bae0815ef0
[MLIR][KernelGen] Legalize `atan` to approximation
...
PiperOrigin-RevId: 335417836
2020-10-05 08:05:52 -07:00
7f84a86cf5
[MLIR][KerneGen] Lower `tf.Atan` all the way to LLVM
...
PiperOrigin-RevId: 335394668
2020-10-05 05:07:13 -07:00
c9f4755863
Integrate LLVM at llvm/llvm-project@22664a3251
...
Updates LLVM usage to match
[22664a325167](https://github.com/llvm/llvm-project/commit/22664a325167 )
PiperOrigin-RevId: 335391717
2020-10-05 04:42:21 -07:00
d927e32451
[HLO] Clean-up dynamic allocation in hlo-legalize-to-lhlo pass.
...
PiperOrigin-RevId: 335385243
2020-10-05 03:55:24 -07:00
7367eac074
Add folder for mhlo::remainder
...
PiperOrigin-RevId: 335372628
2020-10-05 02:20:01 -07:00
c9e249c124
Integrate LLVM at llvm/llvm-project@88c9162c9d
...
Updates LLVM usage to match
[88c9162c9d47](https://github.com/llvm/llvm-project/commit/88c9162c9d47 )
PiperOrigin-RevId: 335144150
2020-10-02 19:32:46 -07:00
5146c29cbf
Integrate LLVM at llvm/llvm-project@0f0cbcc4b1
...
Updates LLVM usage to match
[0f0cbcc4b166](https://github.com/llvm/llvm-project/commit/0f0cbcc4b166 )
PiperOrigin-RevId: 335119390
2020-10-02 16:08:17 -07:00
c708bfd6d0
[MLIR] Add cbrt, reduce-precision, and bitcast ops to MHLO.
...
PiperOrigin-RevId: 335109804
2020-10-02 15:13:18 -07:00
bcf6fbf612
Integrate LLVM at llvm/llvm-project@8825fec37e
...
Updates LLVM usage to match
[8825fec37e73](https://github.com/llvm/llvm-project/commit/8825fec37e73 )
Adapt to upstream subview / subtensor changes.
PiperOrigin-RevId: 335023452
2020-10-02 07:34:47 -07:00
a3a8c498e1
Integrate LLVM at llvm/llvm-project@bfd7ee92cc
...
Updates LLVM usage to match
[bfd7ee92ccec](https://github.com/llvm/llvm-project/commit/bfd7ee92ccec )
PiperOrigin-RevId: 335003324
2020-10-02 04:39:31 -07:00
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
5f303440da
[MLIR][KerneGen] Lower `tf.Atan` all the way to LLVM
...
PiperOrigin-RevId: 334843070
2020-10-01 10:25:51 -07:00
aa65e49ad2
Integrate LLVM at llvm/llvm-project@5101e7e8dd
...
Updates LLVM usage to match
[5101e7e8dd01](https://github.com/llvm/llvm-project/commit/5101e7e8dd01 )
PiperOrigin-RevId: 334812211
2020-10-01 07:49:58 -07:00
458e861254
[MLIR][KerneGen] Lower `tf.Atan` all the way to LLVM
...
PiperOrigin-RevId: 334810730
2020-10-01 07:38:40 -07:00
73f461080c
Integrate LLVM at llvm/llvm-project@fcf70e1e3b
...
Updates LLVM usage to match
[fcf70e1e3b1d](https://github.com/llvm/llvm-project/commit/fcf70e1e3b1d )
PiperOrigin-RevId: 334796954
2020-10-01 05:58:50 -07:00
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
4b1809784a
Support collapse_slice_dims in the mhlo.gather->mhlo.slice canonicalizer
...
PiperOrigin-RevId: 334774763
2020-10-01 02:46:49 -07:00
ae51900562
Integrate LLVM at llvm/llvm-project@ce5379f0f0
...
Updates LLVM usage to match
[ce5379f0f067](https://github.com/llvm/llvm-project/commit/ce5379f0f067 )
PiperOrigin-RevId: 334774614
2020-10-01 02:45:28 -07:00
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
e2ffba3f61
Integrate LLVM at llvm/llvm-project@700e63293e
...
Updates LLVM usage to match
[700e63293eea](https://github.com/llvm/llvm-project/commit/700e63293eea )
PiperOrigin-RevId: 334732681
2020-09-30 20:13:43 -07:00
dfe64d3958
Implement InferShapedTypeOpInterface for mhlo.complex
...
Binary companion for 8bcd33e4b7
2020-09-30 12:14:15 -07:00
019c5ef106
Integrate LLVM at llvm/llvm-project@8c05c7c8d8
...
Updates LLVM usage to match
[8c05c7c8d87c](https://github.com/llvm/llvm-project/commit/8c05c7c8d87c )
PiperOrigin-RevId: 334598568
2020-09-30 07:57:41 -07:00
c8919f8419
Implement InferShapedTypeOpInterface and use inferReturnTypes for mhlo.imag and mhlo.real
...
This makes the lhlo lowering work with dynamic shapes.
PiperOrigin-RevId: 334553472
2020-09-30 02:02:26 -07:00
39389587d2
{mhlo.is_finite, lmhlo.is_finite} -> linalg.generic conversion
...
PiperOrigin-RevId: 334414295
2020-09-29 10:50:42 -07:00
336ee14538
Make lmhlo.sign to linalg lowering work for more floating point and integer types
...
PiperOrigin-RevId: 334400341
2020-09-29 09:52:13 -07:00
04bf09382e
Integrate LLVM at llvm/llvm-project@eb9f7c28e5
...
Updates LLVM usage to match
[eb9f7c28e5fe](https://github.com/llvm/llvm-project/commit/eb9f7c28e5fe )
PiperOrigin-RevId: 334388404
2020-09-29 08:54:26 -07:00
6459f12235
Lower mhlo.not to a xor with all ones
...
PiperOrigin-RevId: 334361499
2020-09-29 05:59:40 -07:00
26ac5baae4
Make mhlo.sort return variadic results instead of a tuple
...
Tuple is only used on XLA's sort to return multiple inputs. MLIR supports
multiple inputs, switch to a tuple return.
PiperOrigin-RevId: 334226937
2020-09-28 13:32:23 -07:00
be2ffd2e21
Integrate LLVM at llvm/llvm-project@6caf3fb817
...
Updates LLVM usage to match
[6caf3fb81786](https://github.com/llvm/llvm-project/commit/6caf3fb81786 )
PiperOrigin-RevId: 333820612
2020-09-25 15:28:22 -07:00
eec7eb72fc
Integrate LLVM at llvm/llvm-project@81a408808f
...
Updates LLVM usage to match
[81a408808f66](https://github.com/llvm/llvm-project/commit/81a408808f66 )
PiperOrigin-RevId: 333600462
2020-09-24 14:37:27 -07:00
9125701a38
Integrate LLVM at llvm/llvm-project@ada1e2ffa1
...
Updates LLVM usage to match
[ada1e2ffa117](https://github.com/llvm/llvm-project/commit/ada1e2ffa117 )
PiperOrigin-RevId: 333566305
2020-09-24 12:02:43 -07:00
4ddca0cd03
Integrate LLVM at llvm/llvm-project@a90d72127a
...
Updates LLVM usage to match
[a90d72127a81](https://github.com/llvm/llvm-project/commit/a90d72127a81 )
PiperOrigin-RevId: 333488152
2020-09-24 04:27:08 -07:00
2e142a685c
Integrate LLVM at llvm/llvm-project@beeceb92c0
...
Updates LLVM usage to match
[beeceb92c082](https://github.com/llvm/llvm-project/commit/beeceb92c082 )
PiperOrigin-RevId: 333459746
2020-09-24 00:34:31 -07:00
9c6640cbb6
Only apply GeneralDotOpLoweringPatterns for static shaped inputs
...
PiperOrigin-RevId: 333439680
2020-09-23 21:42:01 -07:00
34c6844dcc
Integrate LLVM at llvm/llvm-project@af1d3e6559
...
Updates LLVM usage to match
[af1d3e655991](https://github.com/llvm/llvm-project/commit/af1d3e655991 )
PiperOrigin-RevId: 333385949
2020-09-23 15:31:15 -07:00
233f1a8a1a
Folders for mhlo.compare
...
Constant evaluation of compare for the case where inputs are either the same
variable or the values are constant.
PiperOrigin-RevId: 333342328
2020-09-23 12:03:48 -07:00
0259f982bf
Integrate LLVM at llvm/llvm-project@2d0de5f9a4
...
Updates LLVM usage to match
[2d0de5f9a4c5](https://github.com/llvm/llvm-project/commit/2d0de5f9a4c5 )
PiperOrigin-RevId: 333341699
2020-09-23 12:01:39 -07:00
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
e13904c92a
Integrate LLVM at llvm/llvm-project@d6ac649ccd
...
Updates LLVM usage to match
[d6ac649ccda2](https://github.com/llvm/llvm-project/commit/d6ac649ccda2 )
PiperOrigin-RevId: 333316098
2020-09-23 10:10:39 -07:00
3ca740103d
[NFC] Specify "reifyReturnTypeShapes" in DeclareOpInterfaceMethod overridenMethods
...
- This causes declaration of reifyReturnTypeShapes to be inserted in the TableGen
generated code without explicitly adding it in extraClassDeclaration.
PiperOrigin-RevId: 333315601
2020-09-23 10:08:32 -07:00
87d9e6951e
[MLIR] Use MLIR provided 0DTensorOf/1DTensorOf instead of custom constraints.
...
- Use MLIR provided constraints for HLO_ScalarIntTensor and HLO_DimensionTensor.
- Update unit tests to expect new error messages.
PiperOrigin-RevId: 333313131
2020-09-23 09:57:12 -07:00
7d01a60de8
[NFC] Fix typos in comments.
...
PiperOrigin-RevId: 333311070
2020-09-23 09:46:02 -07:00
08514eaa5f
[MLIR][KernelGen] Add E2E tests for `tf.Tan` and `tf.Sinh`
...
PiperOrigin-RevId: 333275089
2020-09-23 05:50:54 -07:00
be409a2409
Integrate LLVM at llvm/llvm-project@f212122150
...
Updates LLVM usage to match
[f21212215031](https://github.com/llvm/llvm-project/commit/f21212215031 )
PiperOrigin-RevId: 333168406
2020-09-22 15:08:10 -07:00
3e3bc705ed
Integrate LLVM at llvm/llvm-project@079757b551
...
Updates LLVM usage to match
[079757b551f3](https://github.com/llvm/llvm-project/commit/079757b551f3 )
PiperOrigin-RevId: 333132011
2020-09-22 12:12:05 -07:00
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
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
0bb017098a
Integrate LLVM at llvm/llvm-project@93fd30bac3
...
Updates LLVM usage to match
[93fd30bac334](https://github.com/llvm/llvm-project/commit/93fd30bac334 )
PiperOrigin-RevId: 332436415
2020-09-18 06:05:49 -07:00
4002077261
[MLIR][KernelGen] Lower `tf.Sinh` to MLHLO
...
PiperOrigin-RevId: 332425724
2020-09-18 04:27:07 -07:00
cfbd7a18b2
Integrate LLVM at llvm/llvm-project@c0e7f64685
...
Updates LLVM usage to match
[c0e7f6468578](https://github.com/llvm/llvm-project/commit/c0e7f6468578 )
PiperOrigin-RevId: 332420028
2020-09-18 03:30:34 -07:00
A. Unique TensorFlower
Alexander Belyaev
Tim Shen
Tres Popp
Adrian Kuegel
Chao Xie
Rahul Joshi
Benjamin Kramer
Ahmed S. Taei
Robert Suderman
Mehdi Amini