1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
|
//===- GPUToNVVMPipeline.cpp - Test lowering to NVVM as a sink pass -------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file implements a pass for testing the lowering to NVVM as a generally
// usable sink pass.
//
//===----------------------------------------------------------------------===//
#include "mlir/Conversion/AffineToStandard/AffineToStandard.h"
#include "mlir/Conversion/ArithToLLVM/ArithToLLVM.h"
#include "mlir/Conversion/FuncToLLVM/ConvertFuncToLLVMPass.h"
#include "mlir/Conversion/GPUCommon/GPUCommonPass.h"
#include "mlir/Conversion/GPUToNVVM/GPUToNVVMPass.h"
#include "mlir/Conversion/IndexToLLVM/IndexToLLVM.h"
#include "mlir/Conversion/MathToLLVM/MathToLLVM.h"
#include "mlir/Conversion/NVGPUToNVVM/NVGPUToNVVM.h"
#include "mlir/Conversion/NVVMToLLVM/NVVMToLLVM.h"
#include "mlir/Conversion/ReconcileUnrealizedCasts/ReconcileUnrealizedCasts.h"
#include "mlir/Conversion/SCFToControlFlow/SCFToControlFlow.h"
#include "mlir/Conversion/VectorToSCF/VectorToSCF.h"
#include "mlir/Dialect/GPU/IR/GPUDialect.h"
#include "mlir/Dialect/GPU/Pipelines/Passes.h"
#include "mlir/Dialect/GPU/Transforms/Passes.h"
#include "mlir/Dialect/MemRef/Transforms/Passes.h"
#include "mlir/Pass/PassManager.h"
#include "mlir/Pass/PassOptions.h"
#include "mlir/Transforms/Passes.h"
using namespace mlir;
namespace {
//===----------------------------------------------------------------------===//
// Common pipeline
//===----------------------------------------------------------------------===//
void buildCommonPassPipeline(
OpPassManager &pm, const mlir::gpu::GPUToNVVMPipelineOptions &options) {
pm.addPass(createConvertNVGPUToNVVMPass());
pm.addPass(createGpuKernelOutliningPass());
pm.addPass(createConvertVectorToSCFPass());
pm.addPass(createSCFToControlFlowPass());
pm.addPass(createConvertNVVMToLLVMPass());
pm.addPass(createConvertFuncToLLVMPass());
pm.addPass(memref::createExpandStridedMetadataPass());
GpuNVVMAttachTargetOptions nvvmTargetOptions;
nvvmTargetOptions.triple = options.cubinTriple;
nvvmTargetOptions.chip = options.cubinChip;
nvvmTargetOptions.features = options.cubinFeatures;
nvvmTargetOptions.optLevel = options.optLevel;
nvvmTargetOptions.cmdOptions = options.cmdOptions;
pm.addPass(createGpuNVVMAttachTarget(nvvmTargetOptions));
pm.addPass(createLowerAffinePass());
pm.addPass(createArithToLLVMConversionPass());
ConvertIndexToLLVMPassOptions convertIndexToLLVMPassOpt;
convertIndexToLLVMPassOpt.indexBitwidth = options.indexBitWidth;
pm.addPass(createConvertIndexToLLVMPass(convertIndexToLLVMPassOpt));
pm.addPass(createCanonicalizerPass());
pm.addPass(createCSEPass());
}
//===----------------------------------------------------------------------===//
// GPUModule-specific stuff.
//===----------------------------------------------------------------------===//
void buildGpuPassPipeline(OpPassManager &pm,
const mlir::gpu::GPUToNVVMPipelineOptions &options) {
ConvertGpuOpsToNVVMOpsOptions opt;
opt.useBarePtrCallConv = options.kernelUseBarePtrCallConv;
opt.indexBitwidth = options.indexBitWidth;
pm.addNestedPass<gpu::GPUModuleOp>(createConvertGpuOpsToNVVMOps(opt));
pm.addNestedPass<gpu::GPUModuleOp>(createCanonicalizerPass());
pm.addNestedPass<gpu::GPUModuleOp>(createCSEPass());
pm.addNestedPass<gpu::GPUModuleOp>(createReconcileUnrealizedCastsPass());
}
//===----------------------------------------------------------------------===//
// Host Post-GPU pipeline
//===----------------------------------------------------------------------===//
void buildHostPostPipeline(OpPassManager &pm,
const mlir::gpu::GPUToNVVMPipelineOptions &options) {
GpuToLLVMConversionPassOptions opt;
opt.hostBarePtrCallConv = options.hostUseBarePtrCallConv;
opt.kernelBarePtrCallConv = options.kernelUseBarePtrCallConv;
pm.addPass(createGpuToLLVMConversionPass(opt));
GpuModuleToBinaryPassOptions gpuModuleToBinaryPassOptions;
gpuModuleToBinaryPassOptions.compilationTarget = options.cubinFormat;
pm.addPass(createGpuModuleToBinaryPass(gpuModuleToBinaryPassOptions));
pm.addPass(createConvertMathToLLVMPass());
pm.addPass(createCanonicalizerPass());
pm.addPass(createCSEPass());
pm.addPass(createReconcileUnrealizedCastsPass());
}
} // namespace
void mlir::gpu::buildLowerToNVVMPassPipeline(
OpPassManager &pm, const GPUToNVVMPipelineOptions &options) {
// Common pipelines
buildCommonPassPipeline(pm, options);
// GPUModule-specific stuff
buildGpuPassPipeline(pm, options);
// Host post-GPUModule-specific stuff
buildHostPostPipeline(pm, options);
}
void mlir::gpu::registerGPUToNVVMPipeline() {
PassPipelineRegistration<GPUToNVVMPipelineOptions>(
"gpu-lower-to-nvvm-pipeline",
"The default pipeline lowers main dialects (arith, memref, scf, "
"vector, gpu, and nvgpu) to NVVM. It starts by lowering GPU code to the "
"specified compilation target (default is fatbin) then lowers the host "
"code.",
buildLowerToNVVMPassPipeline);
}
|
