[inductor] add lowering for repeat_interleave.Tensor with output size specified (#147160)

v0i0 · v0i0 · commit 22103e4c55f9 · 2025-08-08T15:53:22.000-07:00
ghstack-source-id: 7c98b17 Pull Request resolved: #158462
diff --git a/test/inductor/test_torchinductor.py b/test/inductor/test_torchinductor.py
@@ -13670,6 +13670,35 @@ def fn(inp, repeats, output_size):
         args = (inp, repeats, output_size)
         self.assertEqual(fn(*args), torch.compile(fn)(*args))
 
+    @parametrize("dtype", [torch.int32, torch.int64])
+    @parametrize("nd", [1, 2])
+    def test_repeat_interleave_Tensor_decomp(self, dtype, nd):
+        # https://github.com/pytorch/pytorch/issues/147160
+        def f(input, repeats):
+            return torch.repeat_interleave(input, repeats, dim=0, output_size=3) + 1
+
+        input = torch.tensor([[1, 2], [3, 4]], dtype=dtype, device=self.device)
+        input = torch.arange(1, 2**nd + 1, dtype=dtype, device=self.device).reshape(
+            [2] * nd
+        )
+        repeat = torch.tensor([1, 2], device=self.device)
+
+        if input.device.type == "mps" and dtype == torch.int64:
+            raise unittest.SkipTest(
+                "torch.compile fails this test with mps & int64, "
+                "see https://github.com/pytorch/pytorch/issues/159408"
+            )
+
+        f_compiled = torch.compile(f)
+        output, (code,) = run_and_get_code(f_compiled, input, repeat)
+        reference = f(input, repeat)
+        self.assertEqual(output, reference)
+        # we don't lower when the cpp_wrapper is used because it cannot generate
+        # proper examples during autotune
+        can_lower = (not config.cpp_wrapper) and (input.device.type != "mps")
+        has_lowered = not re.search(r"repeat_interleave.Tensor", code)
+        self.assertEqual(has_lowered, can_lower)
+
     # end of class CommonTemplate - add new tests here
 
 
diff --git a/test/inductor/test_torchinductor_codegen_dynamic_shapes.py b/test/inductor/test_torchinductor_codegen_dynamic_shapes.py
@@ -348,7 +348,7 @@ def run(*ex, **kwargs):
     "test_rand_like_deterministic_dynamic_shapes": TestFailure(
         ("cpu", "cuda", "xpu"), is_skip=True
     ),
-    "test_repeat_interleave_2_dynamic_shapes": TestFailure(("cpu", "cuda", "xpu")),
+    "test_repeat_interleave_2_dynamic_shapes": TestFailure(("cpu", "xpu")),
     "test_slice_mutation2_dynamic_shapes": TestFailure(
         ("cpu", "cuda", "xpu"), is_skip=True
     ),
diff --git a/torch/_inductor/decomposition.py b/torch/_inductor/decomposition.py
@@ -1154,3 +1154,25 @@ def rrelu_with_noise_functional(
     else:
         negative_slope = (lower + upper) / 2
         return aten.leaky_relu(self, negative_slope), torch.Tensor()
+
+
+@register_decomposition(aten.repeat_interleave.Tensor)
+def repeat_interleave_Tensor(
+    repeat: torch.Tensor,
+    output_size: Optional[int] = None,
+) -> torch.Tensor:
+    if config.triton.autotune_at_compile_time:
+        # We can't compile-time auto-tune this because
+        # it expects specific data in `repeat`
+        return NotImplemented
+    if output_size is None or type(output_size) is not int:
+        return NotImplemented
+    if repeat.device.type == "mps":
+        return NotImplemented
+    assert repeat.dtype in [torch.int32, torch.int64]
+    assert repeat.ndim == 1
+    cumsum = repeat.cumsum(0)
+    pos = torch.arange(output_size, device=repeat.device)
+    return torch.searchsorted(
+        cumsum, pos, out_int32=(repeat.dtype == torch.int32), right=True
+    )
diff --git a/torch/_inductor/lowering.py b/torch/_inductor/lowering.py
@@ -2879,6 +2879,7 @@ def is_aligned(x):
 
 # index_reduce requires fallback when use_scatter_fallback(...) returns True
 make_fallback(aten.index_reduce)
+make_fallback(aten.repeat_interleave.Tensor, override_decomp=True)
 
 
 # Register with type_promotion_kind None.
