Move tests that are mentioned in PR body or commit message to front. Also attempts to find any issues/PRs mentioned in the PR body and search for those too (ex if you link a disable issue and that issue contains the test file that it was failing on)
looking for: dynamo/test_export_mutations
Also removes some printed information in TD
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120621
Approved by: https://github.com/osalpekar
**Summary:**
This commit simplifies the existing decomposition hierarchy
of batch norm ops by adding a single, backend agnostic op:
`batch_norm_with_update`. The existing hierarchy looks like:
```
aten.batch_norm ->
aten._batch_norm_impl_index ->
[
aten.native_batch_norm ->
aten._native_batch_norm_legit (export only) ->
_batch_norm_legit_cpu/cuda (kernels, export only) ->
_batch_norm_cpu/cuda (kernels)
] OR
[ aten.cudnn_batch_norm ] OR
[ aten.miopen_batch_norm ]
```
Aside from complexity, an important problem with the
above decomposition hierarchy is cuda numerics in
export flows. We observed significantly worse convergence
when training a mobilenetv2-like model when using the
`_batch_norm_cuda` kernel instead of the `cudnn_batch_norm`
kernel. This means users who export their models on CPU
first then move the models to cuda later may silently
see worse accuracies even when cudnn is installed,
because they are using the worse kernel. This issue is
summarized in https://github.com/pytorch/pytorch/issues/111384.
Instead, the new hierarchy proposed by consolidating
existing batch norm ops will look like:
```
aten.batch_norm ->
aten.batch_norm_with_update ->
[ _batch_norm_cpu (kernel) ] OR
[ _batch_norm_cuda (kernel) ] OR
[ cudnn_batch_norm (kernel) ] OR
[ miopen_batch_norm (kernel) ]
```
The new op `batch_norm_with_update` hides backend
implementation details and automatically picks the right
kernel based on what is installed. This commit also adds
the following variants to this op:
```
batch_norm_with_update_functional
batch_norm_with_update.out
batch_norm_no_update
batch_norm_no_update.out
batch_norm_backward
```
Note that this commit only adds this op and its variants,
but does not actually change the decomps to produce these
ops in the graph. This will be done after the 2 week FC
window, and the ops used in the old stack is planned to
be removed after the 6 month BC window.
Test Plan: `OpInfo` tests for `batch_norm_with_update`.
Reviewers: albanD, bdhirsh
Subscribers: albanD, bdhirsh, supriyar
Tasks: https://github.com/pytorch/pytorch/issues/111384
Co-authored-by: Tugsbayasgalan Manlaibaatar <tmanlaibaatar@fb.com>
Pull Request resolved: https://github.com/pytorch/pytorch/pull/116092
Approved by: https://github.com/bdhirsh, https://github.com/albanD
Differential Revision: D54447700
## Context
This changeset updates Vulkan SPIR-V codegen to introduce a global SPIR-V shader registry and register shaders dynamically at static initialization time. This change makes it possible to define and link custom shader libraries to the ATen-Vulkan runtime.
Before:
* `gen_vulkan_spv.py` generated two files, `spv.h` and `spv.cpp` which would contain the definition and initialization of Vulkan shader registry variables.
After:
* Introduce the `ShaderRegistry` class in `api/`, which encapsulates functionality of the `ShaderRegistry` class previously defined in the generated `spv.h` file
* Introduce a global shader registry (defined as a static variable in the `api::shader_registry() function`
* Define a `ShaderRegisterInit` class (taking inspiration from `TorchLibraryInit`) that allows for dynamic shader registration
* `gen_vulkan_spv.py` now only generates `spv.cpp`, which defines a static `ShaderRegisterInit` instance that triggers registration of the compiled shaders to the global shader registry.
Benefits:
* Cleaner code base; we no longer have `ShaderRegistry` defined in a generated file, and don't need a separate implementation file (`impl/Registry.*`) to handle shader lookup. All that logic now lives under `api/ShaderRegistry.*`
* Makes it possible to compile and link separate shader libraries, providing similar flexibility as defining and linking custom ATen operators
Pull Request resolved: https://github.com/pytorch/pytorch/pull/121088
Approved by: https://github.com/manuelcandales, https://github.com/jorgep31415
Give TD it's own job so that each shard can get the results from this one job artifact and they will always be in sync with each other/no longer need to worry about consistently issues
* Move test discovery to its own file that is not dependent on torch so it can be run without building torch
* Cannot do cpp test discovery before building pytorch
* Move TD calculation to own file that will create a json file with the final results
* TD is now job/build env agnostic
* TD will rank all tests, including those that test jobs may not want to run (ex it will rank distributed tests along with default tests, even though these tests are never run on the same machine together)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118250
Approved by: https://github.com/huydhn
Current threshold is to cut the bottom 75% of test files, which results in 13 min of tests getting cut.
test_ops, functorch/test_ops, and test_decomp and other really long running test files are not getting cut and make the top 25% to take really long (still 90+ min)
The original plan was to test on rocm but I'm worried about queuing given that cutting 75% of test files only cuts off 13 min, and crossref is rarely referenced by others and people keep talking about getting rid of it, so it's a good alternative
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119426
Approved by: https://github.com/huydhn
Convert from a list/bucket based TD system to just a numbers based TD system. Looks like a massive change but a decent amount of it is tests and removing code.
Main file of interest is interface.py, which Github is collapsing by default due to size
The test files pretty much got rewritten entirely since a lot of the old tests are no longer relevant.
Other notable changes:
* Use Frozenset to make TestRun hashable
* Adds tools/test/heuristics/__init__.py to ensure that unittest can discover the tests
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119901
Approved by: https://github.com/osalpekar, https://github.com/huydhn
Changes sharding to attempt to put all serial tests on as few shards as possible. Parallel tests are then distributed across all shards, with most of which likely ending up on the non serial shards
Example: 8 minutes of serial tests, 20 minutes of parallel tests, 2 proc per machine, 6 machines
-> 8 + 20/2 = 18 total minutes of tests
-> 18 / 6 machines = 3 min per machine
-> all serial tests should fit on 3 machines (3min, 3 min, 2min)
-> majority of parallel tests should go on last 4 machines, one of which is shared with the serial tests
Move serial tests to run first
If I want to move to a purely numbers based sharding, this ensures that parallel tests are run with parallel tests as much as possible instead of interleaving serial + parallel tests, which decreases effectiveness of parallelization, while also ensuring that test reordering is still mostly effective.
See 73e816ee80 for example logs
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119078
Approved by: https://github.com/huydhn
`nargs="?"` accept 0 or 1 argument, but `nargs="*"` accepts 0 or any number of arguments, which is the intended behavior of the tool
Test plan: Run `python tools/build_with_debinfo.py aten/src/ATen/native/cpu/BlasKernel.cpp aten/src/ATen/native/BlasKernel.cpp` and observe that it generates torch_cpu with those two files containing debug information
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120088
Approved by: https://github.com/Skylion007
Changes sharding to attempt to put all serial tests on as few shards as possible. Parallel tests are then distributed across all shards, with most of which likely ending up on the non serial shards
Example: 8 minutes of serial tests, 20 minutes of parallel tests, 2 proc per machine, 6 machines
-> 8 + 20/2 = 18 total minutes of tests
-> 18 / 6 machines = 3 min per machine
-> all serial tests should fit on 3 machines (3min, 3 min, 2min)
-> majority of parallel tests should go on last 4 machines, one of which is shared with the serial tests
Move serial tests to run first
If I want to move to a purely numbers based sharding, this ensures that parallel tests are run with parallel tests as much as possible instead of interleaving serial + parallel tests, which decreases effectiveness of parallelization, while also ensuring that test reordering is still mostly effective.
See 73e816ee80 for example logs
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119078
Approved by: https://github.com/huydhn
Replaces `view_func()` closures with a reified `ViewFunc` data structure. Codegen generates a `ViewFunc` subclass for each view op (e.g. `NarrowViewFunc`) containing state needed to reconstruct the view. The `ViewFunc` API allows for querying and hot-swapping any `SymInt`s or `Tensors` in the state through `get_symints()` / `get_tensors()` / `clone_and_set()`, which will be essential for fake-ification later on.
```cpp
/// Base class for view functions, providing reapplication of a view on a new base.
/// Each view op should get a codegenerated subclass of this class containing
/// any state needed to reconstruct the view. The class also provides convenience
/// accessors for saved SymInts / tensor state. This is useful for e.g. fake-ification,
/// where we want to use symbolic values or fake tensors instead.
struct TORCH_API ViewFunc {
virtual ~ViewFunc() {}
/// Returns any SymInts in the saved state.
virtual std::vector<c10::SymInt> get_symints() const { return {}; }
/// Returns the number of SymInts in the saved state.
virtual size_t num_symints() const { return 0; }
/// Returns any tensors in the saved state.
virtual std::vector<at::Tensor> get_tensors() const { return {}; }
/// Returns the number of tensors in the saved state.
virtual size_t num_tensors() const { return 0; }
/// Reapplies the view on the given base using the saved state.
virtual at::Tensor operator()(const at::Tensor&) const = 0;
/// Returns a clone of this ViewFunc, optionally with the specified saved state.
virtual std::unique_ptr<ViewFunc> clone_and_set(
std::optional<std::vector<c10::SymInt>> = c10::nullopt,
std::optional<std::vector<at::Tensor>> = c10::nullopt) const = 0;
protected:
/// Sets the values of any SymInts in the saved state. The input vector size must
/// match the number of SymInts in the saved state (i.e. the size of the list
/// returned by get_symints()).
virtual void set_symints(std::vector<c10::SymInt>) {}
/// Sets the values of any Tensors in the saved state. The input vector size must
/// match the number of Tensors in the saved state (i.e. the size of the list
/// returned by get_tensors()).
virtual void set_tensors(std::vector<at::Tensor>) {}
};
```
New codegen files:
* `torch/csrc/autograd/generated/ViewFunc.h`
* `torch/csrc/autograd/generated/ViewFuncs.cpp`
The templates for these also contains impls for `ChainedViewFunc` and `ErroringViewFunc` which are used in a few places within autograd.
Example codegen for `slice.Tensor`:
```cpp
// torch/csrc/autograd/generated/ViewFuncs.h
#define SLICE_TENSOR_VIEW_FUNC_AVAILABLE
struct SliceTensorViewFunc : public torch::autograd::ViewFunc {
SliceTensorViewFunc(int64_t dim, c10::optional<c10::SymInt> start, c10::optional<c10::SymInt> end, c10::SymInt step) : dim(dim), start(start), end(end), step(step)
{};
virtual ~SliceTensorViewFunc() override {};
virtual std::vector<c10::SymInt> get_symints() const override;
virtual size_t num_symints() const override;
virtual std::vector<at::Tensor> get_tensors() const override;
virtual size_t num_tensors() const override;
virtual at::Tensor operator()(const at::Tensor&) const override;
virtual std::unique_ptr<ViewFunc> clone_and_set(
std::optional<std::vector<c10::SymInt>> = c10::nullopt,
std::optional<std::vector<at::Tensor>> = c10::nullopt) const override;
protected:
virtual void set_symints(std::vector<c10::SymInt>) override;
virtual void set_tensors(std::vector<at::Tensor>) override;
private:
int64_t dim;
c10::optional<c10::SymInt> start;
c10::optional<c10::SymInt> end;
c10::SymInt step;
};
...
// torch/csrc/autograd/generated/ViewFuncs.cpp
std::vector<c10::SymInt> SliceTensorViewFunc::get_symints() const {
::std::vector<c10::SymInt> symints;
symints.reserve((start.has_value() ? 1 : 0) + (end.has_value() ? 1 : 0) + 1);
if(start.has_value()) symints.insert(symints.end(), *(start));
if(end.has_value()) symints.insert(symints.end(), *(end));
symints.push_back(step);
return symints;
}
size_t SliceTensorViewFunc::num_symints() const {
return static_cast<size_t>((start.has_value() ? 1 : 0) + (end.has_value() ? 1 : 0) + 1);
}
void SliceTensorViewFunc::set_symints(std::vector<c10::SymInt> symints) {
TORCH_INTERNAL_ASSERT(symints.size() == num_symints());
auto i = 0;
if(start.has_value()) start = symints[i];
i += (start.has_value() ? 1 : 0);
if(end.has_value()) end = symints[i];
i += (end.has_value() ? 1 : 0);
step = symints[i];
}
std::vector<at::Tensor> SliceTensorViewFunc::get_tensors() const {
::std::vector<at::Tensor> tensors;
return tensors;
}
size_t SliceTensorViewFunc::num_tensors() const {
return static_cast<size_t>(0);
}
void SliceTensorViewFunc::set_tensors(std::vector<at::Tensor> tensors) {
TORCH_INTERNAL_ASSERT(tensors.size() == num_tensors());
}
at::Tensor SliceTensorViewFunc::operator()(const at::Tensor& input_base) const {
return at::_ops::slice_Tensor::call(input_base, dim, start, end, step);
}
std::unique_ptr<ViewFunc> SliceTensorViewFunc::clone_and_set(
std::optional<std::vector<c10::SymInt>> symints,
std::optional<std::vector<at::Tensor>> tensors) const {
auto output = std::make_unique<SliceTensorViewFunc>(dim, start, end, step);
if (symints.has_value()) {
output->set_symints(std::move(*(symints)));
}
if (tensors.has_value()) {
output->set_tensors(std::move(*(tensors)));
}
return output;
}
```
The `_view_func()` / `_view_func_unsafe()` methods now accept two additional (optional) args for `symint_visitor_fn` / `tensor_visitor_fn`. If these are defined, they are expected to be python callables that operate on a single SymInt / tensor and return a new one. This allows for the hot-swapping needed during fake-ification.
For testing, there are extensive pre-existing tests, and I added a test to ensure that hot-swapping functions correctly.
```sh
python test/test_autograd.py -k test_view_func_replay
python test/test_ops.py -k test_view_replay
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118404
Approved by: https://github.com/ezyang
Replaces `view_func()` closures with a reified `ViewFunc` data structure. Codegen generates a `ViewFunc` subclass for each view op (e.g. `NarrowViewFunc`) containing state needed to reconstruct the view. The `ViewFunc` API allows for querying and hot-swapping any `SymInt`s or `Tensors` in the state through `get_symints()` / `get_tensors()` / `clone_and_set()`, which will be essential for fake-ification later on.
```cpp
/// Base class for view functions, providing reapplication of a view on a new base.
/// Each view op should get a codegenerated subclass of this class containing
/// any state needed to reconstruct the view. The class also provides convenience
/// accessors for saved SymInts / tensor state. This is useful for e.g. fake-ification,
/// where we want to use symbolic values or fake tensors instead.
struct TORCH_API ViewFunc {
virtual ~ViewFunc() {}
/// Returns any SymInts in the saved state.
virtual std::vector<c10::SymInt> get_symints() const { return {}; }
/// Returns the number of SymInts in the saved state.
virtual size_t num_symints() const { return 0; }
/// Returns any tensors in the saved state.
virtual std::vector<at::Tensor> get_tensors() const { return {}; }
/// Returns the number of tensors in the saved state.
virtual size_t num_tensors() const { return 0; }
/// Reapplies the view on the given base using the saved state.
virtual at::Tensor operator()(const at::Tensor&) const = 0;
/// Returns a clone of this ViewFunc, optionally with the specified saved state.
virtual std::unique_ptr<ViewFunc> clone_and_set(
std::optional<std::vector<c10::SymInt>> = c10::nullopt,
std::optional<std::vector<at::Tensor>> = c10::nullopt) const = 0;
protected:
/// Sets the values of any SymInts in the saved state. The input vector size must
/// match the number of SymInts in the saved state (i.e. the size of the list
/// returned by get_symints()).
virtual void set_symints(std::vector<c10::SymInt>) {}
/// Sets the values of any Tensors in the saved state. The input vector size must
/// match the number of Tensors in the saved state (i.e. the size of the list
/// returned by get_tensors()).
virtual void set_tensors(std::vector<at::Tensor>) {}
};
```
New codegen files:
* `torch/csrc/autograd/generated/ViewFunc.h`
* `torch/csrc/autograd/generated/ViewFuncs.cpp`
The templates for these also contains impls for `ChainedViewFunc` and `ErroringViewFunc` which are used in a few places within autograd.
Example codegen for `slice.Tensor`:
```cpp
// torch/csrc/autograd/generated/ViewFuncs.h
#define SLICE_TENSOR_VIEW_FUNC_AVAILABLE
struct SliceTensorViewFunc : public torch::autograd::ViewFunc {
SliceTensorViewFunc(int64_t dim, c10::optional<c10::SymInt> start, c10::optional<c10::SymInt> end, c10::SymInt step) : dim(dim), start(start), end(end), step(step)
{};
virtual ~SliceTensorViewFunc() override {};
virtual std::vector<c10::SymInt> get_symints() const override;
virtual size_t num_symints() const override;
virtual std::vector<at::Tensor> get_tensors() const override;
virtual size_t num_tensors() const override;
virtual at::Tensor operator()(const at::Tensor&) const override;
virtual std::unique_ptr<ViewFunc> clone_and_set(
std::optional<std::vector<c10::SymInt>> = c10::nullopt,
std::optional<std::vector<at::Tensor>> = c10::nullopt) const override;
protected:
virtual void set_symints(std::vector<c10::SymInt>) override;
virtual void set_tensors(std::vector<at::Tensor>) override;
private:
int64_t dim;
c10::optional<c10::SymInt> start;
c10::optional<c10::SymInt> end;
c10::SymInt step;
};
...
// torch/csrc/autograd/generated/ViewFuncs.cpp
std::vector<c10::SymInt> SliceTensorViewFunc::get_symints() const {
::std::vector<c10::SymInt> symints;
symints.reserve((start.has_value() ? 1 : 0) + (end.has_value() ? 1 : 0) + 1);
if(start.has_value()) symints.insert(symints.end(), *(start));
if(end.has_value()) symints.insert(symints.end(), *(end));
symints.push_back(step);
return symints;
}
size_t SliceTensorViewFunc::num_symints() const {
return static_cast<size_t>((start.has_value() ? 1 : 0) + (end.has_value() ? 1 : 0) + 1);
}
void SliceTensorViewFunc::set_symints(std::vector<c10::SymInt> symints) {
TORCH_INTERNAL_ASSERT(symints.size() == num_symints());
auto i = 0;
if(start.has_value()) start = symints[i];
i += (start.has_value() ? 1 : 0);
if(end.has_value()) end = symints[i];
i += (end.has_value() ? 1 : 0);
step = symints[i];
}
std::vector<at::Tensor> SliceTensorViewFunc::get_tensors() const {
::std::vector<at::Tensor> tensors;
return tensors;
}
size_t SliceTensorViewFunc::num_tensors() const {
return static_cast<size_t>(0);
}
void SliceTensorViewFunc::set_tensors(std::vector<at::Tensor> tensors) {
TORCH_INTERNAL_ASSERT(tensors.size() == num_tensors());
}
at::Tensor SliceTensorViewFunc::operator()(const at::Tensor& input_base) const {
return at::_ops::slice_Tensor::call(input_base, dim, start, end, step);
}
std::unique_ptr<ViewFunc> SliceTensorViewFunc::clone_and_set(
std::optional<std::vector<c10::SymInt>> symints,
std::optional<std::vector<at::Tensor>> tensors) const {
auto output = std::make_unique<SliceTensorViewFunc>(dim, start, end, step);
if (symints.has_value()) {
output->set_symints(std::move(*(symints)));
}
if (tensors.has_value()) {
output->set_tensors(std::move(*(tensors)));
}
return output;
}
```
The `_view_func()` / `_view_func_unsafe()` methods now accept two additional (optional) args for `symint_visitor_fn` / `tensor_visitor_fn`. If these are defined, they are expected to be python callables that operate on a single SymInt / tensor and return a new one. This allows for the hot-swapping needed during fake-ification.
For testing, there are extensive pre-existing tests, and I added a test to ensure that hot-swapping functions correctly.
```sh
python test/test_autograd.py -k test_view_func_replay
python test/test_ops.py -k test_view_replay
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118404
Approved by: https://github.com/ezyang
# Motivation
According to [[1/4] Intel GPU Runtime Upstreaming for Device](https://github.com/pytorch/pytorch/pull/116019), as mentioned in [[RFC] Intel GPU Runtime Upstreaming](https://github.com/pytorch/pytorch/issues/114842), this last PR covers the changes under lazy initialization.
# Design
This PR primarily offers the support of multi-processing via lazy initialization. We lazily initialize our runtime avoiding initializing XPU until the first time it is accessed. In our design, we extend `cuda_lazy_init` to `device_lazy_init` which is a device-agnostic API that can support any backend. And change `maybe_initialize_cuda` to `maybe_initialize_device` to support lazy initialization for both CUDA and XPU while maintaining scalability.
# Additional Context
We adopt a similar design to CUDA. So we share some code with CUDA.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/116869
Approved by: https://github.com/EikanWang, https://github.com/jgong5, https://github.com/gujinghui, https://github.com/malfet
ghstack dependencies: #119248
# Motivation
This PR intends to extend `cuda_lazy_init` to `device_lazy_init` which is a device-agnostic API that can support any backend. And change `maybe_initialize_cuda` to `maybe_initialize_device` to support lazy initialization for CUDA while maintaining scalability.
# Design
We maintain a flag for each backend to manage the lazy initialization state separately.
# Additional Context
No need more UTs.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118846
Approved by: https://github.com/malfet
Everyday I move closer and closer to just using numbers
* number of heuristics that marked it as high, probable, low, none etc
* order of heuristics in the `__init__` file as well as how the heuristic ordered the tests
* put heuristics historical edited files and profiling as not trial mode
* briefly sanity checked that all shards of the larger test files (ex test_ops) exist and there are no dups
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118029
Approved by: https://github.com/huydhn
Moves test discovery into a file that doesn't have import torch so test listing can be done without having torch installed.
Helpful when you don't have torch installed (aka me when I'm feeling lazy)
I want to move TD into it's own job that doesn't need to wait for build to finish, so this is part of that.
The first commit is a nothing more than a copy paste of the selected functions/vars into a new file, the second commit has various changes that should be checked.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118574
Approved by: https://github.com/huydhn
Introduces a new op `slice_inverse()`. This is used in the reverse view_func for slice and several other ops (e.g. `split_with_sizes`, `chunk`). It's implemented behind the scenes by a call to `as_strided()`, but it's easier for subclasses to implement the more limited `slice_inverse()` than the full `as_strided()`. This PR:
* Introduces the op itself
* Updates all relevant functional inverses to call `slice_inverse()` instead of `as_strided()` directly
* Makes codegen changes to allow `slice_scatter()` to be the copy variant for `slice_inverse()`
* Need to avoid view_copy codegen (assumes if view name ends in inverse, we don't need to gen one, which is possibly a bad assumption)
@albanD / @soulitzer / @bdhirsh: I'm most interested in your thoughts on the codegen changes and whether this is the right way to go.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/117041
Approved by: https://github.com/bdhirsh
## Context
This change is part of a set of changes that removes all references to the `c10` library in the `api/`, `graph/`, and `impl/` folders of the PyTorch Vulkan codebase. This is to ensure that these components can be built as a standalone library such that they can be used as the foundations of a Android GPU delegate for ExecuTorch.
## Notes for Reviewers
The majority of the changes in this changeset are:
* Replacing instances of `ska::flat_hash_map` with `std::unordered_map`
* `ska::flat_hash_map` is an optimized hash map, but the optimizations shouldn't be too impactful so `std::unordered_map` should suffice. Performance regression testing will be done at the final change in this stack to verify this.
* Replacing `c10::get_hash` with `std::hash` where only one variable is getting hashed or the `utils::hash_combine()` function added to `api/Utils.h` (which was copied from `c10/util/hash.h`)
Differential Revision: [D52662231](https://our.internmc.facebook.com/intern/diff/D52662231/)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/117177
Approved by: https://github.com/yipjustin
ghstack dependencies: #117176
Part 3 of implementation for general [subclass view fake-ification](https://docs.google.com/document/d/1C5taWiplmX7nKiURXDOAZG2W5VNJ2iV0fQFq92H0Cxw).
Changes codegen to generate `view_func()` / `rev_view_func()` by default for python subclasses. With `view_func()` existing more often now, the lazy view rebase logic [here](f10c3f4184/torch/csrc/autograd/variable.cpp (L665-L695)) causes some slight behavior changes for in-place ops on views:
* Additional view nodes are inserted into output graphs, changing their string representation, although they are functionally the same. The extra nodes are removed in AOTAutograd's DCE pass.
* When `t` is a `FunctionalTensor`, calling `t.grad_fn` will now invoke `view_func()`; we need to make sure we're operating in a `FunctionalTensorMode` so the view op calls succeed.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/116512
Approved by: https://github.com/bdhirsh, https://github.com/soulitzer
ghstack dependencies: #115894
