[Inductor] fix device error for NopKernelSchedulerNode (pytorch#141372)

This PR adds device guard support for NopKernelSchedulerNode which may create a tensor. Prior to this PR, we do not codegen device guard for NopKernelSchedulerNode, leading to errors.

Prior to the PR:
```python
def call(args):
    arg0_1, arg1_1, arg2_1, arg3_1, arg4_1, arg5_1, arg6_1, arg7_1, arg8_1, arg9_1, arg10_1 = args
    args.clear()
    assert_size_stride(arg0_1, (1, 1, 2048, 128), (262144, 262144, 128, 1))
    assert_size_stride(arg1_1, (1, 1, 2048, 128), (262144, 262144, 128, 1))
    assert_size_stride(arg2_1, (1, 1, 2048, 128), (262144, 262144, 128, 1))
    assert_size_stride(arg3_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg4_1, (1, 1, 16, 16), (256, 256, 16, 1))
    assert_size_stride(arg5_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg6_1, (1, 1, 16, 16), (256, 256, 16, 1))
    assert_size_stride(arg7_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg8_1, (1, 1, 16, 16), (256, 256, 16, 1))
    assert_size_stride(arg9_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg10_1, (1, 1, 16, 16), (256, 256, 16, 1))
    buf0 = empty_strided_cuda((1, 1, 2048), (2048, 2048, 1), torch.float32) # TODO: ERROR here. Should be cuda:1
    with torch.cuda._DeviceGuard(1):
        torch.cuda.set_device(1)
        buf1 = empty_strided_cuda((1, 1, 2048, 128), (262144, 262144, 128, 1), torch.bfloat16)
        # Topologically Sorted Source Nodes: [flex_attention], Original ATen: []
        stream1 = get_raw_stream(1)
        breakpoint()
        triton_tem_fused_0.run(arg0_1, arg1_1, arg2_1, buf0, arg3_1, arg4_1, arg5_1, arg6_1, buf1, grid=torch._inductor.kernel.flex_attention.flex_attention_grid(1, 1, 2048, 128, meta0), stream=stream1)
        del arg0_1
        del arg1_1
        del arg2_1
        del arg3_1
        del arg4_1
        del arg5_1
        del arg6_1
        del buf0
    return (buf1, )
```

After the PR:
```python
def call(args):
    arg0_1, arg1_1, arg2_1, arg3_1, arg4_1, arg5_1, arg6_1, arg7_1, arg8_1, arg9_1, arg10_1 = args
    args.clear()
    assert_size_stride(arg0_1, (1, 1, 2048, 128), (262144, 262144, 128, 1))
    assert_size_stride(arg1_1, (1, 1, 2048, 128), (262144, 262144, 128, 1))
    assert_size_stride(arg2_1, (1, 1, 2048, 128), (262144, 262144, 128, 1))
    assert_size_stride(arg3_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg4_1, (1, 1, 16, 16), (256, 256, 16, 1))
    assert_size_stride(arg5_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg6_1, (1, 1, 16, 16), (256, 256, 16, 1))
    assert_size_stride(arg7_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg8_1, (1, 1, 16, 16), (256, 256, 16, 1))
    assert_size_stride(arg9_1, (1, 1, 16), (16, 16, 1))
    assert_size_stride(arg10_1, (1, 1, 16, 16), (256, 256, 16, 1))
    with torch.cuda._DeviceGuard(1):
        torch.cuda.set_device(1)
        buf0 = empty_strided_cuda((1, 1, 2048), (2048, 2048, 1), torch.float32) # New: move into device guard
        buf1 = empty_strided_cuda((1, 1, 2048, 128), (262144, 262144, 128, 1), torch.bfloat16)
        # Topologically Sorted Source Nodes: [flex_attention], Original ATen: []
        stream1 = get_raw_stream(1)
        triton_tem_fused_0.run(arg0_1, arg1_1, arg2_1, buf0, arg3_1, arg4_1, arg5_1, arg6_1, buf1, grid=torch._inductor.kernel.flex_attention.flex_attention_grid(1, 1, 2048, 128, meta0), stream=stream1)
        del arg0_1
        del arg1_1
        del arg2_1
        del arg3_1
        del arg4_1
        del arg5_1
        del arg6_1
        del buf0
    return (buf1, )
```

Fixes pytorch#141010

Pull Request resolved: pytorch#141372
Approved by: https://github.com/eellison

Author: BoyuanFeng

test/inductor/test_flex_attention.py

@@ -3303,6 +3303,27 @@ def forward(self, arg0_1: "i32[]", arg1_1: "i32[]", arg2_1: "i32[]", arg3_1: "i3
 """,  # noqa: B950
         )
 
+    @unittest.skipIf(not TEST_MULTIGPU, "detected only one GPU")
+    def test_device_cuda_1(self):
+        class TestModule(torch.nn.Module):
+            def forward(self, q, k, v, block_mask):
+                return flex_attention(q, k, v, block_mask=block_mask)
+
+        q = torch.randn(1, 1, 256, 32, device="cuda:1", dtype=torch.bfloat16)
+        k = torch.randn(1, 1, 256, 32, device="cuda:1", dtype=torch.bfloat16)
+        v = torch.randn(1, 1, 256, 32, device="cuda:1", dtype=torch.bfloat16)
+        mask = create_block_mask(
+            lambda b, h, q_idx, kv_idx: q_idx >= kv_idx,
+            B=None,
+            H=None,
+            Q_LEN=256,
+            KV_LEN=256,
+            device="cuda:1",
+        )
+        mod = torch.compile(TestModule())
+        attn_output = mod(q, k, v, mask)
+        self.assertEqual(attn_output.device, torch.device("cuda:1"))
+
 
 class TestBlockMask(InductorTestCase):
     @supported_platform

test/inductor/test_torchinductor.py

@@ -2494,6 +2494,13 @@ def fn(x):
 
         self.common(fn, (torch.Tensor([]),))
 
+    @requires_multigpu()
+    def test_linspace4(self):
+        def fn(x):
+            return torch.linspace(0, 2, 0, device=f"{GPU_TYPE}:1")
+
+        self.common(fn, (torch.Tensor([]),))
+
     def test_tensor1(self):
         def fn(x):
             return torch.tensor([1], device=x.device) + x, torch.tensor(

torch/_inductor/graph.py

@@ -850,8 +850,12 @@ def register_buffer(self, buffer: ir.Buffer, *, set_name: bool = False) -> str:
         device = buffer.get_device()
         if (
             # Skip empty CPU tensor so that CUDA graphs can succeed, see https://github.com/pytorch/pytorch/pull/114144
-            not (isinstance(buffer, ir.ComputedBuffer) and buffer.is_zero_elements())
-            and device is not None
+            device is not None
+            and not (
+                isinstance(buffer, ir.ComputedBuffer)
+                and buffer.is_zero_elements()
+                and device == torch.device("cpu")
+            )
         ):
             self.add_device_info(device)
 

torch/_inductor/ir.py

@@ -3176,8 +3176,10 @@ def __str__(self) -> str:
         offset = ""
         if self.offset != 0:
             offset = f", offset={self.offset}"
+
+        device_index_str = "" if self.device.index is None else f":{self.device.index}"
         return (
-            f"{type(self).__name__}('{self.device.type}', {self.dtype}, "
+            f"{type(self).__name__}('{self.device.type}{device_index_str}', {self.dtype}, "
             f"size={self.size}, stride={self.stride}{offset})"
         )
 

torch/_inductor/lowering.py

@@ -3033,7 +3033,7 @@ def _new_constant(
         dtype = decode_dtype(dtype) or x.get_dtype()
         device = device or x.get_device()
         size = [sympy.Integer(s) for s in size]
-        return _full(fill_value, device, dtype, size)
+        return _full(fill_value, decode_device(device), dtype, size)
 
     return _new_constant
 
@@ -3045,7 +3045,12 @@ def new_empty(x, size, *, dtype=None, layout=None, device=None, pin_memory=None)
     if device is None:
         device = x.get_device()
     return empty_strided(
-        size, None, dtype=dtype, layout=layout, device=device, pin_memory=pin_memory
+        size,
+        None,
+        dtype=dtype,
+        layout=layout,
+        device=decode_device(device),
+        pin_memory=pin_memory,
     )
 
 
@@ -3059,6 +3064,7 @@ def empty_strided(
     assert_nyi(layout in (None, torch.strided), f"layout={layout}")
     dtype = decode_dtype(dtype) or torch.get_default_dtype()
     device = device or torch.tensor(0.0).device
+    device = decode_device(device)
     pointwise = _full(fill_value=0, device=device, dtype=dtype, size=size)
     pointwise.realize()
     buffer = pointwise.data.data
@@ -3089,7 +3095,12 @@ def new_empty_strided(
     if device is None:
         device = x.get_device()
     return empty_strided(
-        size, stride, dtype=dtype, layout=layout, device=device, pin_memory=pin_memory
+        size,
+        stride,
+        dtype=dtype,
+        layout=layout,
+        device=decode_device(device),
+        pin_memory=pin_memory,
     )
 
 

torch/_inductor/scheduler.py

@@ -3511,9 +3511,7 @@ def _codegen(self) -> None:
 
             self.enter_context(node)
 
-            if not isinstance(node, NopKernelSchedulerNode) and (
-                device := node.get_device()
-            ):
+            if device := node.get_device():
                 if (
                     device != self.current_device
                     or node.is_extern()