Prior to this patch, generating test checks in place put the ATTR
definitions at the very top of the file, above the RUN lines and
autogenerated note. All CHECK lines should below the RUN lines and
autogenerated note.
This change ensures that the attribute definitions are emitted with the
rest of the CHECK lines.
---------
Co-authored-by: Michael Maitland <michaelmaitland@meta.com>
Prior to this PR, the script removed the already existing autogenerated
note if we came across a line that was equal to the note. But the
default note is multiple lines, so there would never be a match.
Instead, check to see if the current line is a substring of the
autogenerated note.
Co-authored-by: Michael Maitland <michaelmaitland@meta.com>
As part of the work on transitioning bufferization dialect, ops, and
associated logic to operate on newly added type interfaces (see
00eaff3e9c), rename the
bufferization.to_memref to highlight the generic nature of the op.
Bufferization process produces buffers while memref is a builtin type
rather than a generic term.
Preserve the current API (to_buffer still produces a memref), however,
as the new type interfaces are not used yet.
At the moment, the `CHECK-SAME` lines generated by
"generate-test-checks.py" (i.e. check-lines that correspond to the
preceeding `CHECK-LABEL` line) are indented to match the label length.
For example,
```mlir
func.func @batch_reduce_matmul_bcast_k_to_fill_missing_dims_A(%arg0: memref<5xf32>, %arg1: memref<2x5x7xf32>, %arg2: memref<3x7xf32>) {
linalg.batch_reduce_matmul indexing_maps = (...)
}
```
will lead to the following:
```mlir
// CHECK-LABEL: func.func @batch_reduce_matmul_bcast_k_to_fill_missing_dims_A(
// CHECK-SAME: %[[VAL_0:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<5xf32>,
// CHECK-SAME: %[[VAL_1:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<2x5x7xf32>,
// CHECK-SAME: %[[VAL_2:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<3x7xf32>) {
// CHECK: linalg.batch_reduce_matmul indexing_maps = (...)
```
This indentation is unnecasarilly deep. With this change, for labales
that are longer than 20 chars, the indentation is trimmed to 4 spaces:
```mlir
// CHECK-LABEL: func.func @batch_reduce_matmul_bcast_k_to_fill_missing_dims_A(
// CHECK-SAME: %[[VAL_0:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<5xf32>,
// CHECK-SAME: %[[VAL_1:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<2x5x7xf32>,
// CHECK-SAME: %[[VAL_2:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<3x7xf32>) {
// CHECK: linalg.batch_reduce_matmul indexing_maps = (...)
```
Following #128083, all `CHECK-SAME` lines generated by
generate-test-checks.py use a strict regex:
```mlir
// CHECK-SAME: %[[VAL_0:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: memref<128x256x512xf32>,
```
However, in most cases this strict form is unnecessary and can obscure
readability. In such cases, the following would be sufficient:
```mlir
// CHECK-SAME: %[[VAL_0:.*]]: memref<128x256x512xf32>,
```
This patch adds a command-line flag to make the strict mode optional. To
enable strict regex matching, use:
```bash
generate-test-checks.py --strict_name_re=true file.mlir
```
A few additions:
- Lines with `{{`: These can show up if serializing non-MLIR info into
string attrs `my.attr = {{proto}, {...}}`. String escape the opening
`{{`, given that check lines are generated this has no effect on
`{{.*}}` etc in generated lines.
- File split line: Normally these are skipped because of their indent
level, but if using `--starts_from_scope=0` to generate checks for the
`module {...} {` line, and since MLIR opt tools emit file split lines by
default, some `CHECK: // -----` lines were emit.
- (edit removed this, fixed by
https://github.com/llvm/llvm-project/pull/134364) AttrAliases: I'm not
sure if I'm missing something for the attribute parser to work
correctly, but I was getting many `#[[?]]` for all dialect attrs. Only
use the attr aliasing if there's a match.
If an attribute is not defined earlier in the same file, but just
referenced from its dialect directly, then currently not the correct
check is being emited.
What would it emit for #toy.shape<[1, 2, 3]>:
Earlier:
// CHECK: #[['?']]<[1, 2, 3]>
Now:
// CHECK: #toy.shape<[1, 2, 3]>
This commit pulls apart the inherent attribute dependence of classes
like EnumAttrInfo and EnumAttrCase, factoring them out into simpler
EnumCase and EnumInfo variants. This allows specifying the cases of an
enum without needing to make the cases, or the EnumInfo itself, a
subclass of SignlessIntegerAttrBase.
The existing classes are retained as subclasses of the new ones, both
for backwards compatibility and to allow attribute-specific information.
In addition, the new BitEnum class changes its default printer/parser
behavior: cases when multiple keywords appear, like having both nuw and
nsw in overflow flags, will no longer be quoted by the operator<<, and
the FieldParser instance will now expect multiple keywords. All
instances of BitEnumAttr retain the old behavior.
When CHECK-SAME checks are split across multiple lines,
the '.*' regexp for the SSA variable name may cause problems, e.g.:
```
// CHECK_LABEL: func.func @whatever(
// CHECK-SAME: %[[VAL_0:.*]]: i32,
// CHECK-SAME: %[[VAL_1:.*]]: i32,
// CHECK-SAME: %[[VAL_2:.*]]: i64)
```
This will not work for `func.func @whatever(%0: i32, %1: i32, %2: i64)`,
because VAL_0 will match to `0: i32, %1`.
This patch adds the erfc op to the math dialect. It also does lowering
of the math.erfc op to libm calls. There is also a f32 polynomial
approximation for the function based on
https://stackoverflow.com/questions/35966695/vectorizable-implementation-of-complementary-error-function-erfcf
This is in turn based on
M. M. Shepherd and J. G. Laframboise, "Chebyshev Approximation of
(1+2x)exp(x^2)erfc x in 0 <= x < INF", Mathematics of Computation, Vol.
36, No. 153, January 1981, pp. 249-253.
The code has a ULP error less than 3, which was tested, and MLIR test
values were verified against the C implementation.
Define keywords for the MLIR syntax. This allows better recognition of
semantic constructs such as SSA value identification e.g. `%foo` which
gives improved motion handling when using 'word based' such as `w, e`.
This is based on the work done for the LLVM IR in
8c46413f34.
Signed-off-by: Jack Frankland <jack.frankland@arm.com>
Use a word boundary, current code was currently failing when parsing the
definition of because it would also match
`CooperativeMatrixOp` from a later mention of
`SPIRV_KHR_CooperativeMatrixOperandsAttr`.
In the context of regular expressions, Python (used to) gracefully
ignore the escape behavior of `\` in some contexts, e.g. for
representing the regular expression `\w+`. However in newer versions of
Python this now gives a warning in the form
```
SyntaxWarning: invalid escape sequence '\w'
```
Fix by explicitly using raw strings instead.
This PR adds `f8E8M0FNU` type to MLIR.
`f8E8M0FNU` type is proposed in [OpenCompute MX
Specification](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf).
It defines a 8-bit floating point number with bit layout S0E8M0. Unlike
IEEE-754 types, there are no infinity, denormals, zeros or negative
values.
```c
f8E8M0FNU
- Exponent bias: 127
- Maximum stored exponent value: 254 (binary 1111'1110)
- Maximum unbiased exponent value: 254 - 127 = 127
- Minimum stored exponent value: 0 (binary 0000'0000)
- Minimum unbiased exponent value: 0 − 127 = -127
- Doesn't have zero
- Doesn't have infinity
- NaN is encoded as binary 1111'1111
Additional details:
- Zeros cannot be represented
- Negative values cannot be represented
- Mantissa is always 1
```
Related PRs:
- [PR-107127](https://github.com/llvm/llvm-project/pull/107127)
[APFloat] Add APFloat support for E8M0 type
- [PR-105573](https://github.com/llvm/llvm-project/pull/105573) [MLIR]
Add f6E3M2FN type - was used as a template for this PR
- [PR-107999](https://github.com/llvm/llvm-project/pull/107999) [MLIR]
Add f6E2M3FN type
- [PR-108877](https://github.com/llvm/llvm-project/pull/108877) [MLIR]
Add f4E2M1FN type
This PR adds `f4E2M1FN` type to mlir.
`f4E2M1FN` type is proposed in [OpenCompute MX
Specification](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf).
It defines a 4-bit floating point number with bit layout S1E2M1. Unlike
IEEE-754 types, there are no infinity or NaN values.
```c
f4E2M1FN
- Exponent bias: 1
- Maximum stored exponent value: 3 (binary 11)
- Maximum unbiased exponent value: 3 - 1 = 2
- Minimum stored exponent value: 1 (binary 01)
- Minimum unbiased exponent value: 1 − 1 = 0
- Has Positive and Negative zero
- Doesn't have infinity
- Doesn't have NaNs
Additional details:
- Zeros (+/-): S.00.0
- Max normal number: S.11.1 = ±2^(2) x (1 + 0.5) = ±6.0
- Min normal number: S.01.0 = ±2^(0) = ±1.0
- Min subnormal number: S.00.1 = ±2^(0) x 0.5 = ±0.5
```
Related PRs:
- [PR-95392](https://github.com/llvm/llvm-project/pull/95392) [APFloat]
Add APFloat support for FP4 data type
- [PR-105573](https://github.com/llvm/llvm-project/pull/105573) [MLIR]
Add f6E3M2FN type - was used as a template for this PR
- [PR-107999](https://github.com/llvm/llvm-project/pull/107999) [MLIR]
Add f6E2M3FN type
This PR adds `f6E2M3FN` type to mlir.
`f6E2M3FN` type is proposed in [OpenCompute MX
Specification](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf).
It defines a 6-bit floating point number with bit layout S1E2M3. Unlike
IEEE-754 types, there are no infinity or NaN values.
```c
f6E2M3FN
- Exponent bias: 1
- Maximum stored exponent value: 3 (binary 11)
- Maximum unbiased exponent value: 3 - 1 = 2
- Minimum stored exponent value: 1 (binary 01)
- Minimum unbiased exponent value: 1 − 1 = 0
- Has Positive and Negative zero
- Doesn't have infinity
- Doesn't have NaNs
Additional details:
- Zeros (+/-): S.00.000
- Max normal number: S.11.111 = ±2^(2) x (1 + 0.875) = ±7.5
- Min normal number: S.01.000 = ±2^(0) = ±1.0
- Max subnormal number: S.00.111 = ±2^(0) x 0.875 = ±0.875
- Min subnormal number: S.00.001 = ±2^(0) x 0.125 = ±0.125
```
Related PRs:
- [PR-94735](https://github.com/llvm/llvm-project/pull/94735) [APFloat]
Add APFloat support for FP6 data types
- [PR-105573](https://github.com/llvm/llvm-project/pull/105573) [MLIR]
Add f6E3M2FN type - was used as a template for this PR
This PR adds `f6E3M2FN` type to mlir.
`f6E3M2FN` type is proposed in [OpenCompute MX
Specification](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf).
It defines a 6-bit floating point number with bit layout S1E3M2. Unlike
IEEE-754 types, there are no infinity or NaN values.
```c
f6E3M2FN
- Exponent bias: 3
- Maximum stored exponent value: 7 (binary 111)
- Maximum unbiased exponent value: 7 - 3 = 4
- Minimum stored exponent value: 1 (binary 001)
- Minimum unbiased exponent value: 1 − 3 = −2
- Has Positive and Negative zero
- Doesn't have infinity
- Doesn't have NaNs
Additional details:
- Zeros (+/-): S.000.00
- Max normal number: S.111.11 = ±2^(4) x (1 + 0.75) = ±28
- Min normal number: S.001.00 = ±2^(-2) = ±0.25
- Max subnormal number: S.000.11 = ±2^(-2) x 0.75 = ±0.1875
- Min subnormal number: S.000.01 = ±2^(-2) x 0.25 = ±0.0625
```
Related PRs:
- [PR-94735](https://github.com/llvm/llvm-project/pull/94735) [APFloat]
Add APFloat support for FP6 data types
- [PR-97118](https://github.com/llvm/llvm-project/pull/97118) [MLIR] Add
f8E4M3 type - was used as a template for this PR
Each LSP server type (mlir-lsp-server, pdll-lsp-server and
tblgen-lsp-server) should have a different "additional_server_args"
entry in the config for passing arguments to the server such as
`--log=verbose`.
This PR adds `f8E3M4` type to mlir.
`f8E3M4` type follows IEEE 754 convention
```c
f8E3M4 (IEEE 754)
- Exponent bias: 3
- Maximum stored exponent value: 6 (binary 110)
- Maximum unbiased exponent value: 6 - 3 = 3
- Minimum stored exponent value: 1 (binary 001)
- Minimum unbiased exponent value: 1 − 3 = −2
- Precision specifies the total number of bits used for the significand (mantissa),
including implicit leading integer bit = 4 + 1 = 5
- Follows IEEE 754 conventions for representation of special values
- Has Positive and Negative zero
- Has Positive and Negative infinity
- Has NaNs
Additional details:
- Max exp (unbiased): 3
- Min exp (unbiased): -2
- Infinities (+/-): S.111.0000
- Zeros (+/-): S.000.0000
- NaNs: S.111.{0,1}⁴ except S.111.0000
- Max normal number: S.110.1111 = +/-2^(6-3) x (1 + 15/16) = +/-2^3 x 31 x 2^(-4) = +/-15.5
- Min normal number: S.001.0000 = +/-2^(1-3) x (1 + 0) = +/-2^(-2)
- Max subnormal number: S.000.1111 = +/-2^(-2) x 15/16 = +/-2^(-2) x 15 x 2^(-4) = +/-15 x 2^(-6)
- Min subnormal number: S.000.0001 = +/-2^(-2) x 1/16 = +/-2^(-2) x 2^(-4) = +/-2^(-6)
```
Related PRs:
- [PR-99698](https://github.com/llvm/llvm-project/pull/99698) [APFloat]
Add support for f8E3M4 IEEE 754 type
- [PR-97118](https://github.com/llvm/llvm-project/pull/97118) [MLIR] Add
f8E4M3 IEEE 754 type
This script looks for existing definitions with the `SPIRV_` prefix, so
that it can preserve them when updating the file. When the commit
2d62833048 changed the prefix from `SPV_`,
the number of characters to strip from matched names was not updated,
which broke this feature. This commit fixes remaining cases that weren't
fixed by 339c87a8a0.
The relationship of this script to the files it is meant to maintain is
still bitrotten in other ways.
This PR adds `f8E4M3` type to mlir.
`f8E4M3` type follows IEEE 754 convention
```c
f8E4M3 (IEEE 754)
- Exponent bias: 7
- Maximum stored exponent value: 14 (binary 1110)
- Maximum unbiased exponent value: 14 - 7 = 7
- Minimum stored exponent value: 1 (binary 0001)
- Minimum unbiased exponent value: 1 − 7 = −6
- Precision specifies the total number of bits used for the significand (mantisa),
including implicit leading integer bit = 3 + 1 = 4
- Follows IEEE 754 conventions for representation of special values
- Has Positive and Negative zero
- Has Positive and Negative infinity
- Has NaNs
Additional details:
- Max exp (unbiased): 7
- Min exp (unbiased): -6
- Infinities (+/-): S.1111.000
- Zeros (+/-): S.0000.000
- NaNs: S.1111.{001, 010, 011, 100, 101, 110, 111}
- Max normal number: S.1110.111 = +/-2^(7) x (1 + 0.875) = +/-240
- Min normal number: S.0001.000 = +/-2^(-6)
- Max subnormal number: S.0000.111 = +/-2^(-6) x 0.875 = +/-2^(-9) x 7
- Min subnormal number: S.0000.001 = +/-2^(-6) x 0.125 = +/-2^(-9)
```
Related PRs:
- [PR-97179](https://github.com/llvm/llvm-project/pull/97179) [APFloat]
Add support for f8E4M3 IEEE 754 type
This patch is part of a larger initiative aimed at fixing floating-point `max` and `min` operations in MLIR: https://discourse.llvm.org/t/rfc-fix-floating-point-max-and-min-operations-in-mlir/72671.
This commit addresses Task 1.2 of the mentioned RFC. By renaming these operations, we align their names with LLVM intrinsics that have corresponding semantics.
This regressed in D154458 due to the added tracking of used variable
names that now also has to be cleared alongside the counter.
Reviewed By: rafaelubalmw, c-rhodes, awarzynski
Differential Revision: https://reviews.llvm.org/D156547
- Option `--variable_names <names>` allows the user to pass names for FileCheck
regexps representing variables. Variable names are separated by commas, and
empty names can be used to generate specific variable names automatically.
For example, `--variable-names arg_0,arg_1,,,result` will produce regexp names
`ARG_0`, `ARG_1`, `VAR_0`, `VAR_1`, `RESULT`, `VAR_2`, `VAR_3`, ...
- Option '--attribute_names <names>' can be used to generate global regexp names
to represent attributes. Useful for affine maps. Same behavior as
'--variable_names'.
- Bug fixed for scope detection of SSA variables in ops with nested regions that
return SSA values (e.g., 'linalg.generic'). Originally, returned SSA values were
inserted in the nested scope.
This version of the tool has been used to generate unit tests for the following
patch: https://reviews.llvm.org/D153291
For example, the main body of the test named 'test_select_2d_one_dynamic' was
generated using the following command:
```
$ mlir-opt -pass-pipeline='builtin.module(func.func(tosa-to-linalg))' test_select_2d_one_dynamic.tosa.mlir | generate-test-checks.py --attribute_names map0,map1,map2 --variable_names arg0,arg1,arg2,const1,arg0_dim1,arg1_dim1,,arg2_dim1,max_dim1,,,arg0_broadcast,,,,,,,arg1_broadcast,,,,,,,arg2_broadcast,,,,,,result
```
Reviewed By: eric-k256
Differential Revision: https://reviews.llvm.org/D154458
Contribute a grammar, along with associated tests, from the upstream
project maintained at https://github.com/artagnon/tree-sitter-mlir. The
new grammar includes several fixes, and successfully parses 60-80% of
MLIR tests in the Arith, Math, ControlFlow, SCF, Tensor, Affine, and
Linalg dialects.
Differential Revision: https://reviews.llvm.org/D144408
This is an ongoing series of commits that are reformatting our
Python code.
Reformatting is done with `black`.
If you end up having problems merging this commit because you
have made changes to a python file, the best way to handle that
is to run git checkout --ours <yourfile> and then reformat it
with black.
If you run into any problems, post to discourse about it and
we will try to help.
RFC Thread below:
https://discourse.llvm.org/t/rfc-document-and-standardize-python-code-style
Differential Revision: https://reviews.llvm.org/D150782
This includes a small runtime acting as callback for the ExecutionEngine
and a C API that makes it possible to control from the debugger.
A python script for LLDB is included that hook a new `mlir` subcommand
and allows to set breakpoints and inspect the current action, the context
and the stack.
Differential Revision: https://reviews.llvm.org/D144817
X. Sun et al. (https://dl.acm.org/doi/10.5555/3454287.3454728) published
a paper showing that an FP format with 4 bits of exponent, 3 bits of
significand and an exponent bias of 11 would work quite well for ML
applications.
Google hardware supports a variant of this format where 0x80 is used to
represent NaN, as in the Float8E4M3FNUZ format. Just like the
Float8E4M3FNUZ format, this format does not support -0 and values which
would map to it will become +0.
This format is proposed for inclusion in OpenXLA's StableHLO dialect: https://github.com/openxla/stablehlo/pull/1308
As part of inclusion in that dialect, APFloat needs to know how to
handle this format.
Differential Revision: https://reviews.llvm.org/D146441
Float8E5M2FNUZ and Float8E4M3FNUZ have been added to APFloat in D141863.
This change adds these types as MLIR builtin types alongside Float8E5M2
and Float8E4M3FN (added in D133823 and D138075).
Reviewed By: krzysz00
Differential Revision: https://reviews.llvm.org/D143744
CMake supports CMAKE_CXX_COMPILER_LAUNCHER since CMake 3.4
so this custom CMake logic we had in LLVM can now be removed.
The only downside with this is that we can't set ccache
options from LLVM CMake, but it's arguable that this doesn't
belong in LLVM but should be done in a script calling the
build.
This was discussed in the forums here:
https://discourse.llvm.org/t/tips-for-incremental-building/67289/4?u=tobiashieta
Reviewed By: phosek
Differential Revision: https://reviews.llvm.org/D143468
This commit adds a significant amount of visualizers attempting
to cover the majority of our visualization needs. It covers:
* Operations/OperationName/Ops/OpInterfaces
* Attributes/Types/Attr|TypeInterfaces/NamedAttribute
* Blocks/Regions
* Various range types (e.g. ValueRange/TypeRange)
* Values/BlockArguments/OpResults
This does require an NFC change to interfaces to rename
the concept field to avoid clash with the base class. It
also requires exposing a few method to the debugger
to help resolve information that is non-trivial to reconstruct.
These methods are re-exported using a debug_Blah naming
scheme to avoid messing with hot methods.
Note that this makes use of the new callback feature in lldb-16
(currently trunk) that allows for providing visualizers based on
a dynamic callback, instead of just the typename. It requires
a very new lldb, but allows for providing good default visualization
for all attributes/operations/types out of the box.
Differential Revision: https://reviews.llvm.org/D139602
This covers `SDot`, `SUDot`, and `UDot`. The `*AccSat` version will be
added in a follow-up revision.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D139242
