Allows linalg structured operations to be handled during spmdization and
sharding propagation.
There is only support for projected permutation indexing maps.
Note: This is a reland of #84035.
The standard specifies that it it UB to specialize the following traits:
- `std::is_integral`
- `std::is_unsigned`
- `std::make_unsigned`
- `std::make_signed`
This patch:
- Removes specializations for `BigInt`
- Transforms SFINAE for `bit.h` functions from template parameter to
return type (This makes specialization easier).
- Adds `BigInt` specialization for `bit.h` functions.
- Fixes code depending on previous specializations.
This adds patterns and a pass to convert the Arith dialect to EmitC. For
now, this covers arithemtic binary ops operating on floating point
types.
It is not checked within the patterns whether the types, such as the
Tensor type, are supported in the respective EmitC operations. If
unsupported types should be converted, the conversion will fail anyway
because no legal EmitC operation can be created. This can clearly be
improved in a follow up, also resulting in better error messages.
Functions for such checks should not solely be used in the conversions
and should also be (re)used in the verifier.
The standard specifies that it it UB to specialize the following traits:
- `std::is_integral`
- `std::is_unsigned`
- `std::make_unsigned`
- `std::make_signed`
This patch:
- Removes specializations for `BigInt`
- Transforms SFINAE for `bit.h` functions from template parameter to
return type (This makes specialization easier).
- Adds `BigInt` specialization for `bit.h` functions.
- Fixes code depending on previous specializations.
was available. However, it turns out that the NVPTX compiler is not part
of every CUDA distribution, so `if_cuda_available` may evaluate to true
without that compiler being present, which breaks the build. This PR
thus sets the macro to 0 always (which was the behavior before #84007).
This is another follow-up of #83004, which made the same change for
`MLIR_CUDA_CONVERSIONS_ENABLED`. As the previous PR, this PR commit
exposes mentioned CMake variable through `mlir-config.h` and uses the
macro that is introduced with the same name. This replaces the macro
`MLIR_NVPTXCOMPILER_ENABLED`, which the CMake files previously defined
manually.
This is another follow-up of #83004, which made the same change for
`MLIR_CUDA_CONVERSIONS_ENABLED`. As the previous PR, this PR commit
exposes mentioned CMake variable through `mlir-config.h` and uses the
macro that is introduced with the same name. This replaces the macro
`MLIR_ENABLE_DEPRECATED_GPU_SERIALIZATION`, which the CMake files
previously defined manually.
This is another follow-up of #83004. `NVVM/Target.cpp` uses the macro
`MLIR_NVPTXCOMPILER_ENABLED`, which is defined in `llvm-config.h` but
did not include that file, yielding a warning when compiled with
`-Wundef`. This PR adds the include.
~~This is another follow-up of #83004, which made the same change for
`MLIR_CUDA_CONVERSIONS_ENABLED`. As the previous PR, this PR commit
exposes mentioned CMake variable through `mlir-config.h` and uses the
macro that is introduced with the same name. This replaces the macro
`MLIR_NVPTXCOMPILER_ENABLED`, which the CMake files previously defined
manually.~~
This is a follow up of #83004, which made the same change for
`MLIR_CUDA_CONVERSIONS_ENABLED`. As the previous PR, this PR commit
exposes mentioned CMake variable through `mlir-config.h` and uses the
macro that is introduced with the same name. This replaces the macro
`MLIR_ROCM_CONVERSIONS_ENABLED`, which the CMake files previously
defined manually.
The [bazel docs](https://bazel.build/rules/lib/globals/bzl#rule)
discourage setting this. The comments about being necessary for headers
or genrules seem to be obsolete, at least for the LLVM tree itself.
The effect of this is that generated files will go to `bazel-bin`
instead of `bazel-genfiles`.
One external use was fixed here:
32bb3b0613.
Summary:
I've noticed one problem is that the user includes `stdint.h` the
compiler will do `#include_next <stdint.h>` potentially into a
conflicting implementation on systems with multiple headers installed.
The `clang` header is standards compliant and works with `clang` and
`gcc` which are both of our targets, so I simply copied it here. This
has the effect of including `stdint.h` on clang / LLVM libc behaving the
same as `-ffreestanding`.
That macro was not defined in some cases and thus yielded warnings if
compiled with `-Wundef`. In particular, they were not defined in the
BUILD files, so the GPU targets were broken when built with Bazel. This
commit exposes mentioned CMake variable through mlir-config.h and uses
the macro that is introduced with the same name. This replaces the macro
MLIR_CUDA_CONVERSIONS_ENABLED, which the CMake files previously defined
manually.
This patch adds the r, R, k, and K conversion specifiers to printf, with
accompanying tests. They are guarded behind the
LIBC_COPT_PRINTF_DISABLE_FIXED_POINT flag as well as automatic fixed
point support detection.
This op is the inverse of all-gather. It is useful to have an explicit
concise representation instead of having a blob of slicing logic.
Add lowering for the op that slices from the tensor based on the
in-group process index.
Make resharding generate an all-slice instead of inserting the slicing
logic directly.
