llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
torch-mlir/lib/Dialect/TorchConversion/Transforms/Passes.cpp

119 lines
5.3 KiB
C++
Raw Normal View History

Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
Dual license the torch-mlir project. This commit (with approval from all contributors) dual licenses the torch-mlir project under both the standard LLVM license and the standard PyTorch license. This will facilitate moving code between torch-mlir and the two upstream projects. The standard file comment is now: ``` // This file is licensed under the Apache License v2.0 with LLVM Exceptions. // See https://llvm.org/LICENSE.txt for license information. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // Also available under a BSD-style license. See LICENSE. ``` See `LICENSE` in the project root for the terms of both licenses.
2021-09-30 00:03:40 +08:00
// Also available under a BSD-style license. See LICENSE.
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
//
//===----------------------------------------------------------------------===//
Move TorchConversion dialect and TorchTo* into torch-mlir
2021-09-24 11:59:12 +08:00
#include "torch-mlir/Dialect/TorchConversion/Transforms/Passes.h"
Bump LLVM at 8361c5da30588d3d4a48eae648f53be1feb5cfad
2022-03-16 18:44:23 +08:00
#include "mlir/Conversion/Passes.h"
#include "mlir/Dialect/Func/Transforms/Passes.h"
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
#include "mlir/Dialect/Linalg/Passes.h"
#include "mlir/Dialect/MemRef/Transforms/Passes.h"
[tosa] Add TosaMakeBroadcastable pass to torch-to-tosa pipeline. Fixes broken e2e test ElementwiseAddModule_basic Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2021-11-24 14:25:59 +08:00
#include "mlir/Dialect/Tosa/Transforms/Passes.h"
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
#include "mlir/Pass/PassManager.h"
#include "mlir/Transforms/Passes.h"
Move TorchConversion dialect and TorchTo* into torch-mlir
2021-09-24 11:59:12 +08:00
#include "torch-mlir/Conversion/TorchToLinalg/TorchToLinalg.h"
#include "torch-mlir/Conversion/TorchToSCF/TorchToSCF.h"
#include "torch-mlir/Conversion/TorchToStd/TorchToStd.h"
[MLIR][TORCH] Add TorchToTMTensor pass This pass is added to lower ops, which can not be lowered via the TorchToLinalg pass, such as `torch.bincount` op. This pass also uses torch-mlir's TMTensor Dialect to lower the complex ops. Also add torch.bincount op lowering with the help of TMTensor dialect Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-03-03 00:48:15 +08:00
#include "torch-mlir/Conversion/TorchToTMTensor/TorchToTMTensor.h"
Bump LLVM at 8361c5da30588d3d4a48eae648f53be1feb5cfad
2022-03-16 18:44:23 +08:00
#include "torch-mlir/Conversion/TorchToTosa/TorchToTosa.h"
[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
#include "torch-mlir/Dialect/Torch/Transforms/Passes.h"
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
using namespace mlir;
Move TorchConversion dialect and TorchTo* into torch-mlir
2021-09-24 11:59:12 +08:00
using namespace mlir::torch;
[tosa] Add TosaMakeBroadcastable pass to torch-to-tosa pipeline. Fixes broken e2e test ElementwiseAddModule_basic Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2021-11-24 14:25:59 +08:00
using namespace mlir::tosa;
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
//===----------------------------------------------------------------------===//
// Pass registration
//===----------------------------------------------------------------------===//
namespace {
#define GEN_PASS_REGISTRATION
Move TorchConversion dialect and TorchTo* into torch-mlir
2021-09-24 11:59:12 +08:00
#include "torch-mlir/Dialect/TorchConversion/Transforms/Passes.h.inc"
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
} // end namespace
Move TorchConversion dialect and TorchTo* into torch-mlir
2021-09-24 11:59:12 +08:00
void mlir::torch::registerTorchConversionPasses() {
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
::registerPasses();
mlir::PassPipelineRegistration<Torch::TorchLoweringPipelineOptions>(
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
"torch-backend-to-linalg-on-tensors-backend-pipeline",
"Pipeline lowering torch backend contract to linalg-on-tensors backend "
"contract.",
TorchConversion::createTorchBackendToLinalgOnTensorsBackendPipeline);
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
mlir::PassPipelineRegistration<Torch::TorchLoweringPipelineOptions>(
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
"torch-backend-to-tosa-backend-pipeline",
"Pipeline lowering torch backend contract to TOSA backend "
"contract.",
TorchConversion::createTorchBackendToTosaBackendPipeline);
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
}
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
void TorchConversion::createTorchBackendToLinalgOnTensorsBackendPipeline(
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
OpPassManager &pm, const Torch::TorchLoweringPipelineOptions &options) {
// Check some invariants to catch errors in a clear way.
pm.addPass(
TorchConversion::createVerifyInvariantsBeforeBackendLoweringPass());
// Lower to linalg + guards which is the input to codegen backends.
// We do this first as it tends to involve pattern-matching against constants,
// (e.g. dimensions which must be constant in a ranked programming model)
// and those constants get somewhat obscured by TorchToStd.
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createConvertTorchToTMTensorPass());
pm.addNestedPass<func::FuncOp>(createConvertTorchToLinalgPass());
pm.addNestedPass<func::FuncOp>(createConvertTorchToSCFPass());
[TORCH][MLIR]Add lowering for control flow operations. 1. This commit adds lowering of "while-like" prim loop to scf.while operation. 2. Adds lowering of "for-like" prim loops to scf.for operation. Signed-Off-By: Prateek Gupta <prateek@nod-labs.com>
2022-03-31 22:24:44 +08:00
pm.addNestedPass<func::FuncOp>(createConvertTorchToStdPass());
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(memref::createExpandOpsPass());
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
if (options.optimize) {
// Clean up any non-canonical code introduced above..
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createCanonicalizerPass());
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
// Resolve `dim` ops on tensors (which currently live in the `memref`
// dialect for some reason -- we don't have memrefs at this level).
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(
memref::createResolveShapedTypeResultDimsPass());
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
// The resolution of `dim` ops tends to create identical ops. CSE them.
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createCSEPass());
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
}
Port the bulk of the remaining code to torch-mlir This leaves no real code outside torch-mlir. This also renames the "npcomp backend contract" to "linalg on tensors backend contract" as the name of the abstraction layer that RefBackend (IREE too) accepts.
2021-09-28 02:36:44 +08:00
// Finish the type conversion from `torch` types to the types of the
// linalg-on-tensors backend contract.
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
pm.addPass(TorchConversion::createFuncBackendTypeConversionPass());
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
TorchConversion::createFinalizingBackendTypeConversionPass());
VerifyBackendContract -> VerifyLinalgOnTensorsBackendContract This moves it into TorchConversion since it is only needed there. This removes the Backend/ directory.
2021-09-24 10:18:56 +08:00
// Verify that we have lowered to the form that linalg on tensors backends
// expect. This fails compilation (signalPassFailure) if the IR is not in the
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
// correct form.
VerifyBackendContract -> VerifyLinalgOnTensorsBackendContract This moves it into TorchConversion since it is only needed there. This removes the Backend/ directory.
2021-09-24 10:18:56 +08:00
pm.addPass(TorchConversion::createVerifyLinalgOnTensorsBackendContractPass());
Add TorchToIREE and factor out TorchConversion dialect. This converts a basic list op (torch.prim.ListConstruct) to the IREE dialect. ``` def forward(self, x: float): return [x, x] ``` turns into: ``` builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> { %0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %0 : !torch.list<!torch.float> } ``` which turns into: ``` builtin.func @forward(%arg0: f64) -> !iree.list<f64> { %c1 = constant 1 : index %c0 = constant 0 : index %c2 = constant 2 : index %0 = iree.list.create %c2 : !iree.list<f64> iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64 iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64 return %0 : !iree.list<f64> } ``` As part of doing this, I realized that it was time to formalize the IR form that we reach right before running TorchTo{Linalg,Std,...}. We now call it the "Torch backend contract". We then lower the "Torch backend contract" to the "npcomp backend contract", which involves the new TorchConversion (`torch_c`) dialect, which holds ops that need to operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE list, etc.) and the `!torch` types. This made more sense, as I realized that if I didn't factor out `torch_c` then the Torch dialect would have a dependency on IREE dialect (we previously didn't notice this was an issue because we only depended on `builtin` types), which seemed wrong to me. Recommended review order: - TorchToIREE.cpp / `TorchToIREE/basic.mlir` - Look at the new structure of createTorchScriptToNpcompBackendPipeline. It now lives in TorchConversion/Transforms/Passes.cpp and cleanly calls into `Torch::createTorchScriptToTorchBackendPipeline` for the frontend lowering to the Torch backend contract. - Mechanical change extracting `torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new TorchConversion dialect, and a few passes specific to the lowering from the Torch backend contract to the npcomp backend contract. - Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that we convert lists as part of operand materialization, we need to use the original operands). Also added test for AtenMaxPool2dOp and fixed m_TorchConstantIntList. - TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that are created as part of operand materialization for conv/max pool/avg pool ops in TorchToLinalg.
2021-08-12 05:40:08 +08:00
}
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
void TorchConversion::createTorchBackendToTosaBackendPipeline(
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
OpPassManager &pm, const Torch::TorchLoweringPipelineOptions &options) {
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
// Check some invariants to catch errors in a clear way.
pm.addPass(
TorchConversion::createVerifyInvariantsBeforeBackendLoweringPass());
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createConvertTorchToTosaPass());
[tosa] Add TosaMakeBroadcastable pass to torch-to-tosa pipeline. Fixes broken e2e test ElementwiseAddModule_basic Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2021-11-24 14:25:59 +08:00
// Perform rank broadcasting so TosaToLinalg pass works
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createTosaMakeBroadcastablePass());
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
if (options.optimize) {
// Clean up any non-canonical code introduced above..
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createCanonicalizerPass());
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
// The resolution of `dim` ops tends to create identical ops. CSE them.
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(createCSEPass());
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
}
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
// Finish the type conversion from `torch` types to the types of the
// TOSA backend contract.
pm.addPass(TorchConversion::createFuncBackendTypeConversionPass());
llvm: bump tag to e1318078 (#781) The updated LLVM code includes a patch to create bfloat16 array attributes, thus enabling a different patch to torch-mlir to flesh out support for the bfloat16 type.
2022-04-27 03:27:51 +08:00
pm.addNestedPass<func::FuncOp>(
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
TorchConversion::createFinalizingBackendTypeConversionPass());
// Verify that we have lowered to the form that TOSA backends
// expect. This fails compilation (signalPassFailure) if the IR is not in the
// correct form.
pm.addPass(TorchConversion::createVerifyTosaBackendContractPass());
Implement the lazytensor package (#331) Implement the `lazytensor` python package for converting lazy computations captured by the Lazy Tensor Core into MLIR. This PR also fixes a few things with `torchfx` and its example
2021-09-29 08:25:06 +08:00
}