These are identified by misc-include-cleaner. I've filtered out those
that break builds. Also, I'm staying away from llvm-config.h,
config.h, and Compiler.h, which likely cause platform- or
compiler-specific build failures.
The motivation is to avoid having to negate `isDynamic*` checks, avoid
double negations, and allow for `ShapedType::isStaticDim` to be used in
ADT functions without having to wrap it in a lambda performing the
negation.
Also add the new functions to C and Python bindings.
These are identified by misc-include-cleaner. I've filtered out those
that break builds. Also, I'm staying away from llvm-config.h,
config.h, and Compiler.h, which likely cause platform- or
compiler-specific build failures.
ArrayRef has a constructor that accepts std::nullopt. This
constructor dates back to the days when we still had llvm::Optional.
Since the use of std::nullopt outside the context of std::optional is
kind of abuse and not intuitive to new comers, I would like to move
away from the constructor and eventually remove it.
This patch migrates away from std::nullopt in favor of ArrayRef<T>()
where we use perfect forwarding. Note that {} would be ambiguous for
perfect forwarding to work.
Along with the changes to rescale op attributes, commit 7208649 dropped
the builtin casts between signed and signless types. However, explicitly
unsigned types are still legal input and output values from the TOSA IR
perspective.
The change adds back the casts when the unsigned<->signless semantics
are explicit in the underlying tensor types. This prevents the
conversion routine from trying to generate illegal `arith` casts that
are constrained to signless types. Whether the `arith` casts themselves
are signed or unsigned should still depend on the rescale's `*_unsigned`
attribute values.
---------
Signed-off-by: Artem Gindinson <gindinson@roofline.ai>
Lowering of tosa.rescale to Linalg unconditionally sign-extend the input
zero-point value, even when unsigned_input is true. This commit refactor
zeropoint handling to share the same logic between input and output
zeropoint.
Lowering of tosa.rescale to Linalg unconditionally sign-extend the input
zero-point value, even when unsigned_input is true. This commit refactor
zeropoint handling to share the same logic between input and output
zeropoint.
In the propagate mode, NaN compare equal to each other so in case of
several NaNs the index of the first one needs to be returned. This
commit changes the index update condition to check that the current
index is not that of a NaN.
The commit also simplifies argmax NaN ignore lowering to only use OGT.
This prevent any update in case of NaN. The only case where the index of
a NaN is returned is when all values are NaN and this is covered by the
fact that the initial index value is 0 so no update will result in 0
being returned.
*Update RescaleOp to use zero-point as operands instead of attributes.
*Check input_zp data type against the input and output_zp data type
against the output.
Signed-off-by: Peng Sun <peng.sun@arm.com>
Co-authored-by: Peng Sun <peng.sun@arm.com>
This commit changes the zero point attribute to an input to align with
the 1.0 spec.
Signed-off-by: Tai Ly <tai.ly@arm.com>
Co-authored-by: Luke Hutton <luke.hutton@arm.com>
Updated the dialect to match TOSA v1.0 specification for ConstOp and
ConstShapeOp (https://www.mlplatform.org/tosa/tosa_spec.html#_const).
Also updated lit tests
---------
Signed-off-by: Jerry Ge <jerry.ge@arm.com>
For non floating point operations NaN propagation mode has no meaning
and can be safely ignored. For non integer types skip the compare and
select materialization for NaN propagation even in "IGNORE" mode. This
fixes a bug where an unchecked `cast<FloatType>()` was called in the
"IGNORE" case even when the operation is acting on integers.
Update the lit tests for the NaN propagation lowering to check that the
propagation logic is not materialized in the case of a non floating
point type e.g. i8.
Signed-off-by: Jack Frankland <jack.frankland@arm.com>
Add support for NaN propagation lowering in the `tosa-to-linalg` and
`tosa-to-linalg-named` conversions by conditionally checking for NaN in
the case of ignore semantics and materializing the appropriate select
operations. Note that the default behviour of "propagate" matches that
of the arith dialect and so in that case we can avoid creating the
checks altogether.
Add appropriate lit tests including negative tests which check the
various comparisons and selects are materialized as appropriate.
This affects the following TOSA operators:
* arg_max
* max_pool_2d
* clamp
* reduce_max
* reduce_min
* maximum
* minimum
Signed-off-by: Jack Frankland <jack.frankland@arm.com>
Move the `scale`, `offset`, and `border` parameters of the RESIZE
operator in the MLIR TOSA dialect from attributes to inputs and update
lit tests appropriately.
Add the verifier of the `tosa::ResizeOp` operation.
---------
Co-authored-by: Tai Ly <tai.ly@arm.com>
Co-authored-by: Luke Hutton <luke.hutton@arm.com>
Change the shift operand for the mul operator to be a required operand.
Also defined shift to be Tosa_ScalarInt8Tensor which requires that it is
a rank-1 tensor
whose shape is [1] (ie, tensor containing a single element)
Signed-off-by: Tai Ly <tai.ly@arm.com>
This changes Tosa ClampOp attributes to min_val and max_val which are
either integer attributes or float attributes, and adds verify checks
that these attribute element types must match element types of input and
output
Co-authored-by: Tai Ly <tai.ly@arm.com>
getInput1Zp() returns an unsigned value which means in case of negative
zero point value the max intermediate value computation currently goes
wrong. Use getInput1ZpAttr() instead which returns an APInt and allows
easy sign extension to int64_t.
The shape operand is changed to input shape type since V1.0
Change-Id: I508cc1d67e9b017048b3f29fecf202cb7d707110
Co-authored-by: Won Jeon <won.jeon@arm.com>
Removed the TOSA quantization attribute used in various MLIR TOSA
dialect operations in favour of using builtin attributes.
Update any lit tests, conversions and transformations appropriately.
Signed-off-by: Tai Ly <tai.ly@arm.com>
Co-authored-by: Tai Ly <tai.ly@arm.com>
The TOSA-v1.0 specification makes the shift attribute of the MUL
(Hammard product) operator an input. Move the `shift` parameter of the
MUL operator in the MILR TOSA dialect from an attribute to an input and
update any lit tests appropriately.
Expand the verifier of the `tosa::MulOp` operation to check the various
constraints defined in the TOSA-v1.0 specification. Specifically, ensure
that all input operands (excluding the optional shift) are of the same
rank. This means that broadcasting tests which previously checked rank-0
tensors would be broadcast are no longer valid and are removed.
Signed-off-by: Jack Frankland <jack.frankland@arm.com>
Co-authored-by: TatWai Chong <tatwai.chong@arm.com>
In the LLVM style guide, we prefer not using braced initializer lists to
call a constructor. Also, we prefer using an equal before the open curly
brace if we use a braced initializer list when initializing a variable.
See
https://llvm.org/docs/CodingStandards.html#do-not-use-braced-initializer-lists-to-call-a-constructor
for more details.
The style guide does not explain the reason well. There is an article
from abseil, which mentions few benefits. E.g., we can avoid the most
vexing parse, etc. See https://abseil.io/tips/88 for more details.
Signed-off-by: hanhanW <hanhan0912@gmail.com>
When deciding whether to emit a map like
`#map = affine_map<(d0, d1, d2, d3) -> (0, d1, d2, d3)>` or `#map =
affine_map<(d0, d1, d2, d3) -> (d0, d1, d2, d3)>` for an operand of a
`linalg.generic` when lowering element-wise TOSA ops, prefer the latter
unless broadcasting of the operand is really needed.
This helps later transformations which often require the affine map to
be a projected permuatation.
Adds:
1. tosa shape type to Tosa dialect e.g., !tosa.shape<4> is a type for
rank-4 shape values (size-4 array of index values)
2. const_shape operator
3. trait TosaShapeOperator, added to tosa shape operators, and a
verifier that all operands and results of operator are tosa shapes
4. trait TosaResolvableShapeOperands, added to all tosa operators, and a
verifier that every tosa shape operand is produced by a tosa shape
operator (indicated by trait TosaShapeOperator)
5. trait TosaShapeOperatorWithSameRanks, added to
Tosa_ElementwiseShapeOp and a verifier that all operands and result
shapes have same ranks
5. changed TileOp's multiples from attribute to input, of !tosa.shape
type.
6. add folder for tosa ConstShape operator
This patch was originally authored by Tai Ly <tai.ly@arm.com>
Signed-off-by: Jerry Ge <Jerry.Ge@arm.com>
Signed-off-by: Tai Ly <tai.ly@arm.com>
Unsigned integer types are uncommon enough in MLIR that there is no
operation to cast a scalar from signless to unsigned and vice versa.
Currently tosa.rescale uses builtin.unrealized_conversion_cast which
does not lower. Instead, this commit introduces optional attributes to
indicate unsigned input or output, named similarly to those in the TOSA
specification. This is more in line with the rest of MLIR where specific
operations rather than values are signed/unsigned.
This PR fixes an issue related to integer overflow when computing
`(intmax+1)` for `i64` during `tosa-to-linalg` pass for `tosa.cast`.
Found this issue while debugging a numerical mismatch for `deeplabv3`
model from `torchvision` represented in `tosa` dialect using the
`TorchToTosa` pipeline in `torch-mlir` repository. `torch.aten.to.dtype`
is converted to `tosa.cast` that casts `f32` to `i64` type. Technically
by the specification, `tosa.cast` doesn't handle casting `f32` to `i64`.
So it's possible to add a verifier to error out for such tosa ops
instead of producing incorrect code. However, I chose to fix the
overflow issue to still be able to represent the `deeplabv3` model with
`tosa` ops in the above-mentioned pipeline. Open to suggestions if
adding the verifier is more appropriate instead.
This commit marks the type converter in `populate...` functions as
`const`. This is useful for debugging.
Patterns already take a `const` type converter. However, some
`populate...` functions do not only add new patterns, but also add
additional type conversion rules. That makes it difficult to find the
place where a type conversion was added in the code base. With this
change, all `populate...` functions that only populate pattern now have
a `const` type converter. Programmers can then conclude from the
function signature that these functions do not register any new type
conversion rules.
Also some minor cleanups around the 1:N dialect conversion
infrastructure, which did not always pass the type converter as a
`const` object internally.
This PR makes tosa.negate op for integer types to use the simplified
calculation branch if input_zp and output_zp values are also zero.
Signed-off-by: Dmitriy Smirnov <dmitriy.smirnov@arm.com>
This implements the lowering of tosa.clamp with unsigned operand to
linalg.
We interpret the `min/max : i64` attributes on `clamp` to be signed.
This means that when the operand has type `ui64`, one cannot represent
limits across the whole range.
This patch generalizes tensor.expand_shape and memref.expand_shape to
consume the output shape as a list of SSA values. This enables us to
implement generic reshape operations with dynamic shapes using
collapse_shape/expand_shape pairs.
The output_shape input to expand_shape follows the static/dynamic
representation that's also used in `tensor.extract_slice`.
Differential Revision: https://reviews.llvm.org/D140821
---------
Signed-off-by: Gaurav Shukla<gaurav.shukla@amd.com>
Signed-off-by: Gaurav Shukla <gaurav.shukla@amd.com>
Co-authored-by: Ramiro Leal-Cavazos <ramiroleal050@gmail.com>
