forward at 0x7fac563e9c60, file "/home/.venv/lib/python3.10/site-packages/torch/nn/modules/activation.py", line 205>, inlined according trace_rules.lookup MOD_INLINELIST V0421 09:51:39.457000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/.venv/lib/python3.10/site-packages/torch/nn/modules/activation.py:209 in forward (RReLU.forward) (inline depth: 1) V0421 09:51:39.457000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] return F.rrelu(input, self.lower, self.upper, self.training, self.inplace) V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL F [] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR rrelu [LazyVariableTracker(unrealized: <class 'module'>)] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST input [LazyVariableTracker(unrealized: <class 'function'>)] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable()] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR lower [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), UnspecializedBuiltinNNModuleVariable(RReLU)] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>)] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR upper [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>), UnspecializedBuiltinNNModuleVariable(RReLU)] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'float'>)] V0421 09:51:39.471000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR training [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'float'>), UnspecializedBuiltinNNModuleVariable(RReLU)] V0421 09:51:39.472000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST self [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'bool'>)] V0421 09:51:39.472000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR inplace [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'bool'>), UnspecializedBuiltinNNModuleVariable(RReLU)] V0421 09:51:39.472000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 5 [LazyVariableTracker(unrealized: <class 'function'>), TensorVariable(), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'float'>), LazyVariableTracker(unrealized: <class 'bool'>), LazyVariableTracker(unrealized: <class 'bool'>)] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE RETURN_VALUE None [TensorVariable()] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:5915] [0/0] DONE INLINING <code object forward at 0x7fac563e9c60, file "/home/.venv/lib/python3.10/site-packages/torch/nn/modules/activation.py", line 205> V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE STORE_FAST t9128105 [TensorVariable()] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] TRACE starts_line /home/bugs/crash_3c1643d7_reporting.py:16 in model (model) V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1414] [0/0] [__trace_source] return torch.bmm(t9128106, t9128105) V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_GLOBAL torch [] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_ATTR bmm [PythonModuleVariable(<module 'torch' from '/home/.venv/lib/python3.10/site-packages/torch/init.py'>)] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST t9128106 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>)] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE LOAD_FAST t9128105 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>), TensorVariable()] V0421 09:51:39.474000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE CALL_FUNCTION 2 [LazyVariableTracker(unrealized: <class 'builtin_function_or_method'>), TensorVariable(), TensorVariable()] V0421 09:51:39.475000 2617416 torch/_dynamo/symbolic_convert.py:1440] [0/0] [__trace_bytecode] TRACE RETURN_VALUE None [TensorVariable()] I0421 09:51:39.475000 2617416 torch/_dynamo/symbolic_convert.py:5547] [0/0] Step 1: torchdynamo done tracing model (RETURN_VALUE) V0421 09:51:39.475000 2617416 torch/_dynamo/symbolic_convert.py:5551] [0/0] return triggered compile V0421 09:51:39.475000 2617416 torch/_dynamo/output_graph.py:1638] [0/0] COMPILING GRAPH due to GraphCompileReason(reason='return_value', user_stack=[<FrameSummary file /home/bugs/crash_3c1643d7_reporting.py, line 16 in model>], graph_break=False) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] TRACED GRAPH V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] ===== __compiled_fn_1_19eb5920_6223_4bed_82ca_67c576c2f24d ===== V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] /home/.venv/lib/python3.10/site-packages/torch/fx/_lazy_graph_module.py class GraphModule(torch.nn.Module): V0421 09:51:39.477000 2617416 torch/dynamo/output_graph.py:2417] [0/0] [graph_code] def forward(self, G_t9128098: "f32[8, 7, 7][49, 7, 1]cpu", G_m_t9128101_parameters_weight: "f32[7][1]cpu", G_m_t9128101_parameters_bias: "f32[7][1]cpu"): V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [graph_code] g_t9128098 = G_t9128098 V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [graph_code] g_m_t9128101_parameters_weight = G_m_t9128101_parameters_weight V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [graph_code] g_m_t9128101_parameters_bias = G_m_t9128101_parameters_bias V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] # File: /home/bugs/crash_3c1643d7_reporting.py:12 in model, code: t9128106 = torch.sigmoid(t9128098) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [_graph_code] t9128106: "f32[8, 7, 7][49, 7, 1]cpu" = torch.sigmoid(g_t9128098) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [graph_code] # File: /home/bugs/crash_3c1643d7_reporting.py:13 in model, code: t9128101 = m_t9128101(t9128098) V0421 09:51:39.477000 2617416 torch/dynamo/output_graph.py:2417] [0/0] [graph_code] t9128101: "f32[8, 7, 7][49, 7, 1]cpu" = torch.nn.functional.group_norm(g_t9128098, 1, g_m_t9128101_parameters_weight, g_m_t9128101_parameters_bias, 1e-05); g_t9128098 = g_m_t9128101_parameters_weight = g_m_t9128101_parameters_bias = None V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] # File: /home/bugs/crash_3c1643d7_reporting.py:14 in model, code: t9128104 = F.scaled_dot_product_attention(t9128101, t9128101, t9128101) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] t9128104: "f32[8, 7, 7][49, 7, 1]cpu" = torch._C._nn.scaled_dot_product_attention(t9128101, t9128101, t9128101); t9128101 = None V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] # File: /home/bugs/crash_3c1643d7_reporting.py:15 in model, code: t9128105 = _m_t9128105(t9128104) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] t9128105: "f32[8, 7, 7][49, 7, 1]cpu" = torch.nn.functional.rrelu(t9128104, 0.125, 0.3333333333333333, False, False); t9128104 = None V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] # File: /home/bugs/crash_3c1643d7_reporting.py:16 in model, code: return torch.bmm(t9128106, t9128105) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] bmm: "f32[8, 7, 7][49, 7, 1]cpu" = torch.bmm(t9128106, t9128105); t9128106 = t9128105 = None V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] return (bmm,) V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] V0421 09:51:39.477000 2617416 torch/_dynamo/output_graph.py:2417] [0/0] [__graph_code] I0421 09:51:39.478000 2617416 torch/_dynamo/output_graph.py:2661] [0/0] Step 2: calling compiler function inductor V0421 09:51:40.315000 2617416 torch/fx/experimental/symbolic_shapes.py:7777] [0/0] eval 2 [trivial] V0421 09:51:40.340000 2617416 torch/fx/experimental/symbolic_shapes.py:7777] [0/0] eval -1 [trivial] I0421 09:51:40.396000 2617416 torch/_dynamo/convert_frame.py:1901] [0/0] run_gc_after_compile: running gc Traceback (most recent call last): File "/home/bugs/crash_3c1643d7_reporting.py", line 23, in <module> out_compiled = torch.compile(model, backend='inductor')() File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 1038, in compile_wrapper raise e.remove_dynamo_frames() from None # see TORCHDYNAMO_VERBOSE=1 File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/eval_frame.py", line 1024, in compile_wrapper return fn(*args, **kwargs) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 2316, in call result = self._torchdynamo_orig_backend( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 2052, in call result = self._inner_convert( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 729, in call result = _compile( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 1827, in _compile guarded_code, tracer_output = compile_inner(code, one_graph, hooks) File "/home/.venv/lib/python3.10/site-packages/torch/_utils_internal.py", line 96, in wrapper_function return function(*args, **kwargs) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 1500, in compile_inner return _compile_inner(code, one_graph, hooks) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 1534, in _compile_inner dynamo_output = compile_frame( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 1408, in compile_frame bytecode, tracer_output = transform_code_object(code, transform) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/bytecode_transformation.py", line 1608, in transform_code_object tracer_output = transformations(instructions, code_options) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 1380, in transform tracer_output = trace_frame( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 341, in _fn return fn(*args, **kwargs) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 863, in trace_frame run_tracer() File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/convert_frame.py", line 844, in run_tracer tracer.run() File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 1794, in run while self.step(): File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 1459, in step self.dispatch_table[inst.opcode](self, inst) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 5570, in RETURN_VALUE self._return(inst) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/symbolic_convert.py", line 5552, in _return all_stack_locals_metadata = self.output.compile_subgraph( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 1889, in compile_subgraph self.compile_and_call_fx_graph(tx, pass2.graph_output_vars(), root) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 2460, in compile_and_call_fx_graph compiled_fn = self.call_user_compiler(gm, self.example_inputs()) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 2613, in call_user_compiler return self._call_user_compiler(gm, example_inputs) File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/output_graph.py", line 2664, in call_user_compiler compiled_fn = compiler_fn(gm, example_inputs) File "/home/.venv/lib/python3.10/site-packages/torch/dynamo/repro/after_dynamo.py", line 156, in call compiled_gm = compiler_fn(gm, example_inputs) File "/home/.venv/lib/python3.10/site-packages/torch/init.py", line 2461, in call return compile_fx(model, inputs, config_patches=self.config) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 2578, in compile_fx return _maybe_wrap_and_compile_fx_main( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 2655, in _maybe_wrap_and_compile_fx_main return _compile_fx_main( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 2864, in _compile_fx_main raise e.remove_dynamo_frames() from None # see TORCHDYNAMO_VERBOSE=1 File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 2850, in _compile_fx_main return aot_autograd( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/backends/common.py", line 124, in call cg = aot_module_simplified(gm, example_inputs, **self.kwargs) File "/home/.venv/lib/python3.10/site-packages/torch/_functorch/aot_autograd.py", line 1161, in aot_module_simplified compiled_fn, _ = aot_stage2_compile( File "/home/.venv/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/graph_compile.py", line 366, in aot_stage2_compile return aot_stage2_autograd(aot_state, aot_graph_capture) File "/home/.venv/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/graph_compile.py", line 2197, in aot_stage2_autograd fwd_output_strides, compiled_fw_func = _aot_stage2b_fw_compile( File "/home/.venv/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/graph_compile.py", line 1986, in _aot_stage2b_fw_compile return _aot_stage2b_compile_forward_or_inference( File "/home/.venv/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/graph_compile.py", line 2513, in _aot_stage2b_compile_forward_or_inference compiled_fw_func = compiler(fw_module, adjusted_flat_args) File "/home/.venv/lib/python3.10/site-packages/torch/_functorch/_aot_autograd/schemas.py", line 1394, in call output_code = self.compiler_fn(gm, example_inputs) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 2719, in fw_compiler_base return compile_fx_forward( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 2390, in compile_fx_forward return inner_compile( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 826, in compile_fx_inner return wrap_compiler_debug(_compile_fx_inner, compiler_name="inductor")( File "/home/.venv/lib/python3.10/site-packages/torch/_dynamo/repro/after_aot.py", line 273, in debug_wrapper inner_compiled_fn = compiler_fn(gm, example_inputs) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1053, in _compile_fx_inner raise InductorError(e, currentframe()).with_traceback( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1037, in _compile_fx_inner mb_compiled_graph = fx_codegen_and_compile( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1798, in fx_codegen_and_compile return scheme.codegen_and_compile(gm, example_inputs, inputs_to_check, graph_kwargs) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/compile_fx.py", line 1570, in codegen_and_compile compiled_module = graph.compile_to_module() File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/graph.py", line 2499, in compile_to_module return self._compile_to_module() File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/graph.py", line 2505, in _compile_to_module self.codegen_with_cpp_wrapper() if self.cpp_wrapper else self.codegen() File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/graph.py", line 2437, in codegen self._update_scheduler() File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/graph.py", line 2431, in _update_scheduler self.scheduler = Scheduler(self.operations) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 2882, in init self._init(nodes) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 2975, in _init self.nodes = self.fuse_nodes(self.nodes) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 3711, in fuse_nodes nodes = self.fuse_nodes_once(nodes, is_reorder_round=False) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 4637, in fuse_nodes_once self._try_fusion_pairs( File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 4546, in _try_fusion_pairs self.fuse_two_nodes(node1, node2, fused_nodes) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/scheduler.py", line 4373, in fuse_two_nodes node3 = self.get_backend(device).fuse(node1, node2) File "/home/.venv/lib/python3.10/site-packages/torch/_inductor/codegen/cpp.py", line 4773, in fuse assert vars1 == vars2, (vars1, vars2) torch._inductor.exc.InductorError: AssertionError: ((8, 7, 7), (56, 7))

I0421 09:51:40.411000 2617416 torch/_dynamo/eval_frame.py:701] TorchDynamo attempted to trace the following frames: [ I0421 09:51:40.411000 2617416 torch/_dynamo/eval_frame.py:701] * model /home/bugs/crash_3c1643d7_reporting.py:11 I0421 09:51:40.411000 2617416 torch/_dynamo/eval_frame.py:701] ] I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] TorchDynamo compilation metrics: I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] Function, Runtimes (s) I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] _compile.compile_inner, 1.0240 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] compile_attempt_0, 1.0180 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] bytecode_tracing, 0.0961 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] get_fake_value, 0.0209 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] create_proxy, 0.0038 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] wrap_to_fake_tensor_and_record, 0.0022 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] variable_builder_call, 0.0454 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] insert_deferred_runtime_asserts, 0.0012 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] OutputGraph.call_user_compiler, 0.9154 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] _recursive_pre_grad_passes, 0.0024 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.pre_grad_passes.group_batch_fusion_passes, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.pre_grad_passes.efficient_conv_bn_eval_pass, 0.0001 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.pre_grad_passes.apply_gumbel_max_trick_pass, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] inductor_codecache_torch_key, 0.1048 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] create_aot_dispatcher_function, 0.5271 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] aot_collect_metadata, 0.0201 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] aot_trace_joint_graph, 0.0880 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] _recursive_joint_graph_passes, 0.2309 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.joint_graph_passes.remove_noop_ops, 0.0010 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.joint_graph_passes.constant_fold_uniform_value, 0.0029 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.joint_graph_passes.pass_pattern_0, 0.0092 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.joint_graph_passes.pass_pattern_1, 0.0002 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] min_cut_rematerialization_partition, 0.0227 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] compile_fx.<locals>.fw_compiler_base, 0.1543 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] compile_fx_inner, 0.1542 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] fx_codegen_and_compile, 0.1531 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] additional_fake_tensor_prop, 0.0127 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] _recursive_post_grad_passes, 0.0181 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.remove_profiler_ops, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.post_grad_custom_pre_pass, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.remove_noop_ops, 0.0001 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.remove_assert_ops, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.pass_pattern_0, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.pass_pattern_1, 0.0047 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.pass_pattern_2, 0.0003 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.stable_sort, 0.0001 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.move_constructors_to_cuda, 0.0001 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.reinplace_inplaceable_ops, 0.0005 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.decompose_triton_kernel_wrapper_functional, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.decompose_auto_functionalized, 0.0001 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.decompose_scan_to_while_loop, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] pass.post_grad_passes.decompose_map_to_while_loop, 0.0000 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] GraphLowering.run, 0.0634 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] GraphLowering.compile_to_fn, 0.0463 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] GraphLowering.compile_to_module, 0.0462 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] GraphLowering.codegen, 0.0462 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] Scheduler.init, 0.0450 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] Scheduler.fused_nodes, 0.0091 I0421 09:51:40.412000 2617416 torch/_dynamo/utils.py:910] gc, 0.0007 V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats constrain_symbol_range: CacheInfo(hits=0, misses=0, maxsize=None, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats guard_or_defer_runtime_assert: CacheInfo(hits=0, misses=0, maxsize=256, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _inner_evaluate_expr: CacheInfo(hits=150, misses=5, maxsize=256, currsize=5) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _simplify_floor_div: CacheInfo(hits=0, misses=0, maxsize=None, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _maybe_guard_rel: CacheInfo(hits=0, misses=0, maxsize=256, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _find: CacheInfo(hits=0, misses=0, maxsize=None, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats has_hint: CacheInfo(hits=0, misses=0, maxsize=256, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats size_hint: CacheInfo(hits=0, misses=0, maxsize=256, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats simplify: CacheInfo(hits=0, misses=2, maxsize=None, currsize=2) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _update_divisible: CacheInfo(hits=0, misses=0, maxsize=None, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats replace: CacheInfo(hits=0, misses=2, maxsize=None, currsize=2) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _maybe_evaluate_static: CacheInfo(hits=7, misses=2, maxsize=None, currsize=2) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats get_implications: CacheInfo(hits=0, misses=0, maxsize=None, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats get_axioms: CacheInfo(hits=53, misses=1, maxsize=None, currsize=1) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats _maybe_evaluate_static_worker: CacheInfo(hits=0, misses=0, maxsize=None, currsize=0) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats safe_expand: CacheInfo(hits=0, misses=2, maxsize=256, currsize=2) V0421 09:51:40.412000 2617416 torch/fx/experimental/symbolic_shapes.py:189] lru_cache_stats uninteresting_files: CacheInfo(hits=7, misses=1, maxsize=None, currsize=1)

Versions

Collecting environment information... PyTorch version: 2.11.0+cu130 Is debug build: False CUDA used to build PyTorch: 13.0 ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.5 LTS (x86_64) GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0 Clang version: 15.0.0 ([email protected]:llvm/llvm-project.git 4ba6a9c9f65bbc8bd06e3652cb20fd4dfc846137) CMake version: version 3.22.1 Libc version: glibc-2.35

Python version: 3.10.12 (main, Mar 3 2026, 11:56:32) [GCC 11.4.0] (64-bit runtime) Python platform: Linux-6.8.0-94-generic-x86_64-with-glibc2.35 Is CUDA available: False CUDA runtime version: No CUDA CUDA_MODULE_LOADING set to: N/A GPU models and configuration: No CUDA Nvidia driver version: No CUDA cuDNN version: No CUDA Is XPU available: False HIP runtime version: N/A MIOpen runtime version: N/A Is XNNPACK available: True Caching allocator config: N/A

CPU: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 52 bits physical, 57 bits virtual Byte Order: Little Endian CPU(s): 384 On-line CPU(s) list: 0-383 Vendor ID: AuthenticAMD Model name: AMD EPYC 9684X 96-Core Processor CPU family: 25 Model: 17 Thread(s) per core: 2 Core(s) per socket: 96 Socket(s): 2 Stepping: 2 BogoMIPS: 5099.98 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good amd_lbr_v2 nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local user_shstk avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc amd_ibpb_ret arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid overflow_recov succor smca fsrm flush_l1d debug_swap ibpb_exit_to_user Virtualization: AMD-V L1d cache: 6 MiB (192 instances) L1i cache: 6 MiB (192 instances) L2 cache: 192 MiB (192 instances) L3 cache: 2.3 GiB (24 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0-95,192-287 NUMA node1 CPU(s): 96-191,288-383 Vulnerability Gather data sampling: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Mitigation; Safe RET Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected Vulnerability Vmscape: Mitigation; IBPB before exit to userspace

Versions of relevant libraries: [pip3] numpy==2.2.6 [pip3] nvidia-cublas==13.1.0.3 [pip3] nvidia-cuda-cupti==13.0.85 [pip3] nvidia-cuda-nvrtc==13.0.88 [pip3] nvidia-cuda-runtime==13.0.96 [pip3] nvidia-cudnn-cu13==9.19.0.56 [pip3] nvidia-cufft==12.0.0.61 [pip3] nvidia-curand==10.4.0.35 [pip3] nvidia-cusolver==12.0.4.66 [pip3] nvidia-cusparse==12.6.3.3 [pip3] nvidia-cusparselt-cu13==0.8.0 [pip3] nvidia-nccl-cu13==2.28.9 [pip3] nvidia-nvjitlink==13.0.88 [pip3] nvidia-nvtx==13.0.85 [pip3] torch==2.11.0 [pip3] triton==3.6.0 [conda] Could not collect

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @jerryzh168 @aditew01 @chauhang @penguinwu @voznesenskym @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @kadeng @muchulee8 @amjames @aakhundov @coconutruben @jataylo