mutated arguments to triton kernels are fusible into the triton kernel.

Test Plan:
- new test

ghstack-source-id: 2c461fa
Pull Request resolved: #134491

mutated arguments to triton kernels are fusible into the triton kernel.

Test Plan:
- new test

ghstack-source-id: 026d142
Pull Request resolved: #134491

ROCM doesn't trigger the layout optimization that makes the test case
valid so we're going to skip the checks.

Should fix the following (I'll close them later)
- #134481
- #134519

Pull Request resolved: #134690
Approved by: https://github.com/FindHao
ghstack dependencies: #134466, #134490, #134491

Fixes #134119
From user feedback, it's difficult to understand what the tests do. We
clarify the docs more.
Pull Request resolved: #134692
Approved by: https://github.com/albanD
ghstack dependencies: #134466, #134490, #134491, #134690

Fixes #134278

Test Plan:
- tested locally
Pull Request resolved: #134688
Approved by: https://github.com/yushangdi
ghstack dependencies: #134466, #134490, #134491, #134690, #134692

aten.empty is almost always fusible into its consumer, so we never CSE
it. This fixes a bug that looks like the following:

```py
@torch.library.custom_op("_reinplacing::sin_cos", mutates_args={"out_sin", "out_cos"})
def sin_cos(x: torch.Tensor, out_sin: torch.Tensor, out_cos: torch.Tensor) -> None:
    out_sin.copy_(x.sin())
    out_cos.copy_(x.cos())

@torch.compile
def f(x):
    out0 = torch.empty_like(x)
    out1 = torch.empty_like(x)
    sin_cos(x, out0, out1)
    return x.clone(), out0, out1

x = torch.randn(3, requires_grad=True)
f(x)
```

- cse would de-duplicate the empty nodes
- reinplacing would add an additional clone (because it can't write to
  both tensors at the same time)
- the clone lowers into a new buffer + a copy_ kernel
- the copy_ kernel is unnecessary because "empty" is special - all reinplacing needed was an additional
  buffer, it doesn't matter what the values are.

We could attempt to fix this on the reinplacing side but this seemed
better as a partitioner heuristic and the reinplacing fix is a bit more
tricky (we'd need to identify that the op never reads from the empty
node).

Test Plan:
- new test (the old number was 27, the new number is 21, so this PR
  helped).

Pull Request resolved: #134703
Approved by: https://github.com/yf225
ghstack dependencies: #134466, #134490, #134491

…ytorch#134491)

mutated arguments to triton kernels are fusible into the triton kernel.

Test Plan:
- new test

Pull Request resolved: pytorch#134491
Approved by: https://github.com/Chillee
ghstack dependencies: pytorch#134364, pytorch#134466, pytorch#134490

ROCM doesn't trigger the layout optimization that makes the test case
valid so we're going to skip the checks.

Should fix the following (I'll close them later)
- pytorch#134481
- pytorch#134519

Pull Request resolved: pytorch#134690
Approved by: https://github.com/FindHao
ghstack dependencies: pytorch#134466, pytorch#134490, pytorch#134491

Fixes pytorch#134119
From user feedback, it's difficult to understand what the tests do. We
clarify the docs more.
Pull Request resolved: pytorch#134692
Approved by: https://github.com/albanD
ghstack dependencies: pytorch#134466, pytorch#134490, pytorch#134491, pytorch#134690

Fixes pytorch#134278

Test Plan:
- tested locally
Pull Request resolved: pytorch#134688
Approved by: https://github.com/yushangdi
ghstack dependencies: pytorch#134466, pytorch#134490, pytorch#134491, pytorch#134690, pytorch#134692

aten.empty is almost always fusible into its consumer, so we never CSE
it. This fixes a bug that looks like the following:

```py
@torch.library.custom_op("_reinplacing::sin_cos", mutates_args={"out_sin", "out_cos"})
def sin_cos(x: torch.Tensor, out_sin: torch.Tensor, out_cos: torch.Tensor) -> None:
    out_sin.copy_(x.sin())
    out_cos.copy_(x.cos())

@torch.compile
def f(x):
    out0 = torch.empty_like(x)
    out1 = torch.empty_like(x)
    sin_cos(x, out0, out1)
    return x.clone(), out0, out1

x = torch.randn(3, requires_grad=True)
f(x)
```

- cse would de-duplicate the empty nodes
- reinplacing would add an additional clone (because it can't write to
  both tensors at the same time)
- the clone lowers into a new buffer + a copy_ kernel
- the copy_ kernel is unnecessary because "empty" is special - all reinplacing needed was an additional
  buffer, it doesn't matter what the values are.

We could attempt to fix this on the reinplacing side but this seemed
better as a partitioner heuristic and the reinplacing fix is a bit more
tricky (we'd need to identify that the op never reads from the empty
node).

Test Plan:
- new test (the old number was 27, the new number is 21, so this PR
  helped).

Pull Request resolved: pytorch#134703
Approved by: https://github.com/yf225
ghstack dependencies: pytorch#134466, pytorch#134490, pytorch#134491