Commit Graph

116 Commits

Author SHA1 Message Date
A. Unique TensorFlower bba9968ff0 Integrate LLVM at llvm/llvm-project@499bce3aba
Updates LLVM usage to match
[499bce3abab8](https://github.com/llvm/llvm-project/commit/499bce3abab8)

PiperOrigin-RevId: 342921326
2020-11-17 12:30:18 -08:00
A. Unique TensorFlower 7a6a6ffa40 Integrate LLVM at llvm/llvm-project@65f3e121fe
Updates LLVM usage to match
[65f3e121fe4f](https://github.com/llvm/llvm-project/commit/65f3e121fe4f)

PiperOrigin-RevId: 342859791
2020-11-17 07:38:03 -08:00
Stephan Herhut 0c7152e65c Extend fusion root heuristic to also work in partially bufferized programs.
We now follow data flow though tensor_cast, tensor_load and tensor_to_memref
operations.

PiperOrigin-RevId: 342851104
2020-11-17 06:34:48 -08:00
Tres Popp be9ae88eaa Consider aliases through control flow for lhlo-fuse-linalg.
This should handle scf.if and shape.assuming regions,
which we care about in kernel_gen.

PiperOrigin-RevId: 342584762
2020-11-16 01:12:45 -08:00
Ben Vanik 88a5bde51f Fixing M_PI compile error on MSVC.
M_PI and other math constants (used in chlo_legalize_hlo_patterns.td)
are not part of the C++ standard and must be enabled on MSVC
(similar to _GNU_SOURCE adding glibc symbols to posix headers).

PiperOrigin-RevId: 342432987
2020-11-14 10:02:04 -08:00
Smit Hinsu d0901a83e5 Enable fallback legalization for MaxPoolGradGrad and MaxPool3DGradGrad ops
Requires,
* Override for ReducePrecision in HloMlirBuilder
* Sinking of constants for ReduceWindow op

PiperOrigin-RevId: 342330848
2020-11-13 13:49:05 -08:00
Stephan Herhut c344695bca Properly configure patterns in hlo to lhlo conversion with typeconverter.
Otherwise, materializations for partial conversions do not work.

PiperOrigin-RevId: 342215962
2020-11-13 00:58:51 -08:00
Rahul Joshi 1958f228ec [MLIR:HLO] Extend CustomCall to support multiple outputs.
- Extend MHLO CustomCall to have multiple tensors as results.
- Extend LHLO CustomCall to have multiple memrefs for output operands.
- Fix HLO->LHLO and XLA HLO->LHLO mapping for CustomCall to setup the
  operand_segment_sizes attribute correctly.

PiperOrigin-RevId: 342067762
2020-11-12 09:46:25 -08:00
Rahul Joshi 745c8aa0b1 Extract some duplicated code into a helper function.
- Extract code to create result memref's into a ConvertResults function.
- Also fix a bug when using reifyReturnTypes: use correct index for result_shape instead
  of always using the first element.

PiperOrigin-RevId: 341852227
2020-11-11 10:01:00 -08:00
Alexander Belyaev d4f2c767d3 [HLO] Fix HLO DynamicBroadcastInDimOp -> LHLO lowering.
The conversion had a bug in computation of strides and sizes args for std.memref_reinterpret_cast. The previous version also relied on linalg::ReshapeOp to do broadcasting when the rank of the output was higher than the rank of the input. Now the broadcasting is entirely done via descriptor modification and linalg::ReshapeOp was replaced with CopyOp.

PiperOrigin-RevId: 341379871
2020-11-09 04:24:40 -08:00
Robert Suderman a926e0f040 Removed Op(Complex, Real) lowering to address complex type inference issue
Lowerings that depended on operations between real and complex types may
not infer the correct intermediate type. Removing these operations as
they are not technically legally generated operations. Updated tests
to validate this.

PiperOrigin-RevId: 341128903
2020-11-06 15:24:23 -08:00
Tres Popp af4c9774dc Handle rank 1 broadcasts in unranked kernel lowering.
Previously this started at rank 2 after checking for scalars and equal shapes. This resulted in cases such as <1xf32> + <2xf32> being treated as impossible.

PiperOrigin-RevId: 341043965
2020-11-06 07:22:43 -08:00
Alexander Belyaev 3d930d08c2 [HLO] Delete LHLO memref cast ops and migrate to STD ones.
PiperOrigin-RevId: 340663578
2020-11-04 09:26:34 -08:00
Marius Brehler ff9b8c6f65 PR #44499: Add missing dep on MLIRMhloPassIncGen target
Imported from GitHub PR https://github.com/tensorflow/tensorflow/pull/44499

The file `sink_constants_to_control_flow.cc` includes the header
`PassDetail.h`, which itself includes `mhlo_passes.h.inc`. The latter is
not guaranteed to be already generated since there was no dependency set
to MLIRMhloPassIncGen.
Copybara import of the project:

--
0ff51ccc88c1ba049eb2e9555afb54079bea39c9 by Marius Brehler <marius.brehler@iml.fraunhofer.de>:

Add missing dep on MLIRMhloPassIncGen target

The file `sink_constants_to_control_flow.cc` includes the header
`PassDetail.h`, which itself includes `mhlo_passes.h.inc`. The latter is
not guaranteed to be already generated since there was no dependency set
to MLIRMhloPassIncGen.

PiperOrigin-RevId: 340485068
2020-11-03 11:18:48 -08:00
Sean Silva d3ea3abdec Remove `results_escape_functions` from HloLegalizeToLhlo
PiperOrigin-RevId: 340464958
2020-11-03 09:49:56 -08:00
Tres Popp 81e8d778c4 Fix bug using std.rank instead of shape.rank
PiperOrigin-RevId: 339890070
2020-10-30 09:59:24 -07:00
Tres Popp 76b30fd426 Move unranked chlo lowering to transform_unranked_hlo.
Additionally:
- Forward listeners through new if/else op builders.
This corrects an error that led to incomplete legalization of broadcasted op
lowering.
- Use OpConversionPattern to ensure up to date operand values are used.
PiperOrigin-RevId: 339838833
2020-10-30 02:56:44 -07:00
Nicolas Vasilache 880f603239 Drop OperationFolder usage with Linalg fusion.
PiperOrigin-RevId: 339653466
2020-10-29 06:10:14 -07:00
Benjamin Kramer 3bf4277ea4 [MLIR] Add a lmhlo.reduce -> linalg.generic converter
Doesn't support tensors right now, as it's somewhat hairy to support both at
the same time. Since we use a generic lowering the result is messy
and needs a mem2reg pass to eliminate extra load/store/allocas.

PiperOrigin-RevId: 339562971
2020-10-28 16:38:19 -07:00
Thomas Joerg 7363748bae Integrate LLVM at llvm/llvm-project@0fc1aa22ee
Updates LLVM usage to match
[0fc1aa22ee6a](https://github.com/llvm/llvm-project/commit/0fc1aa22ee6a)

PiperOrigin-RevId: 339239851
2020-10-27 06:56:16 -07:00
Benjamin Kramer b7aa01dbe0 Integrate LLVM at llvm/llvm-project@26750a1264
Updates LLVM usage to match
[26750a1264b3](https://github.com/llvm/llvm-project/commit/26750a1264b3)

PiperOrigin-RevId: 339127601
2020-10-26 15:11:52 -07:00
Smit Hinsu 6eda9ed273 Add compare_type optional attribute to CompareOp in HLO dialects
If unspecified, `compare_type` is FLOAT for float element types, SIGNED for signed element types and UNSIGNED for unsigned element types. compare_type can be TOTALORDER for float element types.

- Added import and export support the attribute.
- Restricted legalization from HLO to TF to the default compare types.
- Updated existing usage of the CompareOp

PiperOrigin-RevId: 339099219
2020-10-26 12:58:29 -07:00
Hanhan Wang 444fae9bac [NFC] Make naming style consistent.
Use lowercase with underscores between words instead of camelStyle.

PiperOrigin-RevId: 338722328
2020-10-23 12:23:25 -07:00
Sean Silva c3f4ab06d3 Use OpConversionPattern instead of BufferizeOpConversionPattern
As described in mlir/Transforms/Bufferize.h, patterns that don't need the special methods on a BufferizeTypeConverter should use a regular OpConversionPattern.

PiperOrigin-RevId: 338424819
2020-10-22 01:20:43 -07:00
Stephan Herhut 2e30b59ddc Extend hlo-fuse-linalg slightly to support aliased returns.
The fusion heuristic identifies the root of a fusion by checking whether an
output of a linalg operation is a function result. It did not consider outputs
flowing through aliasing operations (like casts).

PiperOrigin-RevId: 337479910
2020-10-16 04:04:23 -07:00
A. Unique TensorFlower 51cd4200b6 Make LMHLO's Dot have the same power as MHLO's DotGeneral.
PiperOrigin-RevId: 337391565
2020-10-15 15:09:06 -07:00
Adrian Kuegel 0f36979e2c Integrate LLVM at llvm/llvm-project@220de1f32a
Updates LLVM usage to match
[220de1f32add](https://github.com/llvm/llvm-project/commit/220de1f32add)

PiperOrigin-RevId: 337270545
2020-10-15 02:54:24 -07:00
A. Unique TensorFlower 15fe5ca737 Integrate LLVM at llvm/llvm-project@6713332fdd
Updates LLVM usage to match
[6713332fddb7](https://github.com/llvm/llvm-project/commit/6713332fddb7)

PiperOrigin-RevId: 336988895
2020-10-13 17:21:14 -07:00
Adrian Kuegel 53d16d9f8e Integrate LLVM at llvm/llvm-project@93377888ae
Updates LLVM usage to match
[93377888ae89](https://github.com/llvm/llvm-project/commit/93377888ae89)

PiperOrigin-RevId: 336669842
2020-10-12 08:42:31 -07:00
Tres Popp f6af1fc134 Support hlo to lhlo buffer placement through shape.assuming ops.
PiperOrigin-RevId: 336287728
2020-10-09 07:14:12 -07:00
Stephan Herhut d986bd7ad7 Use tensor_cast instead of mhlo::reshape in the lowering of unranked binary operations.
We know that the value already is a scalar and we just want to update the type, so no need to reshape anything.

PiperOrigin-RevId: 336252315
2020-10-09 01:46:48 -07:00
A. Unique TensorFlower 3736c5542f [MLIR][KernelGen] Fix unranked codegeneration in kernel generator
PiperOrigin-RevId: 335847086
2020-10-07 05:39:55 -07:00
Lucy Fox 84277ed784 Remove unused `using` statements.
PiperOrigin-RevId: 335517711
2020-10-05 15:53:59 -07:00
A. Unique TensorFlower a33441907c [MLIR][KernelGen] Add E2E test for `tf.Acos`
PiperOrigin-RevId: 335419310
2020-10-05 08:14:36 -07:00
A. Unique TensorFlower bae0815ef0 [MLIR][KernelGen] Legalize `atan` to approximation
PiperOrigin-RevId: 335417836
2020-10-05 08:05:52 -07:00
A. Unique TensorFlower 7f84a86cf5 [MLIR][KerneGen] Lower `tf.Atan` all the way to LLVM
PiperOrigin-RevId: 335394668
2020-10-05 05:07:13 -07:00
Alexander Belyaev d927e32451 [HLO] Clean-up dynamic allocation in hlo-legalize-to-lhlo pass.
PiperOrigin-RevId: 335385243
2020-10-05 03:55:24 -07:00
Adrian Kuegel 3eb767b43d Add missing lowering step for IsFiniteOp.
Also add a BUILD target for generating the GPU kernel.

PiperOrigin-RevId: 334993362
2020-10-02 03:09:00 -07:00
Chao Xie 5f303440da [MLIR][KerneGen] Lower `tf.Atan` all the way to LLVM
PiperOrigin-RevId: 334843070
2020-10-01 10:25:51 -07:00
A. Unique TensorFlower 458e861254 [MLIR][KerneGen] Lower `tf.Atan` all the way to LLVM
PiperOrigin-RevId: 334810730
2020-10-01 07:38:40 -07:00
A. Unique TensorFlower 049ca060a1 [MLIR][KernelGen] Legalize `atan2` to approximation
Legalize `atan2` analogously to XLA.  `atan2` is first reduced to `atan` on the
interval [-1, 1] and subsequently approximated.  This CL also adds e2e tests for
trigonometric approximations.

PiperOrigin-RevId: 334794336
2020-10-01 05:34:48 -07:00
Rahul Joshi bce128b070 Introduce CustomCall operation in LHLO Dialect
- And add conversion from MHLO CustomCall to LHLO CustomCall
- According to XLA documentation, the called function should not be side effecting,
  so marking the argument MemRefs as MemRead.

PiperOrigin-RevId: 334737196
2020-09-30 20:56:18 -07:00
Ahmed S. Taei 39389587d2 {mhlo.is_finite, lmhlo.is_finite} -> linalg.generic conversion
PiperOrigin-RevId: 334414295
2020-09-29 10:50:42 -07:00
Benjamin Kramer 6459f12235 Lower mhlo.not to a xor with all ones
PiperOrigin-RevId: 334361499
2020-09-29 05:59:40 -07:00
Ahmed S. Taei 9c6640cbb6 Only apply GeneralDotOpLoweringPatterns for static shaped inputs
PiperOrigin-RevId: 333439680
2020-09-23 21:42:01 -07:00
A. Unique TensorFlower 08e0d09463 [MLIR][KernelGen] Rename `legalize-tanh-to-approximation` to `legalize-trigonometric-to-approximation`
To add more approximation lowerings in the future, generalize the pass name.

PiperOrigin-RevId: 333340075
2020-09-23 11:53:45 -07:00
Benjamin Kramer dd92c8ef61 Integrate LLVM at llvm/llvm-project@7e78d89052
Updates LLVM usage to match
[7e78d89052b1](https://github.com/llvm/llvm-project/commit/7e78d89052b1)

PiperOrigin-RevId: 333090785
2020-09-22 09:08:23 -07:00
Mehdi Amini 7abd557a61 Add a header for table-gen generated pass for MHLO and use it in SinkConstantsToControlFlowPass
A non globally registered pass should define `getName()` in order to generate correct crash reproducers.
This is something we get "for free" when using the TableGen generated base class.
We should also migrate the other passes to the same mechanism and remove the static
global registration.

PiperOrigin-RevId: 332976907
2020-09-21 18:01:57 -07:00
A. Unique TensorFlower 4002077261 [MLIR][KernelGen] Lower `tf.Sinh` to MLHLO
PiperOrigin-RevId: 332425724
2020-09-18 04:27:07 -07:00
A. Unique TensorFlower 2fbbbe9cf1 [MLIR][KernelGen] Lower `tf.Acos` to LMHLO.
- Add ranked code generation for `mhlo.compare/select`
- Add bufferization for `tensor_cast`
- Add lowerings for `Atan2Op`

PiperOrigin-RevId: 332407734
2020-09-18 01:40:18 -07:00