Commit Graph

1629 Commits (7731211d02200127e97e259163df911f046d1fbd)
 

Author SHA1 Message Date
Vivek Khandelwal 2597c481f6 [MLIR][TORCH] Add E2E support for aten.new_empty op
This commit decomposes `aten.new_empty` op into `aten.empty.memory_format` op.

This commit also made a dtype fix to the constant tensor allocation like ops.
Earlier the dtype for the result was inferred from the result type; now, it's
being evaluated as per the original definition of the op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-30 13:21:01 +05:30
Sean Silva 140babd952 Add minimal support for Union types.
A recent PyTorch commit made ConstantPad2d call a helper function with a
`Union[int, float]` type annotated. This commit adds minimal support for
representing and dealing with that.
https://github.com/pytorch/pytorch/pull/73287

Changes:
- Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`,
  along with the importer and CAPI code.
- Add support in isValidSubtype for union types.
- Adding a canonicalizer for `torch.derefine` to help simplify some code
  that derefines to a UnionType (this also fixes #664).

There is still more work to do for really supporting UnionType well,
such as canonicalizing UnionType's so that they can be compared with
pointer equality.
2022-03-29 17:45:48 -07:00
Sean Silva 4f61b1fce1 Try to get the release packages publishing again.
As per the docs on:
https://github.com/eregon/publish-release

> Note that the release must *not be marked as prerelease* for this to work.

For some reason, we were marking the release as pre-release before and
this was working, but the docs here seem pretty clear, so I'm going to
try it.
2022-03-30 00:35:02 +00:00
Sean Silva 3a96078571 Pin the CI to the latest working PyTorch.
I am investigating the breakage.

Also, fix "externals" rename in setup.py and some cases where we weren't
using `requirements.txt` consistently.

Also, fix a case where the packaging script would get confused due to
".." in the path name.
2022-03-29 15:02:17 -07:00
Liam Fitzpatrick f2269ced80
Improve list index normalization SimplifyShapeCalculations. (#710)
The reified code to compute the shape of torch.aten.constant_pad_nd
uses negative indices when setting list elements. This was not
converted to a positive offset in one place in SimplifyShapeCalculations
which prevented computation of the static shape.
2022-03-29 22:21:47 +02:00
Maksim Levental 25ba51b2af
This commit decomposes aten._reshape_alias op into aten.view op. (#690) 2022-03-28 23:54:28 -05:00
Maksim Levental eecbf0bab6
Eager mode description in the README and small example and ResNet18 example. (#707) 2022-03-28 23:54:06 -05:00
Sean Silva 520725cdc5 Fix bad rename from "pseudo" to "valsem". 2022-03-28 20:40:42 +00:00
Sean Silva 776426ea4e [SimplifyShapeCalculations] Fix AbstractlyInterpretListOpsWithinABlock
The logic in the rewriting phase had a bug in case of a read-only op
coming before mutation ops. The logic would use the op itself as the
"latest literal", but that is not correct, because later on we replace
the op itself with the *final* "latest literal", assuming that all uses
of the op have been rewritten -- that was working in general, except for
any read-only ops at the beginning.

Big thanks to @ljfitz for the tiny reproducer!

Fixes #704
2022-03-28 13:18:35 -07:00
Sean Silva 52c330cca2 Fix some more uses of "e2e" that I missed in the last commit. 2022-03-28 19:09:56 +00:00
Maksim Levental 3e999beaea
Small bug fixes in eager mode (#691) 2022-03-28 13:31:07 -05:00
Sean Silva 1960ba76fb Remove "e2e" name from `examples/torchscript_resnet18_e2e.py`
That was back from an earlier stage in the project when e2e was a big
deal because we didn't have anything working e2e yet :)
2022-03-28 18:26:54 +00:00
Sean Silva 0378c75b35 Centralize all test serialization logic. 2022-03-28 10:17:13 -07:00
Sean Silva e59a91620a Tidy up README and examples
- update diagram to use the name "Eager Mode" instead of
  `torch.dispatch`, which wasn't a very accurate name
- rename `resnet_inference.ipynb` to
  `torchscript_resnet_inference.ipynb` - this is in preparation to LTC
  and Eager Mode versions
- remove mention of TorchFX - turns out that all TorchFX modules are
  actually scriptable modules, so there is literally "zero code" vs
  using the TorchScript path
- remove LazyTensorCore example, and instead point at the current
  in-development `torch_mlir_ltc_backend` branch.

Note: there were actually some pretty useful utilities built out in the
examples directory, but they now live inside the Eager Mode
`python/torch_mlir/eager_mode/ir_building.py` (and need to be rolled
into a proper home with the upcoming rewrite of our top-level
`torch_mlir.compile` API).
2022-03-28 10:05:58 -07:00
Ahmed S. Taei 8383497704
[NFC] Rename external -> externals (#699) 2022-03-26 09:12:27 -07:00
Anup Gangwar 5d7a6c2976
[tosa] Support for Aten[Unsqueeze|Contiguous|Dropout|Reshape|View] ops (#700) 2022-03-25 14:15:07 -07:00
Sean Silva 6b637a9fd9 Move e2e test definitions into the `torch_mlir_e2e_test` package
This is the first step to making the e2e framework convenient to use
by downstream backends.
2022-03-25 13:56:41 -07:00
Vivek Khandelwal 88c216da13 [MLIR][TORCH] Add support for same input and output shapes for view op
This commit adds support for the cases of view op where the rank and
the shapes of the input and result are equal.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-25 22:26:10 +05:30
Gaurav Shukla 02b6d04eb4 [LINALG] Add E2E support for `aten.zero_` op
This commit adds decomposition of `aten.zero_` op.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-03-25 12:46:50 +05:30
Sean Silva 94df096c11
Add note to not edit upstream_shape_helpers.py 2022-03-24 09:32:19 -07:00
Prashant Kumar 730cdcd071 Add hugging face `albert-base-v2` in torchscript_e2e_heavydep_tests
`albert-base-v2` for sequence classification is added in e2e_heavy_test.
2022-03-24 17:43:24 +05:30
Ramiro Leal-Cavazos e966112c8d
Add final cast to TorchToLinalg conversions missing it (#692)
In order to make sure that the TorchToLinalg conversions leave the
graph in a valid state, the final result of the conversion has to be
casted to the result type of the op. This commit adds this cast to ops
that did not have it.
2022-03-23 13:52:32 -07:00
Qiang Fu f7c7bb800c
Add non-default dtype support for a few elementwise math ops. (#687)
* fix type inference
* fix Torch2Linalg conversion
* add test cases
2022-03-23 13:35:43 -07:00
max fe8ac57e6d This PR implements an eager mode backend for PyTorch through the torch-mlir framework. This is accomplished by overriding the `__torch_dispatch__` class method on wrapper subclass `TorchMLIRTensor(torch.Tensor)`.
Effectively, this mode works by compiling op by op as the NN is eagerly executed by PyTorch. Entailed in that compilation is building a representation of the op that can be `torch.jit.script`ed, importing using `ModuleBuilder`, and then executing (e.g., with `RefBackendLinalgOnTensorsBackend`). This mode includes a fallback to conventional PyTorch if anything in the torch-mlir compilation process fails (e.g., unsupported op).

Currently, all e2e tests pass execpt for two that involve an upstream PyTorch bug (https://github.com/pytorch/pytorch/issues/74400).

High priority next steps:

1. A compile cache in order to speed up reruns of the same NN.
2. Integration with IREE (though not in this repo).
3. Integration with `torch.distributed`.
2022-03-22 14:42:57 -07:00
Ahmed Taei f9d34596e8 [NFC] Split BackendTypeConversion -> (BackendTypeConversion, BackendTypeConversionPasses) 2022-03-22 13:56:18 -07:00
Sean Silva 6a7cf0c304 Update Torch-MLIR architecture diagram
Torch FX was never really a different path, since all FX modules are
actually valid TorchScript modules. Instead, replace it with the new
torch.dispatch work that we are building.
2022-03-22 11:51:52 -07:00
Gaurav Shukla 7c3ba25238 [LINALG] Add decomposition of `aten.dropout` op
- This commit adds decomposition of `aten.dropout` op. It also covers the
  training mode of the same op.
- It also adds lowering of `aten.sub.float` op.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-03-22 13:14:49 +05:30
Sean Silva 729402c3f4 Reduce compilation time for TorchOps.cpp.inc
The `assemblyFormat` stuff (which generates unrolled, per-op C++ code)
was taking up a lot of compile time, and all the ops are essentially
printed with the same logic. So this PR makes them all call the same
helper function. This is done by using
`let hasCustomAssemblyFormat = 1` and then implementing `FooOp::parse`
and `FooOp::print`.

Additionally, the `Generated*Ops.td` files are all collapsed into just
`GeneratedTorchOps.td` (there is no reason to have the files separate,
since the files are very large anyway so one is always having to search
within them -- editors don't care that the file to search is now a bit
bigger :) ).

This reduces TorchOpsODSGenerated.cpp compile time (which is now
GeneratedTorchOps.cpp) from 39 to 31 seconds on my machine. This is
actually less than I expected, but this PR is an overall cleanup to the
code anyway. The next step will be to introduce (better) functionality
upstream for sharding the TorchOps.cpp.inc file, so that we can truly
parallelize the O(#ops) costs. This is also necessary, because after
this PR, TorchDialect.cpp is now the slowest file to compile, due to the
`addOperations<... all the ops ...>` call, which needs to be shareded
too.
2022-03-21 14:42:26 -07:00
Vivek Khandelwal 5b9bdfaf3f [MLIR][TORCH] Add E2E support for aten._to_copy op
This commit decomposes `aten._to_copy` op into
`valsem.aten.copy` op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-21 19:12:37 +05:30
Vivek Khandelwal 13383b03b8 [MLIR][TORCH] Add value tensor variant to aten::copy_ op
This commit adds the op `ValsemVariantAtenCopyOp` that represents
`AtenCopy_Op` without the underscore. This is needed to make sure
that the `ReduceOpVariants` pass turns the in-place op into an op
that takes value tensors as inputs, otherwise the
`MaximizeValueSemantics` pass will not be able to add value
semantics correctly.

This commit also adds the lowering of `ValsemVariantAtenCopyOp`.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-21 19:12:37 +05:30
Vivek Khandelwal 4c0cd5c23d [MLIR][TORCH] Add E2E support for aten.expand_as op
This commit decomposes `aten.expand_as` op into `aten.broadcast_to` op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-21 12:47:39 +05:30
Vigilans 63fb1e5aad Bump LLVM at 8361c5da30588d3d4a48eae648f53be1feb5cfad 2022-03-18 13:16:14 -04:00
Ramiro Leal-Cavazos 218b4875d5
Make conditions for type refinement of static cast less strict (#680)
This commit adds support for type refinement when
`torch.tensor_static_info_cast`s are involved, even when there are
users of the casted tensor that don't allow type refinements.

Originally the canonicalization pattern for
`torch.tensor_static_info_cast` would check if all the users of the
casted tensor allowed type refinements before making any changes. This
means that if at least one of the users did not allow type
refinements, the pattern would fail. This becomes an issue when doing
shape calculations because the calculations need the shape information
of each input tensor to be available before the calculation can be
simplified.
2022-03-18 09:10:12 -07:00
Prateek Gupta 7256c9e395 [TORCH][MLIR] Fix the return types of `aten.native_layer_norm`.
This commit fixes the 2nd and 3rd return types of the `aten.native_layer_norm`.
Previously the mean and rSTD were returned with reduction dims removed.
This commit fixes this and keeps the reduction dims of the results.

Signed-Off-By: Prateek Gupta <prateek@nord-labs.com>
2022-03-17 12:08:32 +05:30
Sean Silva 3b66b4925a Make TorchOps.cpp faster to iterate on.
The ODS-generated code included via the `TorchOps.cpp.inc` file takes a
very long time to compile. This PR isolates it into its own file so that
the build system can cache it.

This PR creates a new file `TorchOpsODSGenerated.cpp` just to include
the `TorchOps.cpp.inc` file. Doing so required moving to the "new" way
to define verifiers, since the static `verify` free functions in
TorchOps.cpp weren't accessible from the .inc file after it was moved to
`TorchOpsODSGenerated.cpp`.

On my machine, this drops the build time of TorchOps.cpp (such as when
iterating on a canonicalizer) from >40 seconds to <10 seconds.
10 seconds still isn't great though, but at least it isn't "go get a
coffee" type of waiting.
2022-03-16 09:33:12 -07:00
Vivek Khandelwal 8da7d90611 [MLIR][TORCH] Add E2E support for aten.index_put op
This commit decomposes `aten.index_put` op into
`valsem.aten.index_put_impl` op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-16 22:02:02 +05:30
Vivek Khandelwal 3d95c3d6c9 [MLIR][TORCH] Add value tensor variant to aten::_index_put_impl_
This commit adds the op `ValsemVariantAtenIndexPutImplOp` that represents
`Aten_IndexPutImpl_Op` without the underscore. This is needed to
make sure that the `ReduceOpVariants` pass turns the in-place op
into an op that takes value tensors as inputs, otherwise the
`MaximizeValueSemantics` pass will not be able to add value
semantics correctly.

This commit also adds the lowering of `ValsemVariantAtenIndexPutImplOp` op.

This commit also updates the `torch.bincount` op test cases.
2022-03-16 22:02:02 +05:30
Yi Zhang 8a4388ea7b Fix convert_to_loops.mlir format 2022-03-16 11:42:37 -04:00
Ramiro Leal-Cavazos 0bcc6d1075
Add maximize-value-semantics support for multiple non-value tensor inputs (#659)
This commit adds value semantics support for ops such as
`aten.view_as` and `aten.expand_as` that take two non-value 
tensors as input.
2022-03-15 18:13:45 -07:00
Sean Silva 92da4988f0 Improve "pseudo" op terminology.
The term "pseudo" is very vague and was getting confusing (I felt I had
to explain it in every comment referencing it). Instead, rework the
"pseudo" ops to instead be named:

- MLIR Syntax: `torch.valsem.*`
- C++ / ODS: `ValsemVariant*Op`

This makes it clear what the concept is, and avoids confusion with other
things that might be called "pseudo", since these are very specific and
should be 100% consistently named w.r.t. the non-valsem-variant ops that
they correspond to.
2022-03-15 17:57:52 -07:00
Sean Silva 7ea50a537a Avoid `using` the `torch_upstream` namespace.
This is code that we always want to treat as "foreign" and not get too
comfortable using in many functions. One way to accomplish that is to
make it a bit clunkier to use.

Also, fix Utils.cpp to match the LLVM/MLIR coding conventions (don't
define functions inside namespaces -- prefer `using` and explicit
qualification).
2022-03-15 17:24:17 -07:00
Sean Silva 84a9693006 Elide `!torch.` prefix in nested dialect types.
This leads to much more succinct types in many cases:

```
!torch.list<!torch.int>
!torch.list<int>

!torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>>
!torch.tuple<list<int>, list<int>>

!torch.optional<!torch.list<!torch.int>>
!torch.optional<list<int>>

!torch.list<list<list<tensor>>>
!torch.list<!torch.list<!torch.list<!torch.tensor>>>
```

I would like to take this further and allow omitting the `!torch.`
prefix in all cases, but that's harder -- for example, we currently use
`FuncOp` for functions, and so I don't think we can customize the
printing there. It seems like it will be a longer road to getting that
level of customization.
2022-03-15 17:24:08 -07:00
Sean Silva 3734f69119 Remove basic_mt from the heavydep tests
This was an aspirational goal at an earlier stage in the project where
the focus was heavily on programs with state, control flow, and
lists/dicts. We will circle back to such programs likely 2022H2 at some
point, but for now, having this test doesn't add much, since basically
nothing works or is being worked on.
2022-03-15 15:25:53 -07:00
Sean Silva a5fe0cf063 Introduce new shape library design.
See the documentation in `docs/shape_lib.md` and
`docs/adding_a_shape_function.md` for an overview of the system.

This completely overhauls how we represent shape functions. In
particular, RefineTypes does not infer shapes anymore (only dtypes).
Shape functions are now written in (TorchScript'able) Python.

Recommended review order:

1. Read `docs/shape_lib.md` and `docs/adding_a_shape_function.md`.
1. Code and tests for ReifyShapeCalculations, DropShapeCalculations.
1. Code and tests for SimplifyShapeCalculations.
1. shape_lib_gen.py
1. Code and tests for new RefineTypes pass.
1. Random folders/canonicalizers in TorchOps.cpp and associated test in
   `canonicalize.mlir`.
1. New ReadOnly trait inferred from the registry.
1. Any miscellaneous remaining stuff.

Example `-print-ir-after-all` for ElementwiseUnaryModule:
[IR lowering dump](https://gist.github.com/silvasean/e4dc8cbc8d00aac7819602e3cbd8e212).

Example `-print-ir-after-all` for ElementwiseBinaryModule:
[IR lowering dump](https://gist.github.com/silvasean/daf6860ecced732af3568af6b1899113).
2022-03-15 12:41:58 -07:00
Sean Silva 5d9222383c Split up TorchToLinalg.cpp
This helps keep things organized and also exposes more parallelism to
the build system. It seems though that most of the compile time is
actually spent in the headers though, so the wall time doesn't decrease
as much as I had hoped (and now that the headers are being included
multiple times, the cpu time actually increases a lot, sadly -- will try
to dig into this).
2022-03-14 10:19:41 -07:00
Prashant Kumar b6d13301fc [TORCH] Fix the location of packed_params.
The location of packed_params.h is changed in aten src.
2022-03-14 17:52:19 +05:30
Ramiro Leal-Cavazos 51e267aa37
Combine maximize-value-semantics rewrite patterns into one pattern (#642)
This commit replaces the two rewrite patterns of
maximize-value-semantics with a single pattern that captures the
behavior of both as well as other edge cases previously not
supported. The new pattern works by first performing alias analysis on
a subgraph to see if pattern is applicable, then rewriting all
non-value tensors to value tensors in a single go.
2022-03-10 09:36:52 -08:00
Yi Zhang 3510b2ba9d Fix scatter op bufferization to alway copy original tensor 2022-03-09 18:19:44 -05:00
Prateek Gupta 3d9ba5e525 [MLIR][TORCH] Add E2E support for aten.erf op.
Signed-Off-By: Prateek Gupta <prateek@nod-labs.com>
2022-03-09 22:22:03 +05:30
Vivek Khandelwal 1a2a9e066f [MLIR][TORCH] Add TorchToTMTensor pass
This pass is added to lower ops, which can not be lowered
via the TorchToLinalg pass, such as `torch.bincount` op.
This pass also uses torch-mlir's TMTensor Dialect to lower the
complex ops.

Also add torch.bincount op lowering with the help of TMTensor dialect

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-08 22:52:34 +05:30