Commit Graph

126 Commits (d4ea39b616ff38162a340292e53928b645a6494b)

Author SHA1 Message Date
Prashant Kumar d4ea39b616 Convert bool to float or integer type.
Conversion of torch.bool tensor type to float and integer type is
handled.
2022-02-07 21:22:22 +05:30
Anup Gangwar f9f97ea184 * [tosa] Support for AtenNativeLayerNormOp
* [tosa] Support for AtenPermuteOp

Signed-off-by: Anup Gangwar <anup.gangwar@arm.com>
2022-02-04 14:46:31 -05:00
Prashant Kumar ccf546f14c Add aten::nll_loss_backward op
The lowering of aten::nll_loss_backward op has been added
from torch to linalg dialect. The changes has been made as
a part of -torch-convert-to-linalg pass.

Signed-off-by: Prashant Kumar prashant@nod-labs.com
2022-02-04 21:57:53 +05:30
Prashant Kumar 68acc8696e Modify softmax decomposition to be more numerically stable.
The softmax decomposition is modified according to https://github.com/pytorch/functorch/blob/main/functorch/_src/decompositions.pytorch
to account for numerical stability. Also, modified aten.argmax lowering
to handle negative dimension.
2022-02-03 21:20:36 +05:30
Gaurav Shukla 0079901039 [TORCH][MLIR] Add E2E support for `aten.reshape` op
This commit decomposes `aten.reshape` into `aten.view` op in the case of
value tensor type operand.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-02-02 20:41:47 +05:30
Suraj Sudhir 1b505cbac5
RefineTypes fixes for TOSA backend (#557)
Handles Linear, Adaptive_AvgPool2D and FlattenUsintInts
Adds ResNet18 static model for TOSA

Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2022-02-01 14:08:54 -08:00
Yi Zhang 0cb216a1ad [Torch][Linalg] Add basic support for RNG
This PR include the following pieces:
- Add torch `Generator` type. `Generator` type is converted to i64 in
refbackend type converter.
- Add seed managment support for the default global generator.
`torch_c.getNextSeed` op is used to get the seed. On refbackend, the
`torch_c.getNextSeed` is lowered to load/store from [0] of global
variable `default_generator` memref<i64> in `InsertRngGlobals` pass.
- Add `aten.uniform_` and testing as an example op for RNG ops. Add
`torch.pseudo.aten.uniform` op. It has the same operands and return as
the `aten.uniform_` from the op registry except for value semantics.
2022-01-31 18:56:42 -05:00
Suraj Sudhir 0f083e770a
[tosa] Add maxpool2d and adaptive_avgpool2d support (#550)
Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2022-01-31 13:34:09 -08:00
Yi Zhang 5d9a15263a [TORCH] Add aten.std e2e support 2022-01-31 15:17:49 -05:00
Prashant Kumar e58b66bc3b Add lowering of `aten.max.dim` op.
Lowering of `aten.max.dim` op has been added.
2022-01-31 21:41:22 +05:30
Anup Gangwar 454fa9d123
* [tosa] Support for AtenFlattenUsingIntsOp (#548) 2022-01-28 21:38:56 -08:00
Liam Fitzpatrick 8bc028af05 Fold __is__ and unchecked_cast of derefine
The added e2e maxpool testcase from #545 was not getting a static shape
due to an unfolded prim.If when RefineTypes was called. This was because
of unfolded torch.iaten.__is__ and torch.prim.unchecked_cast operators
with torch.derefine operands.
2022-01-28 17:54:40 -05:00
Anup Gangwar 7a5736facd
* [tosa] Support for AtenReshapeOp (#543)
* [tosa] Support for AtenBatchNormOp

Signed-off-by: Anup Gangwar <anup.gangwar@arm.com>

Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-27 14:38:59 -08:00
Gaurav Shukla 13b9fd62c6 [TBE] Add a test module for table batch embedding
This commit adds a test module specifically for table batch embedding
algorithm. This test case is in reference to the FBGEMM table batch
embedding:
https://github.com/pytorch/FBGEMM/blob/main/fbgemm_gpu/bench/split_table_batched_embeddings_benchmark.py#L270

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-01-28 02:24:28 +05:30
Suraj Sudhir eb06d21765
[tosa] Implement conv2d support (#541)
Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2022-01-26 19:16:13 -08:00
Nirvedh 17a4843cf7 Adding an e2e test for histogram binning calibration 2022-01-25 18:27:20 -05:00
Anup Gangwar f8080bd1c5
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531)
* [tosa] Support for AtenCeilOp and AtenReciprocalOp
* [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant
* [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants

Signed-off-by: Anup Gangwar <anup.gangwar@arm.com>

Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-20 10:58:30 -08:00
Suraj Sudhir 5d6c4f48dc
[tosa] Enable tosa-to-linalg-named so Matmul works again (#530)
Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2022-01-19 12:10:04 -08:00
Vivek Khandelwal 6fe70c7794 [MLIR][TORCH] Add E2E support for aten.index.Tensor op
This commit adds lowering of `aten.index.Tensor` op

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-01-19 13:37:56 +05:30
Suraj Sudhir 5ded7d096f
[tosa] Add tosa-to-standard before tosa-to-linalg pass (#524)
Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2022-01-14 11:05:11 -08:00
Anup Gangwar abd61b4974 * Workaround for Issue 521, remove createTosaToStandard from Passes.cpp and
disable ElementwisePowModule_basic
* Update nll_loss_forward to align to the change in PyTorch

Signed-off-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-12 14:30:58 -06:00
Anup Gangwar d69d29b7a6 * [tosa] Support for AtenPowTensorScalarOp with constant Scalar as input
Signed-off-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-11 22:55:54 -05:00
Liam Fitzpatrick 077e55d756 Add support for constant_pad_nd
Note that to enable folding of the code coming from an example
like the ConstantPad2dStaticModule e2e test, support for other
operations had to be added/improved:
- aten::neg.int
- aten::eq.float
- aten::eq.str
- prim::Uninitialized
2022-01-11 10:25:25 -05:00
Vivek Khandelwal 35cf8d18f7 Add support for two return values
This commit adds support for two return values of type
memref f32 and i64.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-01-11 11:07:10 +05:30
Vivek Khandelwal ca662dc9cc [MLIR][TORCH] Add E2E support for aten.threshold, aten.threshold_backward op
This commit adds lowering of `aten.threshold` op
This commit adds lowering of `aten.threshold_backward` op

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-01-10 11:56:56 +05:30
Yi Zhang 7cf7b91664 [MLIR][TORCH] Fix tensor literal int elem type to be signless
The element type of tensor literal should be signless when converted to
builtin tensor types.
2022-01-07 16:34:24 -05:00
Yi Zhang 732a76f45c Make broadcasting result shape more static
This involes the following 2 parts:
- Change refine type to propagate more static shape info.
- Get as much static shape info as possible when creating the result
tensor when converting to linalg.
2022-01-06 18:39:27 -05:00
Suraj Sudhir b4842d9863
[tosa] Implement squeeze.dim support (#511)
Templated variants for squeeze and squeeze.dim
2022-01-06 08:31:29 -08:00
Gaurav Shukla 3c40539b34 [TORCH][MLIR] Add E2E support for `aten.[ones_like|zeros_like]`
- This commit adds E2E support for `aten.ones_like` and
  `aten.zeros_like` ops.
- Adds support for non-None `dtype` argument of `aten.empty_like` op.
- All the unit test cases related to constant tensor allocation like ops
  are moved to a different file named `constant_alloc.py`.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-01-06 20:24:40 +05:30
Liam Fitzpatrick ccfdfd1b80 Refine static shapes for conv2d and maxpool2d 2022-01-03 11:09:23 -06:00
Vivek Khandelwal 4486de5ef3 [MLIR][TORCH] Add E2E support for torch.arange op
This commit adds lowering of `aten.arange.start_step` op.
This commit decomposes `aten.arange` and `aten.arange.start` into
`aten.arange.start_step` op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2021-12-27 22:45:48 +05:30
Gaurav Shukla a83004c806 [TORCH][MLIR] Fold trivial cases of `aten.to.dtype` and `aten.view` op
- It folds `aten.to.dtype` when the input tensor type and result type
  are exactly same.
- It folds `aten.view` when the rank of both the input tensor type and
  result type is unity.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-24 13:32:34 +05:30
Prashant Kumar 9e1ecf2c0b Add Add and Sub scalar op conversions.
`aten.add.Scalar` and `aten.sub.Scalar` op conversions have been added.
The changes have been made as a part of `-convert-torch-to-linalg` pass.
2021-12-22 21:41:49 +05:30
Nirvedh 3cb46cecef Added aten::t() Op 2021-12-22 10:57:10 -05:00
xndcn 5eed562e19 add aten.sub.int/aten.mul.int lowering in TorchToStd 2021-12-17 10:35:15 -08:00
Yi Zhang d8ba68119e Lower aten::view with linalg.collapse and linalg.expand
We only handle the expanding OR collapsing cases, we do not handle
expanding And collapsing happening at the same time or cases where
it's neither collapsing nor expanding like view of [2,3] for
3x2 tensor.

It's assumed that if a shape list element is got from
`aten.size(tensor, dim)` the corresponding dim is not splitted or
collapsed. This assumption makes it easier to deal with dynamic shapes.
2021-12-16 17:58:20 -05:00
Gaurav Shukla bc9abbc1c9 [TORCH][MLIR] Add E2E support for `aten.empty_like` op
This commit adds decomposition of `aten.empty_like` into `aten.empty`
op.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-16 20:17:39 +05:30
Gaurav Shukla eddc09aa55 [TORCH][MLIR] Add E2E support for `aten.eq` and `aten.lt` ops
- Added E2E support for `aten.eq.Tensor` and `aten.lt.Tensor` ops. Both
  the operands are expected to be of the same type, i.e., type promotion
  is not addressed as a part of this commit.
- Added E2E support for `aten.eq.Scalar` and `aten.lt.Scalar` ops.
  Tensor operand type to Scalar operand type promotion has not been
  handled in this commit.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-16 18:47:22 +05:30
Suraj Sudhir 0cd95b5c68
[tosa] Support for Torch.squeeze (#487) 2021-12-15 21:40:29 -08:00
Daniel Garvey 396ab35c9d
Small fixes for slice edge cases (#476) 2021-12-15 15:54:41 -06:00
Prashant Kumar ab81f871e4 Add aten.tensor.int and aten.tensor.float op lowerings.
Add the required lowerings and correct test cases.
These op produce zero-d tensors and it was incorrectly mentioned in
refine types to produce 1d tensor of size 1.
2021-12-15 17:21:34 +05:30
Anup Gangwar cce490d71d
* [tosa] Support for Rsqrt legalization (#480)
Signed-off-by: Anup Gangwar <anup.gangwar@arm.com>

Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2021-12-14 10:03:58 -08:00
Prashant Kumar 6dabf185f5 Add support for int types in gtScalar op.
Support for integer types in gtScalar op has been added.
The code share same logic with gtTensor op and can be merged
which is added as a TODO.
2021-12-14 01:29:52 +05:30
Gaurav Shukla 8d4879feb0 [TORCH][MLIR] Add and templatize lowering of [`aten.zeros|aten.ones|aten.empty`] ops
- Templatize `aten.zeros` and `aten.ones` ops lowering.
- Add E2E support for `aten.empty` op.
- Add Integer type support in `aten.mul.Scalar` op lowering.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-14 00:07:11 +05:30
Prashant Kumar 528354de84 Add `aten.gt.Tensor` op
`aten.gt.Tensor` op has been added in torch dialect and the
lowering of the op has been done to the linalg dialect.

Signed-off-by: Prashant Kumar <prashant@nod-labs.com>
2021-12-13 00:08:52 +05:30
Gaurav Shukla a778f990e9 [TORCH][MLIR] Add E2E support for `aten.ceil` op
This commit adds lowering of `aten.ceil` op as a part of element-wise
ops lowering.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-12 01:15:47 +05:30
Prateek Gupta cfc8de36f8
[MLIR][TORCH] Add E2E support for `aten.native_layer_norm`. (#470)
This commit adds support for aten.native_layer_norm operation. Here
the previous code for aten.layer_norm is tweaked a little bit to
accomodate both mean and variance values alongwith the layer norm
value. This commit also adds decomposition of aten.layer_norm into
aten.native_layer_norm, which was previously getting lowered directly
to linalg.

Signed-Off-By: Prateek Gupta<prateek@nod-labs.com>
2021-12-10 19:06:19 +05:30
Gaurav Shukla 5a47f92390 [TORCH][MLIR] Add E2E support for `aten.squeeze.dim` op
This commit adds lowering of `aten.squeeze.dim` op into
`linalg.TensorCollapseShape` op. Here, the dim(th) dimension of the
input tensor is not supposed to be dynamic.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-10 17:01:20 +05:30
Vivek Khandelwal 8130354c09 [MLIR][TORCH] Add E2E support for aten.index_select op
This commit adds lowering of `aten.index_select` op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2021-12-09 23:13:36 +05:30
Gaurav Shukla f34eb66124 [TORCH][MLIR] Add E2E support for [`aten.gt.Scalar`|`aten.where.self`]
This commit adds lowering of `aten.gt.Scalar` and `aten.where.self` as a
part of element-wise ops lowering.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2021-12-09 12:47:10 +05:30