llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
torch-mlir/test/Dialect/Torch/ops.mlir

193 lines
8.5 KiB
MLIR
Raw Normal View History

[torch-mlir earthmoving (1/N)] C/C++ code movement. This creates the `external/torch-mlir` directory as an LLVM_EXTERNAL_PROJECTS-compatible project (analogous to `iree-dialects`) and completes movement/rename of all pure MLIR C/C++ compiler code into there. The next step will be to move all the Python code / code that links/includes PyTorch C++ code (which currently lives in `frontends/pytorch`) into a subdirectory here. I call this "earthmoving" because it is mostly mechanical changes and renames. As a quick summary (we can change this down the road easily) - C++ `mlir::NPCOMP::Torch -> mlir::torch::Torch` - CAPI `npcompTorchListTypeGet -> torchMlirTorchListTypeGet` - preprocessor `#ifndef NPCOMP_ -> #ifndef TORCHMLIR_` - CMake `NPCOMPFoo -> TorchMLIRFoo` The goal of this is to create a standalone project creating a center of mass for entry into the MLIR ecosystem from PyTorch, suitable in scope for eventual inclusion/ownership in PyTorch. The idea is that `external/torch-mlir` will some day be pulled out into its own repository, and then npcomp will simply pull it in as a submodule. Layering-wise, what lives in `torch-mlir` lowers code from PyTorch (currently TorchScript, but TorchFX or pytorch/xla-style tracing are possible extensions) down to what we have been calling the "Torch backend contract" which is cleaned up IR (inlining, simplifcation, conversion to value tensors, ...) entirely in the `torch` dialect. This is the branching off point for further lowering, of which npcomp takes one opinion (outside `torch-mlir` of course!), namely the `TorchConversion` dialect/transforms which lower to IR suitable for IREE and other linalg-on-tensors based lower-level compilers. Summary of changes: - move `{include,lib,test}/Dialect/Torch` into `torch-mlir` - move relevant parts of CAPI into `torch-mlir`. - leave a few things related to the `torch-mlir` Python build commented out, which should be resolved in a subsequent change.
2021-09-10 03:24:10 +08:00
// RUN: torch-mlir-opt %s | torch-mlir-opt | FileCheck %s
Add a torch.kernel_call op and associated predicates.
2020-09-30 05:17:34 +08:00
[torch-mlir][torch] add encoding field to torch type (#2799) This adds an encoding field to the torch type, using the interfaces for printing, parsing, and verification. Note that although this change prepares adding sparsity to the torch type (as illustrated by the round trip and invalid tests), nothing in this change depends on the actual contents of the encoding field!
2024-01-26 02:04:04 +08:00
// CHECK: #[[$ENCODING:.*]] = #sparse_tensor.encoding<{ map = (d0, d1) -> (d0 : dense, d1 : compressed) }>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @torch.operator(
func.func @torch.operator(%arg0: !torch.tensor, %arg1: !torch.tensor) -> !torch.tensor {
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: torch.operator "ns.unqual.overload"(%arg0, %arg1) : (!torch.tensor, !torch.tensor) -> !torch.tensor
%0 = torch.operator "ns.unqual.overload"(%arg0, %arg1) : (!torch.tensor, !torch.tensor) -> !torch.tensor
return %0 : !torch.tensor
Add a torch.kernel_call op and associated predicates.
2020-09-30 05:17:34 +08:00
}
Add initial TorchScript module importer It turns out that this was easiest to structure as a general IValue importer, since torch module are just one of the possible IValue's. We import the IValue object graph in a braindead fashion into basicpy ops and a new `torch.nn_module` op that is used to model the attributes/methods of a torch::jit::Module IValue. See `Torch/ops.mlir` for an example, and also check out the .py import tests in `frontends/pytorch/test/module_import`. As part of this change, a few housekeeping tasks: - extract some helpers from graph_importer.cpp - more helpers around the C API - misc touchups
2021-01-28 08:35:44 +08:00
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @torch.linear_params.create(%arg0: !torch.tensor, %arg1: !torch.tensor) -> (!torch.LinearParams, !torch.LinearParams) {
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
%with_bias = torch.linear_params.create %arg0, %arg1 : !torch.tensor, !torch.tensor
%without_bias = torch.linear_params.create %arg0 : !torch.tensor
Get simple quantized model importing. This is enough to import the program and get it through the compilation pipeline. It of course fails at the VerifyBackendContract pass since there is a lot missing, but the final IR for a simple quantized MLP is looking pretty decent already: [IR](https://gist.github.com/silvasean/f76bccd76e9b193d396cfb2f9a11f54d) Main changes: - Add support for importing torch quantized tensors, including `torch.per_tensor_affine.create` op and `!torch.qint8` element type. - Add support for importing `LinearPackedParamsBase` (basically a weight + optional bias, but requires `torch.linear_params.create` op + `!torch.LinearParams` type to model it). This was less painful than I expected, as it has the necessary methods to opaquely unpack itself. I factored things so it should be easy to extend to other custom classes like `ConvPackedParamsBase`. - Add minimal boilerplate for importing `quantized::*` ops, with `quantized::linear` being a motivating example. - Add e2e test with simple quantized MLP (courtesy of @phoenix-meadowlark). This is somewhat of an abuse of `!numpy.ndarray` / `tensor`, as really the proper semantics of `!torch.qint8` dtype on a Torch tensor is "check the quantizer object of the tensor for side data (scale/offset, possibly per-channel) that defines the full semantics of the tensor". We don't have any such notion of "side data" for `!numpy.ndarray` / `tensor`, let alone anything that would have the associated behavior of keying off the dtype to determine if the side data is present. This will be fixed by a proper `!torch.tensor` type.
2021-05-20 02:40:48 +08:00
return %with_bias, %without_bias : !torch.LinearParams, !torch.LinearParams
}
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.default() -> !torch.tensor
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.default() -> !torch.tensor
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.default_explicit() -> !torch.tensor{{$}}
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.default_explicit() -> !torch.tensor<*,unk>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.value_semantic() -> !torch.vtensor{{$}}
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.value_semantic() -> !torch.vtensor<*,unk>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.dtype() -> !torch.tensor<*,si32>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.dtype() -> !torch.tensor<*,si32>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.ranked() -> !torch.tensor<[?,?,?],unk>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.ranked() -> !torch.tensor<[?,?,?],unk>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.some_sizes_known() -> !torch.tensor<[?,2,?,4],unk>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.some_sizes_known() -> !torch.tensor<[?,2,?,4],unk>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
// CHECK: @tensor.fully_determined() -> !torch.vtensor<[1,2,3,4],f32>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tensor.fully_determined() -> !torch.vtensor<[1,2,3,4],f32>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
[torch-mlir][torch] add encoding field to torch type (#2799) This adds an encoding field to the torch type, using the interfaces for printing, parsing, and verification. Note that although this change prepares adding sparsity to the torch type (as illustrated by the round trip and invalid tests), nothing in this change depends on the actual contents of the encoding field!
2024-01-26 02:04:04 +08:00
// CHECK: @tensor.sparse() -> !torch.vtensor<[64,64],f32,#[[$ENCODING]]>
#CSR = #sparse_tensor.encoding<{ map = (d0, d1) -> (d0 : dense, d1 : compressed) }>
func.func private @tensor.sparse() -> !torch.vtensor<[64,64],f32,#CSR>
Add `!torch.tuple<T1, T2>` type. This further eliminates the need for the `basicpy` dependency. This required adding `torch.prim.TupleConstruct` to replace `basicpy.build_tuple`.
2021-06-15 09:06:38 +08:00
// CHECK: @tuple.empty() -> !torch.tuple<>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tuple.empty() -> !torch.tuple<>
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
// CHECK: @tuple.one_element() -> !torch.tuple<tensor>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tuple.one_element() -> !torch.tuple<tensor>
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
// CHECK: @tuple.two_elements() -> !torch.tuple<tensor, tensor>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @tuple.two_elements() -> !torch.tuple<tensor, tensor>
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
Add minimal support for Union types. A recent PyTorch commit made ConstantPad2d call a helper function with a `Union[int, float]` type annotated. This commit adds minimal support for representing and dealing with that. https://github.com/pytorch/pytorch/pull/73287 Changes: - Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`, along with the importer and CAPI code. - Add support in isValidSubtype for union types. - Adding a canonicalizer for `torch.derefine` to help simplify some code that derefines to a UnionType (this also fixes #664). There is still more work to do for really supporting UnionType well, such as canonicalizing UnionType's so that they can be compared with pointer equality.
2022-03-30 06:57:31 +08:00
// CHECK: @union.empty() -> !torch.union<>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @union.empty() -> !torch.union<>
Add minimal support for Union types. A recent PyTorch commit made ConstantPad2d call a helper function with a `Union[int, float]` type annotated. This commit adds minimal support for representing and dealing with that. https://github.com/pytorch/pytorch/pull/73287 Changes: - Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`, along with the importer and CAPI code. - Add support in isValidSubtype for union types. - Adding a canonicalizer for `torch.derefine` to help simplify some code that derefines to a UnionType (this also fixes #664). There is still more work to do for really supporting UnionType well, such as canonicalizing UnionType's so that they can be compared with pointer equality.
2022-03-30 06:57:31 +08:00
// CHECK: @union.one_element() -> !torch.union<tensor>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @union.one_element() -> !torch.union<tensor>
Add minimal support for Union types. A recent PyTorch commit made ConstantPad2d call a helper function with a `Union[int, float]` type annotated. This commit adds minimal support for representing and dealing with that. https://github.com/pytorch/pytorch/pull/73287 Changes: - Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`, along with the importer and CAPI code. - Add support in isValidSubtype for union types. - Adding a canonicalizer for `torch.derefine` to help simplify some code that derefines to a UnionType (this also fixes #664). There is still more work to do for really supporting UnionType well, such as canonicalizing UnionType's so that they can be compared with pointer equality.
2022-03-30 06:57:31 +08:00
// CHECK: @union.two_elements() -> !torch.union<tensor, tensor>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @union.two_elements() -> !torch.union<tensor, tensor>
Add minimal support for Union types. A recent PyTorch commit made ConstantPad2d call a helper function with a `Union[int, float]` type annotated. This commit adds minimal support for representing and dealing with that. https://github.com/pytorch/pytorch/pull/73287 Changes: - Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`, along with the importer and CAPI code. - Add support in isValidSubtype for union types. - Adding a canonicalizer for `torch.derefine` to help simplify some code that derefines to a UnionType (this also fixes #664). There is still more work to do for really supporting UnionType well, such as canonicalizing UnionType's so that they can be compared with pointer equality.
2022-03-30 06:57:31 +08:00
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
// CHECK: @dict() -> !torch.dict<str, tensor>
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @dict() -> !torch.dict<str, tensor>
Add `!torch.tuple<T1, T2>` type. This further eliminates the need for the `basicpy` dependency. This required adding `torch.prim.TupleConstruct` to replace `basicpy.build_tuple`.
2021-06-15 09:06:38 +08:00
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @torch.tensor.literal() {
func.func @torch.tensor.literal() {
Add `torch.vtensor.literal` op. This op is much better behaved than the `torch.tensor.literal` op (which is the new name of the `torch.tensor` op). In particular `torch.tensor.literal`: - always has a maximally refined type. - always has value semantics. - can be constant folded / CSE'd. ReduceOpVariants is changed to perform the transformation from `torch.tensor.literal` to `torch.vtensor.literal` (which in general involves static information casts and copies. This new op also allowed tightening up `torch.tensor.literal` to only accept NonValueTensorType (instead of any tensor type). This new ".literal" name is more descriptive. It was getting too confusing seeing an op called just `torch.tensor` (we originally called it that because that's the name of the similar function in the Torch Python API, but it just doesn't fit here).
2021-06-17 23:52:13 +08:00
// CHECK: torch.tensor.literal(dense<4.200000e+01> : tensor<3x2xf32>) : !torch.tensor
%0 = torch.tensor.literal(dense<42.0> : tensor<3x2xf32>) : !torch.tensor
// CHECK: torch.tensor.literal(dense<4.200000e+01> : tensor<3x2xf32>) : !torch.tensor<[3,2],f32>
%1 = torch.tensor.literal(dense<42.0> : tensor<3x2xf32>) : !torch.tensor<[3,2],f32>
return
}
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @torch.vtensor.literal() {
func.func @torch.vtensor.literal() {
Add `torch.vtensor.literal` op. This op is much better behaved than the `torch.tensor.literal` op (which is the new name of the `torch.tensor` op). In particular `torch.tensor.literal`: - always has a maximally refined type. - always has value semantics. - can be constant folded / CSE'd. ReduceOpVariants is changed to perform the transformation from `torch.tensor.literal` to `torch.vtensor.literal` (which in general involves static information casts and copies. This new op also allowed tightening up `torch.tensor.literal` to only accept NonValueTensorType (instead of any tensor type). This new ".literal" name is more descriptive. It was getting too confusing seeing an op called just `torch.tensor` (we originally called it that because that's the name of the similar function in the Torch Python API, but it just doesn't fit here).
2021-06-17 23:52:13 +08:00
// CHECK: torch.vtensor.literal(dense<4.200000e+01> : tensor<3x2xf32>) : !torch.vtensor<[3,2],f32>
%0 = torch.vtensor.literal(dense<42.0> : tensor<3x2xf32>) : !torch.vtensor<[3,2],f32>
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
return
}
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @derefine(%arg0: !torch.tensor) -> !torch.optional<tensor> {
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
%0 = torch.derefine %arg0 : !torch.tensor to !torch.optional<tensor>
return %0 : !torch.optional<tensor>
Add basic MLP's to the e2e curriculum. These tests pass on the reference backend. - Add aten.linear op + shape xfer function + ATen->Linalg lowering. - Note: this needs to be more automated, and needs to cover more cases. - Current not implemented caveats: - size-1 broadcasting for bias vector (either static-size-1 or ? case) - higher-rank aten.linear ops (not produced by torch.nn.Linear though) - type promotion (still don't even know the exact rules here) - Add folder for torch.derefine op. Now the inliner can clean it up as it inlines. (call boundaries are a main place we need to insert torch.derefine) This is brittle -- the other important case is control flow which will need to be handled via an extension to RefineTypes.cpp (as will more robust call handling). River has an in-flight patch to update it to the new dataflow framework so I didn't want to do anything intrusive here. - Also adjust torch.derefine syntax to use the keyword `to` instead of `->`, as most type-only, cast-like ops do.
2021-04-27 02:42:41 +08:00
}
Add initial TorchScript module importer It turns out that this was easiest to structure as a general IValue importer, since torch module are just one of the possible IValue's. We import the IValue object graph in a braindead fashion into basicpy ops and a new `torch.nn_module` op that is used to model the attributes/methods of a torch::jit::Module IValue. See `Torch/ops.mlir` for an example, and also check out the .py import tests in `frontends/pytorch/test/module_import`. As part of this change, a few housekeeping tasks: - extract some helpers from graph_importer.cpp - more helpers around the C API - misc touchups
2021-01-28 08:35:44 +08:00
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @torch.prim.If(%arg0: !torch.bool, %arg1: !torch.int) -> !torch.int {
Add `!torch.int` type. This replaces the ad-hoc use of `i64` throughout the Torch layer, and helps to keep it crystal clear the distinction between `!torch.int` (which is modeling the Python `int` type) and the various types that serve as dtypes of tensors, which are a totally different type universe. Changes: - `!torch.int` type and C bindings. - Change `torch.constant.int` parser to not need the `: i64` at the end. - `m_TorchConstantInt` matcher to aid with matching constants. - BackendTypeConversion changes for `!torch.int` -> `i64` type conversion. - Refactor finalizing patterns in FinalizingBackendTypeConversionPass (they were getting very repetitive). - Mechanical rewriting of `!torch.int` to `i64` in all the tests, and `AnyTorchIntType` to `Torch_IntType` in the `.td` files.
2021-06-17 06:53:15 +08:00
%0 = torch.prim.If %arg0 -> (!torch.int) {
%1 = torch.aten.add.int %arg1, %arg1 : !torch.int, !torch.int -> !torch.int
torch.prim.If.yield %1 : !torch.int
Add `torch.prim.If` This removes the use of `scf.if`, which required laundering back and forth between `i1` and `!torch.bool` in the frontend. We will eventually lower this op to `scf.if`, but this results in a cleaner IR and layering at the frontend.
2021-06-17 01:23:26 +08:00
} else {
Add `!torch.int` type. This replaces the ad-hoc use of `i64` throughout the Torch layer, and helps to keep it crystal clear the distinction between `!torch.int` (which is modeling the Python `int` type) and the various types that serve as dtypes of tensors, which are a totally different type universe. Changes: - `!torch.int` type and C bindings. - Change `torch.constant.int` parser to not need the `: i64` at the end. - `m_TorchConstantInt` matcher to aid with matching constants. - BackendTypeConversion changes for `!torch.int` -> `i64` type conversion. - Refactor finalizing patterns in FinalizingBackendTypeConversionPass (they were getting very repetitive). - Mechanical rewriting of `!torch.int` to `i64` in all the tests, and `AnyTorchIntType` to `Torch_IntType` in the `.td` files.
2021-06-17 06:53:15 +08:00
%1 = torch.aten.mul.int %arg1, %arg1 : !torch.int, !torch.int -> !torch.int
torch.prim.If.yield %1 : !torch.int
Add `torch.prim.If` This removes the use of `scf.if`, which required laundering back and forth between `i1` and `!torch.bool` in the frontend. We will eventually lower this op to `scf.if`, but this results in a cleaner IR and layering at the frontend.
2021-06-17 01:23:26 +08:00
}
Add `!torch.int` type. This replaces the ad-hoc use of `i64` throughout the Torch layer, and helps to keep it crystal clear the distinction between `!torch.int` (which is modeling the Python `int` type) and the various types that serve as dtypes of tensors, which are a totally different type universe. Changes: - `!torch.int` type and C bindings. - Change `torch.constant.int` parser to not need the `: i64` at the end. - `m_TorchConstantInt` matcher to aid with matching constants. - BackendTypeConversion changes for `!torch.int` -> `i64` type conversion. - Refactor finalizing patterns in FinalizingBackendTypeConversionPass (they were getting very repetitive). - Mechanical rewriting of `!torch.int` to `i64` in all the tests, and `AnyTorchIntType` to `Torch_IntType` in the `.td` files.
2021-06-17 06:53:15 +08:00
return %0 : !torch.int
Add `torch.prim.If` This removes the use of `scf.if`, which required laundering back and forth between `i1` and `!torch.bool` in the frontend. We will eventually lower this op to `scf.if`, but this results in a cleaner IR and layering at the frontend.
2021-06-17 01:23:26 +08:00
}
Add `!torch.bool` type. This finishes removing the dependence on the basicpy dialect! Changes: - Add `!torch.bool` type and replace use of `!basicpy.BoolType` in Torch-related code. - Rename BuiltinTensorize to BackendTypeConversion since now it handles bool conversions (and, when we add !torch.int and !torch.float, it will handle those as well), and generalize the related utilities (I also moved them to Torch/Transforms since they aren't really part of Torch/IR). - Add `torch.to_i1` and `torch.from_i1` ops for materializations - [cleanup] Reorganize `torch.constant.*` ops in TorchOps.td - Remove dependency of `torch` dialect on `basicpy` dialect and also `std` dialect. For `std`, we use some call related ops, but the `torch` dialect itself never produces them (we have passes that do though). This is fairly mechanical. Recommended review order: - New stuff in Torch/IR - New BuiltinTypeConversion files. - Mechnical fixups elsewhere.
2021-06-16 07:47:53 +08:00
// CHECK: %true = torch.constant.bool true
%true = torch.constant.bool true
// CHECK: %false = torch.constant.bool false
%false = torch.constant.bool false
Make torch.constant.float print a little nicer. This printing is chosen to be similar to how MLIR prints the values by default.
2021-06-23 05:25:16 +08:00
Add `torch.constant.int` and `torch.constant.float`. - This removes reliance on basicpy.numeric_constant. - Also, add OpAsmOpInterface to the `torch.constant.none` and `torch.constant.str` ops.
2021-06-16 03:42:51 +08:00
// CHECK: %int3 = torch.constant.int 3
%int3 = torch.constant.int 3
Make torch.constant.float print a little nicer. This printing is chosen to be similar to how MLIR prints the values by default.
2021-06-23 05:25:16 +08:00
// CHECK: %int-3 = torch.constant.int -3
%int-3 = torch.constant.int -3
[Torch] Fix torch.constant.int operation parsing (#3476) Due to the custom operation parser, the print and parser were expecting two different forms. One having the dictionary before the value and the other after. Following the format of the other constants ops, the constant.int will follow the `value attr-dict` format. Updated the parser accordingly.
2024-06-28 22:06:52 +08:00
// CHECK: %int5 = torch.constant.int 5 {test = "value"}
%int5 = torch.constant.int 5 {test = "value"}
Make torch.constant.float print a little nicer. This printing is chosen to be similar to how MLIR prints the values by default.
2021-06-23 05:25:16 +08:00
// CHECK: %float1.000000e00 = torch.constant.float 1.000000e+00
%float1.000000e00 = torch.constant.float 1.000000e+00
// CHECK: %float-1.000000e00 = torch.constant.float -1.000000e+00
%float-1.000000e00 = torch.constant.float -1.000000e+00
// CHECK: %float1.000000e-10 = torch.constant.float 1.000000e-10
%float1.000000e-10 = torch.constant.float 1.000000e-10
// CHECK: %float1.000000e10 = torch.constant.float 1.000000e+10
%float1.000000e10 = torch.constant.float 1.000000e+10
// CHECK: %float4.250000e01 = torch.constant.float 4.250000e+01
%float4.250000e01 = torch.constant.float 4.250000e+01
Add `torch.vtensor.literal` op. This op is much better behaved than the `torch.tensor.literal` op (which is the new name of the `torch.tensor` op). In particular `torch.tensor.literal`: - always has a maximally refined type. - always has value semantics. - can be constant folded / CSE'd. ReduceOpVariants is changed to perform the transformation from `torch.tensor.literal` to `torch.vtensor.literal` (which in general involves static information casts and copies. This new op also allowed tightening up `torch.tensor.literal` to only accept NonValueTensorType (instead of any tensor type). This new ".literal" name is more descriptive. It was getting too confusing seeing an op called just `torch.tensor` (we originally called it that because that's the name of the similar function in the Torch Python API, but it just doesn't fit here).
2021-06-17 23:52:13 +08:00
%tensor = torch.tensor.literal(dense<1.000000e+00> : tensor<1xf32>) : !torch.tensor
Add `torch.constant.int` and `torch.constant.float`. - This removes reliance on basicpy.numeric_constant. - Also, add OpAsmOpInterface to the `torch.constant.none` and `torch.constant.str` ops.
2021-06-16 03:42:51 +08:00
// CHECK: %none = torch.constant.none
Introduce native `!torch.none` type. - Add `torch.constant.none` op to construct it (naming is chosen to be analogous to Torch's representation of a prim::Constant with NoneType, rather than using the "singleton" terminology of Basicpy).
2021-06-15 02:36:10 +08:00
%none = torch.constant.none
Add `torch.constant.int` and `torch.constant.float`. - This removes reliance on basicpy.numeric_constant. - Also, add OpAsmOpInterface to the `torch.constant.none` and `torch.constant.str` ops.
2021-06-16 03:42:51 +08:00
// CHECK: %str = torch.constant.str "some str"
Add !torch.str type. - Remove dependence on `!basicpy.BytesType`. - Add `torch.constant.str "s"` analogous to `torch.constant.none`.
2021-06-15 23:29:06 +08:00
%str = torch.constant.str "some str"
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func private @f(%arg0: !torch.nn.Module<"test">) {
Add initial TorchScript module importer It turns out that this was easiest to structure as a general IValue importer, since torch module are just one of the possible IValue's. We import the IValue object graph in a braindead fashion into basicpy ops and a new `torch.nn_module` op that is used to model the attributes/methods of a torch::jit::Module IValue. See `Torch/ops.mlir` for an example, and also check out the .py import tests in `frontends/pytorch/test/module_import`. As part of this change, a few housekeeping tasks: - extract some helpers from graph_importer.cpp - more helpers around the C API - misc touchups
2021-01-28 08:35:44 +08:00
return
}
Implement GlobalizeObjectGraph transformation. This required restructuring of how we model TorchScript on import. The main difference is that now we split out a `torch.class_type` that holds methods and declarations of the types of each slot. This is more consistent with TorchScript (our previous representation was "denormalized"). Recommended reading order: 1. check out the description of `torch.class_type` in `TorchOps.td` and look at `test/Dialect/Torch/ops.mlir` and `frontends/pytorch/test/module_import/` to familiarize with the new representation. - Just look at the new IR. The diff between the old names and new names is confusing. 2. check out `test/Dialect/Torch/globalize-object-graph*.mlir` and read along with the pass description in `include/npcomp/Dialect/Torch/Transforms/Passes.td` 3. Read the code in `GlobalizeObjectGraph.cpp` and miscellaneous changes in `ivalue_importer.cpp`, `TorchOps.cpp`, etc.
2021-02-18 03:28:51 +08:00
torch.class_type @empty {}
%submodule = torch.nn_module {} : !torch.nn.Module<"empty">
torch.class_type @test {
Add `!torch.bool` type. This finishes removing the dependence on the basicpy dialect! Changes: - Add `!torch.bool` type and replace use of `!basicpy.BoolType` in Torch-related code. - Rename BuiltinTensorize to BackendTypeConversion since now it handles bool conversions (and, when we add !torch.int and !torch.float, it will handle those as well), and generalize the related utilities (I also moved them to Torch/Transforms since they aren't really part of Torch/IR). - Add `torch.to_i1` and `torch.from_i1` ops for materializations - [cleanup] Reorganize `torch.constant.*` ops in TorchOps.td - Remove dependency of `torch` dialect on `basicpy` dialect and also `std` dialect. For `std`, we use some call related ops, but the `torch` dialect itself never produces them (we have passes that do though). This is fairly mechanical. Recommended review order: - New stuff in Torch/IR - New BuiltinTypeConversion files. - Mechnical fixups elsewhere.
2021-06-16 07:47:53 +08:00
torch.attr "b" : !torch.bool
Add `!torch.int` type. This replaces the ad-hoc use of `i64` throughout the Torch layer, and helps to keep it crystal clear the distinction between `!torch.int` (which is modeling the Python `int` type) and the various types that serve as dtypes of tensors, which are a totally different type universe. Changes: - `!torch.int` type and C bindings. - Change `torch.constant.int` parser to not need the `: i64` at the end. - `m_TorchConstantInt` matcher to aid with matching constants. - BackendTypeConversion changes for `!torch.int` -> `i64` type conversion. - Refactor finalizing patterns in FinalizingBackendTypeConversionPass (they were getting very repetitive). - Mechanical rewriting of `!torch.int` to `i64` in all the tests, and `AnyTorchIntType` to `Torch_IntType` in the `.td` files.
2021-06-17 06:53:15 +08:00
torch.attr "i" : !torch.int
Add a !torch.float type. This removes the dependence of the `torch` dialect on the low-level builtin types. Now the `torch` dialect is a standalone layer, suitable for targeting from higher-level Python abstractions without any premature lowering to primitive types.
2021-06-17 23:24:31 +08:00
torch.attr "f" : !torch.float
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
torch.attr "t" : !torch.tensor
Implement GlobalizeObjectGraph transformation. This required restructuring of how we model TorchScript on import. The main difference is that now we split out a `torch.class_type` that holds methods and declarations of the types of each slot. This is more consistent with TorchScript (our previous representation was "denormalized"). Recommended reading order: 1. check out the description of `torch.class_type` in `TorchOps.td` and look at `test/Dialect/Torch/ops.mlir` and `frontends/pytorch/test/module_import/` to familiarize with the new representation. - Just look at the new IR. The diff between the old names and new names is confusing. 2. check out `test/Dialect/Torch/globalize-object-graph*.mlir` and read along with the pass description in `include/npcomp/Dialect/Torch/Transforms/Passes.td` 3. Read the code in `GlobalizeObjectGraph.cpp` and miscellaneous changes in `ivalue_importer.cpp`, `TorchOps.cpp`, etc.
2021-02-18 03:28:51 +08:00
torch.attr "submodule" : !torch.nn.Module<"empty">
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
torch.attr "ob" : !torch.optional<bool>
Add !torch.str type. - Remove dependence on `!basicpy.BytesType`. - Add `torch.constant.str "s"` analogous to `torch.constant.none`.
2021-06-15 23:29:06 +08:00
torch.attr "s" : !torch.str
Add initial TorchScript module importer It turns out that this was easiest to structure as a general IValue importer, since torch module are just one of the possible IValue's. We import the IValue object graph in a braindead fashion into basicpy ops and a new `torch.nn_module` op that is used to model the attributes/methods of a torch::jit::Module IValue. See `Torch/ops.mlir` for an example, and also check out the .py import tests in `frontends/pytorch/test/module_import`. As part of this change, a few housekeeping tasks: - extract some helpers from graph_importer.cpp - more helpers around the C API - misc touchups
2021-01-28 08:35:44 +08:00
torch.method "method", @f
}
Implement GlobalizeObjectGraph transformation. This required restructuring of how we model TorchScript on import. The main difference is that now we split out a `torch.class_type` that holds methods and declarations of the types of each slot. This is more consistent with TorchScript (our previous representation was "denormalized"). Recommended reading order: 1. check out the description of `torch.class_type` in `TorchOps.td` and look at `test/Dialect/Torch/ops.mlir` and `frontends/pytorch/test/module_import/` to familiarize with the new representation. - Just look at the new IR. The diff between the old names and new names is confusing. 2. check out `test/Dialect/Torch/globalize-object-graph*.mlir` and read along with the pass description in `include/npcomp/Dialect/Torch/Transforms/Passes.td` 3. Read the code in `GlobalizeObjectGraph.cpp` and miscellaneous changes in `ivalue_importer.cpp`, `TorchOps.cpp`, etc.
2021-02-18 03:28:51 +08:00
torch.nn_module {
Add `!torch.bool` type. This finishes removing the dependence on the basicpy dialect! Changes: - Add `!torch.bool` type and replace use of `!basicpy.BoolType` in Torch-related code. - Rename BuiltinTensorize to BackendTypeConversion since now it handles bool conversions (and, when we add !torch.int and !torch.float, it will handle those as well), and generalize the related utilities (I also moved them to Torch/Transforms since they aren't really part of Torch/IR). - Add `torch.to_i1` and `torch.from_i1` ops for materializations - [cleanup] Reorganize `torch.constant.*` ops in TorchOps.td - Remove dependency of `torch` dialect on `basicpy` dialect and also `std` dialect. For `std`, we use some call related ops, but the `torch` dialect itself never produces them (we have passes that do though). This is fairly mechanical. Recommended review order: - New stuff in Torch/IR - New BuiltinTypeConversion files. - Mechnical fixups elsewhere.
2021-06-16 07:47:53 +08:00
torch.slot "b", %true : !torch.bool
Add `!torch.int` type. This replaces the ad-hoc use of `i64` throughout the Torch layer, and helps to keep it crystal clear the distinction between `!torch.int` (which is modeling the Python `int` type) and the various types that serve as dtypes of tensors, which are a totally different type universe. Changes: - `!torch.int` type and C bindings. - Change `torch.constant.int` parser to not need the `: i64` at the end. - `m_TorchConstantInt` matcher to aid with matching constants. - BackendTypeConversion changes for `!torch.int` -> `i64` type conversion. - Refactor finalizing patterns in FinalizingBackendTypeConversionPass (they were getting very repetitive). - Mechanical rewriting of `!torch.int` to `i64` in all the tests, and `AnyTorchIntType` to `Torch_IntType` in the `.td` files.
2021-06-17 06:53:15 +08:00
torch.slot "i", %int3 : !torch.int
Make torch.constant.float print a little nicer. This printing is chosen to be similar to how MLIR prints the values by default.
2021-06-23 05:25:16 +08:00
torch.slot "f", %float1.000000e00 : !torch.float
Introduce `!torch.tensor` / `!torch.vtensor` types. This removes our reliance on the numpy dialect and avoids our off-label use of the builtin tnesor type for modeling unknown dtypes. The `!torch.vtensor` (`ValueTensorType`) type is a value-semantic tensor. The `!torch.tensor` (`NonValueTensorType`) type is a non-value-semantic tensor. The new types look as follows syntactically: ``` // Least-static-information, non-value-semantic tensor. !torch.tensor // Explicit form of least-static-information variant. !torch.tensor<*,unk> // Least-static-information, value-semantic tensor. !torch.vtensor // Explicit form of least-static-information variant. !torch.vtensor<*,unk> // Fixed-set of allowable element types, with first-class support for // Torch's frontend signedness semantics. !torch.tensor<*,si32> // First-class support for unknown dtypes. !torch.tensor<[?,?,?],unk> // Standard MLIR representation of `?` for unknown dimensions. !torch.tensor<[?,2,?,4],unk> // Statically shaped / dtyped example. !torch.vtensor<[1,2,3,4],f32> ``` This required fairly significant changes throughout the compiler, but overall it is a big cleanup. We now have a much clearer layering of "the Torch frontend lowering" vs "lowering to std + linalg + etc.". At the C++ level, there is `ValueTensorType`, `NonValueTensorType`. We also have a helper `BaseTensorType` (kind of like ShapedType) which interoperates with those two. Included changes: - New `torch.tensor(dense<0.0> : tensor<5xf32>) : !torch.tensor` op for creating torch tensor literals in the frontend. - Consistently use signedness for the types (except i1 which I didn't touch -- we need to sort out the situation with !basicpy.BoolType there anyway so will be attending to that soon) - Frontend can annotate whether an argument to the function has value semantics. We currently require this, as our backend contract does not currently allow us to even model the non-value-semantic case. Before, the value-semantic assumption was randomly injected in the middle of the pass pipeline. - Move ArrayToTensor (now called MaximizeValueSemantics) and RefinePublicReturn passes to torch dialect. - The TorchToStd and TorchToLinalg passes are now type conversions from `!torch.vtensor` to `tensor` and use the dialect conversion infra. The overall conversion pipeline is set up following the best practices of the "Type Conversions the Not-So-Hard Way" talk. This required introducing `torch-func-builtin-tensorize` and `torch-finalizing-builtin-tensorize` passes analogous to the upstream bufferization passes with the corresponding names (mostly just copypasta from there). - Misc Torch-level canonicalizations -- we now cleanly layer the lowering to std later in the pipeline, so we are gradually lessening our reliance on random std constant folding before we get to that point. Recommended review order: - New types in TorchTypes.td/TorchTypes.h/TorchDialect.cpp - New ops in TorchOps.td / TorchOps.cpp - Less important / more mechanical stuff - Frontend changes. - Pass changes/additions in `Torch/Transforms` and `Conversion/`
2021-05-21 08:07:18 +08:00
torch.slot "t", %tensor : !torch.tensor
Implement GlobalizeObjectGraph transformation. This required restructuring of how we model TorchScript on import. The main difference is that now we split out a `torch.class_type` that holds methods and declarations of the types of each slot. This is more consistent with TorchScript (our previous representation was "denormalized"). Recommended reading order: 1. check out the description of `torch.class_type` in `TorchOps.td` and look at `test/Dialect/Torch/ops.mlir` and `frontends/pytorch/test/module_import/` to familiarize with the new representation. - Just look at the new IR. The diff between the old names and new names is confusing. 2. check out `test/Dialect/Torch/globalize-object-graph*.mlir` and read along with the pass description in `include/npcomp/Dialect/Torch/Transforms/Passes.td` 3. Read the code in `GlobalizeObjectGraph.cpp` and miscellaneous changes in `ivalue_importer.cpp`, `TorchOps.cpp`, etc.
2021-02-18 03:28:51 +08:00
torch.slot "submodule", %submodule : !torch.nn.Module<"empty">
Introduce native `!torch.none` type. - Add `torch.constant.none` op to construct it (naming is chosen to be analogous to Torch's representation of a prim::Constant with NoneType, rather than using the "singleton" terminology of Basicpy).
2021-06-15 02:36:10 +08:00
torch.slot "ob", %none : !torch.none
Add !torch.str type. - Remove dependence on `!basicpy.BytesType`. - Add `torch.constant.str "s"` analogous to `torch.constant.none`.
2021-06-15 23:29:06 +08:00
torch.slot "s", %str : !torch.str
Implement GlobalizeObjectGraph transformation. This required restructuring of how we model TorchScript on import. The main difference is that now we split out a `torch.class_type` that holds methods and declarations of the types of each slot. This is more consistent with TorchScript (our previous representation was "denormalized"). Recommended reading order: 1. check out the description of `torch.class_type` in `TorchOps.td` and look at `test/Dialect/Torch/ops.mlir` and `frontends/pytorch/test/module_import/` to familiarize with the new representation. - Just look at the new IR. The diff between the old names and new names is confusing. 2. check out `test/Dialect/Torch/globalize-object-graph*.mlir` and read along with the pass description in `include/npcomp/Dialect/Torch/Transforms/Passes.td` 3. Read the code in `GlobalizeObjectGraph.cpp` and miscellaneous changes in `ivalue_importer.cpp`, `TorchOps.cpp`, etc.
2021-02-18 03:28:51 +08:00
} : !torch.nn.Module<"test">
Introduce new shape library design. See the documentation in `docs/shape_lib.md` and `docs/adding_a_shape_function.md` for an overview of the system. This completely overhauls how we represent shape functions. In particular, RefineTypes does not infer shapes anymore (only dtypes). Shape functions are now written in (TorchScript'able) Python. Recommended review order: 1. Read `docs/shape_lib.md` and `docs/adding_a_shape_function.md`. 1. Code and tests for ReifyShapeCalculations, DropShapeCalculations. 1. Code and tests for SimplifyShapeCalculations. 1. shape_lib_gen.py 1. Code and tests for new RefineTypes pass. 1. Random folders/canonicalizers in TorchOps.cpp and associated test in `canonicalize.mlir`. 1. New ReadOnly trait inferred from the registry. 1. Any miscellaneous remaining stuff. Example `-print-ir-after-all` for ElementwiseUnaryModule: [IR lowering dump](https://gist.github.com/silvasean/e4dc8cbc8d00aac7819602e3cbd8e212). Example `-print-ir-after-all` for ElementwiseBinaryModule: [IR lowering dump](https://gist.github.com/silvasean/daf6860ecced732af3568af6b1899113).
2022-03-10 08:44:22 +08:00
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @shape_calculations(%arg0: !torch.vtensor) -> !torch.vtensor {
Introduce new shape library design. See the documentation in `docs/shape_lib.md` and `docs/adding_a_shape_function.md` for an overview of the system. This completely overhauls how we represent shape functions. In particular, RefineTypes does not infer shapes anymore (only dtypes). Shape functions are now written in (TorchScript'able) Python. Recommended review order: 1. Read `docs/shape_lib.md` and `docs/adding_a_shape_function.md`. 1. Code and tests for ReifyShapeCalculations, DropShapeCalculations. 1. Code and tests for SimplifyShapeCalculations. 1. shape_lib_gen.py 1. Code and tests for new RefineTypes pass. 1. Random folders/canonicalizers in TorchOps.cpp and associated test in `canonicalize.mlir`. 1. New ReadOnly trait inferred from the registry. 1. Any miscellaneous remaining stuff. Example `-print-ir-after-all` for ElementwiseUnaryModule: [IR lowering dump](https://gist.github.com/silvasean/e4dc8cbc8d00aac7819602e3cbd8e212). Example `-print-ir-after-all` for ElementwiseBinaryModule: [IR lowering dump](https://gist.github.com/silvasean/daf6860ecced732af3568af6b1899113).
2022-03-10 08:44:22 +08:00
%0 = torch.shape.calculate {
%0 = torch.aten.tanh %arg0 : !torch.vtensor -> !torch.vtensor
torch.shape.calculate.yield %0 : !torch.vtensor
} shapes {
Elide `!torch.` prefix in nested dialect types. This leads to much more succinct types in many cases: ``` !torch.list<!torch.int> !torch.list<int> !torch.tuple<!torch.list<!torch.int>, !torch.list<!torch.int>> !torch.tuple<list<int>, list<int>> !torch.optional<!torch.list<!torch.int>> !torch.optional<list<int>> !torch.list<list<list<tensor>>> !torch.list<!torch.list<!torch.list<!torch.tensor>>> ``` I would like to take this further and allow omitting the `!torch.` prefix in all cases, but that's harder -- for example, we currently use `FuncOp` for functions, and so I don't think we can customize the printing there. It seems like it will be a longer road to getting that level of customization.
2022-03-16 07:22:56 +08:00
%0 = torch.aten.size %arg0 : !torch.vtensor -> !torch.list<int>
torch.shape.calculate.yield.shapes %0 : !torch.list<int>
Introduce new shape library design. See the documentation in `docs/shape_lib.md` and `docs/adding_a_shape_function.md` for an overview of the system. This completely overhauls how we represent shape functions. In particular, RefineTypes does not infer shapes anymore (only dtypes). Shape functions are now written in (TorchScript'able) Python. Recommended review order: 1. Read `docs/shape_lib.md` and `docs/adding_a_shape_function.md`. 1. Code and tests for ReifyShapeCalculations, DropShapeCalculations. 1. Code and tests for SimplifyShapeCalculations. 1. shape_lib_gen.py 1. Code and tests for new RefineTypes pass. 1. Random folders/canonicalizers in TorchOps.cpp and associated test in `canonicalize.mlir`. 1. New ReadOnly trait inferred from the registry. 1. Any miscellaneous remaining stuff. Example `-print-ir-after-all` for ElementwiseUnaryModule: [IR lowering dump](https://gist.github.com/silvasean/e4dc8cbc8d00aac7819602e3cbd8e212). Example `-print-ir-after-all` for ElementwiseBinaryModule: [IR lowering dump](https://gist.github.com/silvasean/daf6860ecced732af3568af6b1899113).
2022-03-10 08:44:22 +08:00
} : !torch.vtensor
return %0 : !torch.vtensor
}
Add minimal support for Union types. A recent PyTorch commit made ConstantPad2d call a helper function with a `Union[int, float]` type annotated. This commit adds minimal support for representing and dealing with that. https://github.com/pytorch/pytorch/pull/73287 Changes: - Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`, along with the importer and CAPI code. - Add support in isValidSubtype for union types. - Adding a canonicalizer for `torch.derefine` to help simplify some code that derefines to a UnionType (this also fixes #664). There is still more work to do for really supporting UnionType well, such as canonicalizing UnionType's so that they can be compared with pointer equality.
2022-03-30 06:57:31 +08:00
[custom op] Generalize shape library logic to work with dtypes (#1594) * [custom op] Generalize shape library logic to work with dtypes This commit generalizes the shape library logic, so that dtype rules for ops can also be expressed using the same mechanism. In other words, each op can now have a shape function and a dtype function specified in Python that is imported during lowering to calculate the shapes and dtypes throught a program. For more information about how to specify a dtype function, see the updated `docs/adding_a_shape_and_dtype_function.md`. For those not familiar with how the shape library works, the file `docs/calculations_lib.md` provides an overview.
2022-12-14 00:25:41 +08:00
func.func @dtype_calculations(%arg0: !torch.vtensor) -> !torch.vtensor {
%0 = torch.dtype.calculate {
%1 = torch.aten.tanh %arg0 : !torch.vtensor -> !torch.vtensor
torch.dtype.calculate.yield %1 : !torch.vtensor
} dtypes {
%2 = torch.prim.dtype %arg0 : !torch.vtensor -> !torch.int
torch.dtype.calculate.yield.dtypes %2 : !torch.int
} : !torch.vtensor
return %0 : !torch.vtensor
}
func.func @promote_dtypes(%ranks: !torch.list<optional<int>>, %dtypes: !torch.list<int>) -> !torch.int {
%0 = torch.promote_dtypes %ranks, %dtypes : (!torch.list<optional<int>>, !torch.list<int>) -> !torch.int
return %0 : !torch.int
}
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @number_type_subtypes(%arg0: !torch.tensor, %arg1: !torch.list<int>, %arg2: !torch.union<float, int>) {
Add minimal support for Union types. A recent PyTorch commit made ConstantPad2d call a helper function with a `Union[int, float]` type annotated. This commit adds minimal support for representing and dealing with that. https://github.com/pytorch/pytorch/pull/73287 Changes: - Adding support for `!torch.union<T1, T2, T3>`/`Torch::UnionType`, along with the importer and CAPI code. - Add support in isValidSubtype for union types. - Adding a canonicalizer for `torch.derefine` to help simplify some code that derefines to a UnionType (this also fixes #664). There is still more work to do for really supporting UnionType well, such as canonicalizing UnionType's so that they can be compared with pointer equality.
2022-03-30 06:57:31 +08:00
%0 = torch.aten.constant_pad_nd %arg0, %arg1, %arg2 : !torch.tensor, !torch.list<int>, !torch.union<float, int> -> !torch.tensor
return
}
Add support for ScalarType::QUInt8 I ran into this while poking around at https://github.com/llvm/torch-mlir/issues/959
2022-06-25 08:20:26 +08:00
func.func private @tensor_legal_dtype$torch.qint8() -> !torch.tensor<*,!torch.qint8>
func.func private @tensor_legal_dtype$torch.quint8() -> !torch.tensor<*,!torch.quint8>
[ONNX] add int16 quantization support (#3446) There is currently no int16 quantization support in torch. This patch adds a new mlir type to correspond to the missing "torch.qint16" type, and enables lowering of quantization-related onnx ops using int16 types. In follow-up patches, custom quantization logic for ops like aten.matmul/aten.mm/aten.convolution may need to be revisited to allow support for qint16. The passes in FuseQuantizedOps.cpp may also need slight modifications.
2024-06-12 13:07:22 +08:00
func.func private @tensor_legal_dtype$torch.qint16() -> !torch.tensor<*,!torch.qint16>
[Torch Dialect] fix isValidSubtype with dynamic dim (#2018)
2023-04-11 16:02:18 +08:00
func.func @prim_list_construct$valid_shape_subtype(%arg0: !torch.vtensor<[1,53,56,96],f16>, %arg1: !torch.vtensor<[1,3,56,96],f16>) -> !torch.list<vtensor<[1,?,56,96],f16>> {
%arg2 = "torch.prim.ListConstruct"(%arg0, %arg1) : (!torch.vtensor<[1,53,56,96],f16>, !torch.vtensor<[1,3,56,96],f16>) -> !torch.list<vtensor<[1,?,56,96],f16>>
return %arg2 : !torch.list<vtensor<[1,?,56,96],f16>>
}
Add verification for torch permute op (#2551) - adds support for an optional verifier to the generated torch op tablegen (GeneratedTorchOps.td) - uses the above to add a verifier for the torch permute op. Motivation: I hit an unclear error from linalg while developing a decomposition pass for pixel_shuffle. The error would have been clearer if the problem had been detected earlier in the invalid aten.permute op. Testing: new tests added. To run added tests, from the base directory run ``` ./build/bin/llvm-lit test/Dialect/Torch/invalid.mlir ```
2023-11-16 03:47:54 +08:00
// Check that verification passes with '-1' as a permutation index.
func.func @torch.permute$negative_index_valid (%arg0: !torch.vtensor<[1,2,3],f32>) -> !torch.vtensor<[1,2,3],f32> {
%intm1 = torch.constant.int -1
%int0 = torch.constant.int 0
%int1 = torch.constant.int 1
%perm = torch.prim.ListConstruct %int0, %int1, %intm1 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
%3 = torch.aten.permute %arg0, %perm : !torch.vtensor<[1,2,3],f32>, !torch.list<int> -> !torch.vtensor<[1,2,3],f32>
return %3 : !torch.vtensor<[1,2,3],f32>
}