Register nonzero for meta device for FBLSim (pytorch#144727)

Summary:
Fix `nonzero is not registered to meta` issue:
```
"NotImplementedError: aten::nonzero: attempted to run this operator with Meta tensors, but there was no fake impl or Meta kernel registered".
```

Reviewed By: ezyang

Differential Revision: D66525640

Pull Request resolved: pytorch#144727
Approved by: https://github.com/ezyang

test/test_meta.py

@@ -13,6 +13,7 @@
 import torch.utils._python_dispatch
 from torch._dispatch.python import enable_python_dispatcher
 from torch._ops import OpOverload, OpOverloadPacket
+from torch.fx.experimental import _config as exp_config
 from torch.testing import make_tensor
 from torch.testing._internal.common_utils import unMarkDynamoStrictTest
 from torch.testing._internal.common_utils import (
@@ -1794,6 +1795,16 @@ def run(device):
         elif cpu_err is not None and meta_err is None:
             raise RuntimeError("cpu failed, but meta didn't.") from cpu_err
 
+    def test_nonzero(self):
+        t = torch.randn(2, 3, 4, device='meta')
+        with exp_config.patch(meta_nonzero_assume_all_nonzero=True):
+            nz = t.nonzero()
+        self.assertEqual(nz.dtype, torch.int64)
+        self.assertEqual(nz.device.type, 'meta')
+        self.assertEqual(nz.shape, torch.Size([24, 3]))
+        self.assertEqual(nz.stride(), torch.Size([1, 24]))
+
+
 
 instantiate_device_type_tests(TestMeta, globals())
 

torch/_meta_registrations.py

@@ -36,6 +36,7 @@
     out_wrapper,
 )
 from torch._refs import _broadcast_shapes, _maybe_broadcast
+from torch.fx.experimental import _config as exp_config
 from torch.utils import _pytree as pytree
 
 
@@ -3115,6 +3116,25 @@ def nonzero_static(self, *, size: int, fill_value: int = -1):
     return self.new_empty((size, self.dim()), dtype=torch.long)
 
 
+@register_meta([torch.ops.aten.nonzero.default, torch.ops.aten.nonzero.out])
+@out_wrapper()
+def nonzero(self):
+    torch._check_not_implemented(
+        exp_config.meta_nonzero_assume_all_nonzero,
+        lambda: "The register_meta function for torch.nonzero() raises unimplemented by default, "
+        "as a correct data-independent implementation does not exist. This implementation "
+        "returns a fake value, assuming all elements of the tensor are non-zero. "
+        "To enable this registration, please set "
+        "'torch.fx.experimental._config.meta_nonzero_assume_all_nonzero' to True.",
+    )
+    return torch.empty_strided(
+        (self.numel(), self.dim()),
+        (1, self.numel()),
+        dtype=torch.long,
+        device=self.device,
+    )
+
+
 @register_meta([aten.index.Tensor, aten._unsafe_index.Tensor])
 def meta_index_Tensor(self, indices):
     torch._check(bool(indices), lambda: "at least one index must be provided")

torch/fx/experimental/_config.py

@@ -82,6 +82,12 @@
 # This flag changes whether we should use the same symbolic variable to represent input sizes that are the same.
 use_duck_shape = True
 
+# Controls the registration of torch.nonzero() on the meta device.
+# When True, nonzero returns a tensor with shape (self.numel(), self.dim())
+# assuming all elements are none-zero.
+# Default is False to prevent unintended registration. Set to True to enable.
+meta_nonzero_assume_all_nonzero = False
+
 from torch.utils._config_module import install_config_module