Keep on one line to make linter happy

Revert `torch._dynamo.eval_frame`

bug fix

update vision commit hash

Update on "Autotune pointwise/reduction in max_autotune mode"

cc mlazos soumith voznesenskym yanboliang penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx peterbell10 desertfire

[ghstack-poisoned]

Update on "Move ValueRanges into its own module"

I am going to use it in ShapeEnv shortly.

Signed-off-by: Edward Z. Yang <ezyangmeta.com>

cc mlazos soumith voznesenskym yanboliang penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx peterbell10 desertfire

[ghstack-poisoned]

Update on "inductor: fix guard_equals"


Fixes pytorch#94268.

In the code before pytorch#92609, there was an assertion in the `guard_equals` function.
```python
assert self.size_hint(expr) == 0, (expr, self.size_hint(expr))
```

In pytorch#92609, `guard_equals` has been changed to
```python
def guard_equals(self, left: Expr, right: Expr) -> Expr:
    self.shape_env.evaluate_expr(sympy.Eq(left, right))
    return left
```
Considering the case where `left` and `right` are both concrete values for example, `left = 10` and `right = 20`. In the current code, `self.shape_env.evaluate_expr(sympy.Eq(left, right))` will directly return `False`:
https://github.com/pytorch/pytorch/blob/a81cf49d9733b04a2931c85a154ab0bb698650b3/torch/fx/experimental/symbolic_shapes.py#L1380-L1385

This returned value is not used anywhere and the `guard_equals` function will still `return left` in this case even though `left != right`.

cc mlazos soumith voznesenskym yanboliang penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx peterbell10 desertfire

[ghstack-poisoned]

Add necessary XPU hooks

These XPU hooks let PyTorch* know whether has XPU or not

Signed-off-by: Huang, Xunsong <[email protected]>

Update comments

Revert line change breaking bool scatter