Commit Graph

19 Commits

Author SHA1 Message Date
Adrian Kuegel 4033a56750 Add special cases for SelectOp rank specialization.
We now use the same special cases for all ops with arity >= 2.
For binary ops, we now have only one special case if at least one of the
operands has exactly one element. In that case, we reshape both operands to
rank 1. Before, we had separate special cases whether the left-hand side
or the right-hand side have a scalar shape.

PiperOrigin-RevId: 366005835
2021-03-31 04:28:51 -07:00
A. Unique TensorFlower 39650a5d5a Remove rank 1 specialization from TransformUnrankedHloPass.
For binary ops, we already special-case rank 0 vs rank 1, and same shape. So we
don't need to special-case a maximum rank of 1.

PiperOrigin-RevId: 360891955
2021-03-04 05:24:53 -08:00
Adrian Kuegel 62b357b601 Remove rank 1 specialization from TransformUnrankedHloPass.
For binary ops, we already special-case rank 0 vs rank 1, and same shape. So we
don't need to special-case a maximum rank of 1.

PiperOrigin-RevId: 360881387
2021-03-04 04:04:11 -08:00
Adrian Kuegel 0683db3b24 Legalize MinimumBroadcastShapes op.
Use it in TransformUnrankedHloPass, which allows to reduce the maximum
rank for rank specialized broadcast from 6 to 5.

PiperOrigin-RevId: 360415743
2021-03-02 06:39:01 -08:00
A. Unique TensorFlower ac0552f127 [MLIR][HLO] Remove duplicate `PopulateTransformUnrankedHloPatterns`
PiperOrigin-RevId: 359046173
2021-02-23 07:50:47 -08:00
Benjamin Kramer a9cc1dcfa0 [mlir][hlo] Add basic rank-specialization for select
This just blows up everything to ranked (up to 6) and is probably quite slow.
This is sufficient to make kernelgen compile SelectV2.

PiperOrigin-RevId: 358777728
2021-02-22 02:41:12 -08:00
Benjamin Kramer b42def4612 [mlir][hlo] Refactor rank specialization to allow an arbitrary number of inputs
This actually simplifies the code a bit.

PiperOrigin-RevId: 358201038
2021-02-18 09:53:03 -08:00
Adrian Kuegel 824bc9c425 Improve broadcast transformation to treat dynamic shapes with 1 element as scalar.
A shape that contains exactly one element is effectively a scalar. This leads
to a speedup in cases where we have a binary op with one operand that is
effectively a scalar, because we can use the fast path.

PiperOrigin-RevId: 357515552
2021-02-14 23:25:41 -08:00
A. Unique TensorFlower d77c9ad6fa [MLIR][CHLO] Add `is_inf`, `is_pos_inf`, and `is_neg_inf` to CHLO dialect
Also add the respective lowerings to MHLO.

PiperOrigin-RevId: 354101955
2021-01-27 09:00:56 -08:00
Tres Popp ba0346b071 Integrate LLVM at llvm/llvm-project@96ef4f307d
Updates LLVM usage to match
[96ef4f307df2](https://github.com/llvm/llvm-project/commit/96ef4f307df2)

PiperOrigin-RevId: 352786460
2021-01-20 07:09:47 -08:00
A. Unique TensorFlower 9a1abaa212 Integrate LLVM at llvm/llvm-project@e2d7d3cb0e
Updates LLVM usage to match
[e2d7d3cb0ead](https://github.com/llvm/llvm-project/commit/e2d7d3cb0ead)

PiperOrigin-RevId: 351915841
2021-01-14 18:05:33 -08:00
A. Unique TensorFlower b0bf2ef45b Integrate LLVM at llvm/llvm-project@c3acda0798
Updates LLVM usage to match
[c3acda0798f9](https://github.com/llvm/llvm-project/commit/c3acda0798f9)

PiperOrigin-RevId: 348896724
2020-12-23 23:53:54 -08:00
Tres Popp d327fc5737 [kernel_gen] Lower max rank specialization from 6 to 5
We don't care much about rank 6 broadcasting operations and this lowers compile times significantly.

PiperOrigin-RevId: 346046601
2020-12-07 02:18:38 -08:00
Tres Popp 7c3f049c8e [kernel_gen] Lower max rank specialization from 6 to 5
We don't care much about rank 6 broadcasting operations and this lowers compile times significantly.

PiperOrigin-RevId: 345466476
2020-12-03 09:19:25 -08:00
Adrian Kuegel 6a71a84302 Support different input/output type for TransformUnrankedHlo.
Also generate the tf.Equal kernel, now that it works.

PiperOrigin-RevId: 344402014
2020-11-26 04:20:34 -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
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
A. Unique TensorFlower 69b80d8deb [MLIR] Extend unranked transformation to CHLO dialect
PiperOrigin-RevId: 332026604
2020-09-16 09:49:18 -07:00