557e56362e
[MLIR][KernelGen] Simplify rank specialization tests with smaller target rank
...
For the tests rank specialize only up to rank 3. The remaining cases for higher
ranks are analogous.
PiperOrigin-RevId: 377024370
2021-06-02 03:48:07 -07:00
968d4b8709
PR #49598 : [MLIR][DISC] legalize tensor_load inserted during hlo-to-lhlo conversion
...
Imported from GitHub PR https://github.com/tensorflow/tensorflow/pull/49598
This PR implements logic for lowering memref.tensor_load ops that are
inserted during `mhlo-legalize-to-lmhlo`
Copybara import of the project:
--
80eb377af4e02182e1aecc943a41ca5d7d1c2100 by Wenyi Zhao <reyizero@gmail.com>:
[MLIR][DISC] legalize tensor_load inserted during hlo-to-lhlo conversion
This PR implements logic for lowering memref.tensor_load ops that are
inserted during `mhlo-legalize-to-lmhlo`.
--
ac452fe3dcd591211cd5c59be9189fe2f7153b41 by Wenyi Zhao <reyizero@gmail.com>:
minor fix
--
6b36017f8632a06adbc3e05a62975fa641d0260f by Wenyi Zhao <reyizero@gmail.com>:
minor refine
--
846005cc76d0033112e47825c2e9a97790b6925f by Wenyi Zhao <reyizero@gmail.com>:
minor fix
--
f6a4becaa287d5ca323b2d152a4d0ae053730fd9 by Wenyi Zhao <reyizero@gmail.com>:
fix
--
5555749f60f7fce8f57962860ef65efccf0362ba by Wenyi Zhao <reyizero@gmail.com>:
fix
--
8873b9b6d9315c1199ca9f7c133ecf377ecd2fa6 by Wenyi Zhao <reyizero@gmail.com>:
fix
PiperOrigin-RevId: 376942547
2021-06-01 16:27:56 -07:00
5baf6e7709
Integrate LLVM at llvm/llvm-project@97d234935f
...
Updates LLVM usage to match
[97d234935f15](https://github.com/llvm/llvm-project/commit/97d234935f15 )
PiperOrigin-RevId: 376840022
2021-06-01 08:38:32 -07:00
d1828625ab
[MLIR][KernelGen] Make maximum supported rank in rank specialization configurable
...
The maximum supported target rank of 5 is sufficient for all operations but
`select`. Make the maximum target rank configurable in the rank specialization.
This reduces the number of generated kernels for operations that don't require
it.
PiperOrigin-RevId: 376822496
2021-06-01 06:54:31 -07:00
c7c245eaf1
[MLIR][KernelGen] Add MLIR-generated Xlogy kernel
...
Add the first MLIR-generated kernel that relies on an in-TF lowering. Fusion for
this kernel relies on the generalized rank specialization for operation groups.
PiperOrigin-RevId: 376805435
2021-06-01 04:48:18 -07:00
6a570502b6
Integrate LLVM at llvm/llvm-project@f000c4cfb6
...
Updates LLVM usage to match
[f000c4cfb66c](https://github.com/llvm/llvm-project/commit/f000c4cfb66c )
PiperOrigin-RevId: 376798033
2021-06-01 03:51:22 -07:00
f16e5a3a67
[MLIR][HLO] Use canonicalization patterns in broadcast propagation pass
...
Replace local canonicalization patterns with those from upstream.
PiperOrigin-RevId: 376794178
2021-06-01 03:14:31 -07:00
adff130b12
Integrate LLVM at llvm/llvm-project@41d7909368
...
Updates LLVM usage to match
[41d7909368be](https://github.com/llvm/llvm-project/commit/41d7909368be )
PiperOrigin-RevId: 376785495
2021-06-01 02:10:18 -07:00
31536431e0
[MLIR][HLO] Eliminate duplicate broadcastable constraints
...
PiperOrigin-RevId: 376718433
2021-05-31 13:50:23 -07:00
0f341012c6
[MLIR][HLO] Eliminate duplicate broadcastable constraints
...
PiperOrigin-RevId: 376715240
2021-05-31 13:08:02 -07:00
511a1db4f3
[MLIR][HLO] Use canonicalization patterns in broadcast propagation pass
...
Replace local canonicalization patterns with those from upstream.
PiperOrigin-RevId: 376708719
2021-05-31 12:01:26 -07:00
5f5db13715
[MLIR][HLO] Use canonicalization patterns in broadcast propagation pass
...
Replace local canonicalization patterns with those from upstream.
PiperOrigin-RevId: 376707588
2021-05-31 11:44:50 -07:00
1f786eb934
Integrate LLVM at llvm/llvm-project@c36ff6424f
...
Updates LLVM usage to match
[c36ff6424f24](https://github.com/llvm/llvm-project/commit/c36ff6424f24 )
PiperOrigin-RevId: 376687137
2021-05-31 08:16:39 -07:00
cc1b22e618
[HLO][Linalg] Support scalar broadcasts in point-wise converter
...
This is needed for operations that support this limited form of broadcasting,
namely `mhlo.select`.
PiperOrigin-RevId: 376655844
2021-05-31 03:50:23 -07:00
832f39b871
Integrate LLVM at llvm/llvm-project@577fea4e1a
...
Updates LLVM usage to match
[577fea4e1a13](https://github.com/llvm/llvm-project/commit/577fea4e1a13 )
PiperOrigin-RevId: 376640971
2021-05-31 01:54:46 -07:00
8110683c0b
Integrate LLVM at llvm/llvm-project@9968896cd6
...
Updates LLVM usage to match
[9968896cd62a](https://github.com/llvm/llvm-project/commit/9968896cd62a )
PiperOrigin-RevId: 376382711
2021-05-28 12:09:55 -07:00
e422ce5a9a
Integrate LLVM at llvm/llvm-project@8086f9d87e
...
Updates LLVM usage to match
[8086f9d87ee8](https://github.com/llvm/llvm-project/commit/8086f9d87ee8 )
PiperOrigin-RevId: 376289371
2021-05-27 18:20:22 -07:00
402b74ed7f
Fix type bug in mhlo.dynamic-update-slice lowering.
...
The operand type can be f32. We should not use operand type to do clamp
operations.
PiperOrigin-RevId: 376286524
2021-05-27 17:53:49 -07:00
b5f444232f
Integrate LLVM at llvm/llvm-project@d47dd11071
...
Updates LLVM usage to match
[d47dd1107132](https://github.com/llvm/llvm-project/commit/d47dd1107132 )
PiperOrigin-RevId: 376269354
2021-05-27 16:06:54 -07:00
6ea8f1ddf8
Integrate LLVM at llvm/llvm-project@eba69b59d1
...
Updates LLVM usage to match
[eba69b59d1a3](https://github.com/llvm/llvm-project/commit/eba69b59d1a3 )
PiperOrigin-RevId: 376175310
2021-05-27 08:37:41 -07:00
e319f87857
Integrate LLVM at llvm/llvm-project@c467585682
...
Updates LLVM usage to match
[c467585682dc](https://github.com/llvm/llvm-project/commit/c467585682dc )
PiperOrigin-RevId: 376155243
2021-05-27 06:25:17 -07:00
5deed55dc3
Integrate LLVM at llvm/llvm-project@d82f2a123f
...
Updates LLVM usage to match
[d82f2a123f9c](https://github.com/llvm/llvm-project/commit/d82f2a123f9c )
PiperOrigin-RevId: 376140722
2021-05-27 04:25:36 -07:00
b32f885ad7
[MLIR][KernelGen] Enable cluster rank specialization
...
Replace the previously used `TransformUnrankedHloPass` which rank-specializes
only one operation at a time. The new generalized rank specialization clusters
compatible operations and rank-specializes them collectively.
PiperOrigin-RevId: 376127752
2021-05-27 02:44:31 -07:00
a4fa6afa07
[mlir][hlo] Avoid dyn_cast_or_null when called with getDefiningOp result (NFC)
...
PiperOrigin-RevId: 376110457
2021-05-27 00:20:42 -07:00
d939a156d8
Integrate LLVM at llvm/llvm-project@5852582532
...
Updates LLVM usage to match
[5852582532b3](https://github.com/llvm/llvm-project/commit/5852582532b3 )
PiperOrigin-RevId: 376059531
2021-05-26 17:01:04 -07:00
28c411606f
Add support for lowering mhlo.dynamic-update-slice ops to Linalg and std ops.
...
PiperOrigin-RevId: 376042810
2021-05-26 15:31:05 -07:00
1c7415ba0b
Integrate LLVM at llvm/llvm-project@c5c1ec7945
...
Updates LLVM usage to match
[c5c1ec7945ff](https://github.com/llvm/llvm-project/commit/c5c1ec7945ff )
PiperOrigin-RevId: 376014620
2021-05-26 13:21:07 -07:00
af01c08ce6
Allow tuple as results of mhlo.custom_call op (NFC)
...
PiperOrigin-RevId: 376009518
2021-05-26 12:56:10 -07:00
26a0053d7d
Remove linalg.indexed_generic from mhlo lowerings to linalg
...
IndexedGeneric is going away. Transition to using linalg.Index instead.
PiperOrigin-RevId: 376002501
2021-05-26 12:24:23 -07:00
ca09dabf1a
Integrate LLVM at llvm/llvm-project@ea91a8cbab
...
Updates LLVM usage to match
[ea91a8cbab93](https://github.com/llvm/llvm-project/commit/ea91a8cbab93 )
PiperOrigin-RevId: 375987155
2021-05-26 11:15:30 -07:00
4ebcebf31c
[MLIR][HLO] Exploit scalar properties in rank specialization lowering
...
Take advantage of the fact that scalars are already ranked and that they are
neutral elements to broadcasting. Do not reshape scalars, do not consider them
for broadcasting, and materialize ranked operations on scalars accordingly.
PiperOrigin-RevId: 375968371
2021-05-26 09:59:13 -07:00
2f8f3d692c
Integrate LLVM at llvm/llvm-project@01120fe5b3
...
Updates LLVM usage to match
[01120fe5b398](https://github.com/llvm/llvm-project/commit/01120fe5b398 )
PiperOrigin-RevId: 375937201
2021-05-26 07:23:01 -07:00
edf5ec8084
Integrate LLVM at llvm/llvm-project@cb65419b1a
...
Updates LLVM usage to match
[cb65419b1ac0](https://github.com/llvm/llvm-project/commit/cb65419b1ac0 )
PiperOrigin-RevId: 375915516
2021-05-26 04:47:24 -07:00
4b3e9373e6
Integrate LLVM at llvm/llvm-project@bc56620b8b
...
Updates LLVM usage to match
[bc56620b8b76](https://github.com/llvm/llvm-project/commit/bc56620b8b76 )
PiperOrigin-RevId: 375829158
2021-05-25 17:17:28 -07:00
1439ec7cae
Integrate LLVM at llvm/llvm-project@0bbb502daa
...
Updates LLVM usage to match
[0bbb502daa90](https://github.com/llvm/llvm-project/commit/0bbb502daa90 )
PiperOrigin-RevId: 375778977
2021-05-25 13:13:08 -07:00
cb46298a07
[MLIR][HLO] Support all smaller ranks in rank specialization cases
...
Rank specialization cases can be applied to all argument tensors of smaller
ranks than the expected maximum rank. This is crucial if all operands are
effectively scalars and the maximum reduced rank is 0.
PiperOrigin-RevId: 375712020
2021-05-25 08:38:53 -07:00
c5af02fd8d
Integrate LLVM at llvm/llvm-project@f3c577ed38
...
Updates LLVM usage to match
[f3c577ed38e5](https://github.com/llvm/llvm-project/commit/f3c577ed38e5 )
PiperOrigin-RevId: 375703642
2021-05-25 07:53:35 -07:00
a847109ac7
Support complex types when converting HLO multiply op.
...
We can lower it to the MulOp in the complex dialect.
PiperOrigin-RevId: 375675079
2021-05-25 04:35:34 -07:00
5816920258
Support complex types when converting HLO divide op.
...
We can lower it to the DivOp in the complex dialect.
Also add tests to hlo-legalize-to-linalg.mlir for CompareOp lowering of complex
types. These were forgotten in a previous commit.
PiperOrigin-RevId: 375669125
2021-05-25 03:43:46 -07:00
8e28008e38
Integrate LLVM at llvm/llvm-project@6359842bc0
...
Updates LLVM usage to match
[6359842bc088](https://github.com/llvm/llvm-project/commit/6359842bc088 )
PiperOrigin-RevId: 375667591
2021-05-25 03:30:27 -07:00
758ae7da6b
Support complex types when converting HLO compare op (EQ/NE).
...
We can lower it to the EqualOp / NotEqualOp in the complex dialect.
PiperOrigin-RevId: 375655092
2021-05-25 01:54:27 -07:00
5504f82f11
Integrate LLVM at llvm/llvm-project@c2f819af73
...
Updates LLVM usage to match
[c2f819af73c5](https://github.com/llvm/llvm-project/commit/c2f819af73c5 )
PiperOrigin-RevId: 375598055
2021-05-24 17:52:41 -07:00
af6092ef13
Integrate LLVM at llvm/llvm-project@d65c32fb41
...
Updates LLVM usage to match
[d65c32fb41b0](https://github.com/llvm/llvm-project/commit/d65c32fb41b0 )
PiperOrigin-RevId: 375537865
2021-05-24 12:57:30 -07:00
b93e54d8a4
PR #49454 : [MLIR][DISC] Upgrade to use the new `reifyReturnTypeShapes` interface.
...
Imported from GitHub PR https://github.com/tensorflow/tensorflow/pull/49454
The new interface is more safe to be used during dialect conversion
(e.g. converting from tensor world to buffer world).
Copybara import of the project:
--
a6968072d59bec3c3bbaef0121d297e807c37c91 by Wenyi Zhao <reyizero@gmail.com>:
[MLIR][DISC] Upgrade to use the new `reifyReturnTypeShapes` interface.
The new interface is more safe to be used during dialect conversion
(e.g. converting from tensor world to buffer world).
--
55e7c6b7f2f99b99e226645a57e2433fae3e90ed by Wenyi Zhao <reyizero@gmail.com>:
minor fix
PiperOrigin-RevId: 375500273
2021-05-24 10:11:55 -07:00
28c4112f35
Downgrade some emitErrors in patterns to notifyMatchFailure.
...
PiperOrigin-RevId: 375160543
2021-05-21 14:09:23 -07:00
1ba4c714c9
Add support for lowering mhlo.scatter ops to Linalg.
...
This only works for updating tensors, not add/min/max computations. It requires
the index depth to be 1 because of the limitation in Linalg. We can not compare
multiple indices without packing indices.
PiperOrigin-RevId: 375137721
2021-05-21 12:17:14 -07:00
f5dc73ab60
Integrate LLVM at llvm/llvm-project@eb6429d0fb
...
Updates LLVM usage to match
[eb6429d0fb94](https://github.com/llvm/llvm-project/commit/eb6429d0fb94 )
PiperOrigin-RevId: 375071389
2021-05-21 06:31:13 -07:00
1f996ab796
Integrate LLVM at llvm/llvm-project@b9076d119a
...
Updates LLVM usage to match
[b9076d119a5b](https://github.com/llvm/llvm-project/commit/b9076d119a5b )
PiperOrigin-RevId: 375055490
2021-05-21 04:09:26 -07:00
97e6103933
[MLIR][HLO] Reshape to scalars in rank specialization
...
Scalars were incorrectly casted to scalar tensors when they have to be reshaped.
PiperOrigin-RevId: 375049088
2021-05-21 03:12:16 -07:00
c3147d76d5
Integrate LLVM at llvm/llvm-project@e3cf7c88c4
...
Updates LLVM usage to match
[e3cf7c88c472](https://github.com/llvm/llvm-project/commit/e3cf7c88c472 )
PiperOrigin-RevId: 375041162
2021-05-21 02:10:15 -07:00
A. Unique TensorFlower
wyzhao
Hanhan Wang
Benjamin Kramer
Adrian Kuegel
Mehdi Amini
Robert Suderman
Stella Laurenzo