1; When EXPENSIVE_CHECKS are enabled, the machine verifier appears between each 2; pass. Ignore it with 'grep -v'. 3; fixme: the following line is added to cleanup bots, will be removed in weeks. 4; RUN: rm -f %S/llc-pipeline.s 5; RUN: llc -O0 -mtriple=amdgcn--amdhsa -disable-verify -debug-pass=Structure < %s 2>&1 \ 6; RUN: | grep -v 'Verify generated machine code' | FileCheck -match-full-lines -strict-whitespace -check-prefix=GCN-O0 %s 7; RUN: llc -O1 -mtriple=amdgcn--amdhsa -disable-verify -debug-pass=Structure < %s 2>&1 \ 8; RUN: | grep -v 'Verify generated machine code' | FileCheck -match-full-lines -strict-whitespace -check-prefix=GCN-O1 %s 9; RUN: llc -O1 -mtriple=amdgcn--amdhsa -disable-verify -amdgpu-scalar-ir-passes -amdgpu-sdwa-peephole \ 10; RUN: -amdgpu-load-store-vectorizer -amdgpu-enable-pre-ra-optimizations -debug-pass=Structure < %s 2>&1 \ 11; RUN: | grep -v 'Verify generated machine code' | FileCheck -match-full-lines -strict-whitespace -check-prefix=GCN-O1-OPTS %s 12; RUN: llc -O2 -mtriple=amdgcn--amdhsa -disable-verify -debug-pass=Structure < %s 2>&1 \ 13; RUN: | grep -v 'Verify generated machine code' | FileCheck -match-full-lines -strict-whitespace -check-prefix=GCN-O2 %s 14; RUN: llc -O3 -mtriple=amdgcn--amdhsa -disable-verify -debug-pass=Structure < %s 2>&1 \ 15; RUN: | grep -v 'Verify generated machine code' | FileCheck -match-full-lines -strict-whitespace -check-prefix=GCN-O3 %s 16 17; REQUIRES: asserts 18 19; GCN-O0:Target Library Information 20; GCN-O0-NEXT:Target Pass Configuration 21; GCN-O0-NEXT:Machine Module Information 22; GCN-O0-NEXT:Target Transform Information 23; GCN-O0-NEXT:Assumption Cache Tracker 24; GCN-O0-NEXT:Profile summary info 25; GCN-O0-NEXT:Argument Register Usage Information Storage 26; GCN-O0-NEXT:Create Garbage Collector Module Metadata 27; GCN-O0-NEXT:Register Usage Information Storage 28; GCN-O0-NEXT:Machine Branch Probability Analysis 29; GCN-O0-NEXT: ModulePass Manager 30; GCN-O0-NEXT: Pre-ISel Intrinsic Lowering 31; GCN-O0-NEXT: AMDGPU Printf lowering 32; GCN-O0-NEXT: FunctionPass Manager 33; GCN-O0-NEXT: Dominator Tree Construction 34; GCN-O0-NEXT: Lower ctors and dtors for AMDGPU 35; GCN-O0-NEXT: Fix function bitcasts for AMDGPU 36; GCN-O0-NEXT: FunctionPass Manager 37; GCN-O0-NEXT: Early propagate attributes from kernels to functions 38; GCN-O0-NEXT: AMDGPU Lower Intrinsics 39; GCN-O0-NEXT: AMDGPU Inline All Functions 40; GCN-O0-NEXT: CallGraph Construction 41; GCN-O0-NEXT: Call Graph SCC Pass Manager 42; GCN-O0-NEXT: Inliner for always_inline functions 43; GCN-O0-NEXT: A No-Op Barrier Pass 44; GCN-O0-NEXT: Lower OpenCL enqueued blocks 45; GCN-O0-NEXT: Lower uses of LDS variables from non-kernel functions 46; GCN-O0-NEXT: FunctionPass Manager 47; GCN-O0-NEXT: Expand Atomic instructions 48; GCN-O0-NEXT: Lower constant intrinsics 49; GCN-O0-NEXT: Remove unreachable blocks from the CFG 50; GCN-O0-NEXT: Expand vector predication intrinsics 51; GCN-O0-NEXT: Scalarize Masked Memory Intrinsics 52; GCN-O0-NEXT: Expand reduction intrinsics 53; GCN-O0-NEXT: AMDGPU Attributor 54; GCN-O0-NEXT: CallGraph Construction 55; GCN-O0-NEXT: Call Graph SCC Pass Manager 56; GCN-O0-NEXT: AMDGPU Annotate Kernel Features 57; GCN-O0-NEXT: FunctionPass Manager 58; GCN-O0-NEXT: AMDGPU Lower Kernel Arguments 59; GCN-O0-NEXT: Lazy Value Information Analysis 60; GCN-O0-NEXT: Lower SwitchInst's to branches 61; GCN-O0-NEXT: Lower invoke and unwind, for unwindless code generators 62; GCN-O0-NEXT: Remove unreachable blocks from the CFG 63; GCN-O0-NEXT: Post-Dominator Tree Construction 64; GCN-O0-NEXT: Dominator Tree Construction 65; GCN-O0-NEXT: Natural Loop Information 66; GCN-O0-NEXT: Legacy Divergence Analysis 67; GCN-O0-NEXT: Unify divergent function exit nodes 68; GCN-O0-NEXT: Lazy Value Information Analysis 69; GCN-O0-NEXT: Lower SwitchInst's to branches 70; GCN-O0-NEXT: Dominator Tree Construction 71; GCN-O0-NEXT: Natural Loop Information 72; GCN-O0-NEXT: Convert irreducible control-flow into natural loops 73; GCN-O0-NEXT: Fixup each natural loop to have a single exit block 74; GCN-O0-NEXT: Post-Dominator Tree Construction 75; GCN-O0-NEXT: Dominance Frontier Construction 76; GCN-O0-NEXT: Detect single entry single exit regions 77; GCN-O0-NEXT: Region Pass Manager 78; GCN-O0-NEXT: Structurize control flow 79; GCN-O0-NEXT: Post-Dominator Tree Construction 80; GCN-O0-NEXT: Natural Loop Information 81; GCN-O0-NEXT: Legacy Divergence Analysis 82; GCN-O0-NEXT: Basic Alias Analysis (stateless AA impl) 83; GCN-O0-NEXT: Function Alias Analysis Results 84; GCN-O0-NEXT: Memory SSA 85; GCN-O0-NEXT: AMDGPU Annotate Uniform Values 86; GCN-O0-NEXT: SI annotate control flow 87; GCN-O0-NEXT: LCSSA Verifier 88; GCN-O0-NEXT: Loop-Closed SSA Form Pass 89; GCN-O0-NEXT: DummyCGSCCPass 90; GCN-O0-NEXT: FunctionPass Manager 91; GCN-O0-NEXT: Safe Stack instrumentation pass 92; GCN-O0-NEXT: Insert stack protectors 93; GCN-O0-NEXT: Dominator Tree Construction 94; GCN-O0-NEXT: Post-Dominator Tree Construction 95; GCN-O0-NEXT: Natural Loop Information 96; GCN-O0-NEXT: Legacy Divergence Analysis 97; GCN-O0-NEXT: AMDGPU DAG->DAG Pattern Instruction Selection 98; GCN-O0-NEXT: MachineDominator Tree Construction 99; GCN-O0-NEXT: SI Fix SGPR copies 100; GCN-O0-NEXT: MachinePostDominator Tree Construction 101; GCN-O0-NEXT: SI Lower i1 Copies 102; GCN-O0-NEXT: Finalize ISel and expand pseudo-instructions 103; GCN-O0-NEXT: Local Stack Slot Allocation 104; GCN-O0-NEXT: Register Usage Information Propagation 105; GCN-O0-NEXT: Eliminate PHI nodes for register allocation 106; GCN-O0-NEXT: SI Lower control flow pseudo instructions 107; GCN-O0-NEXT: Two-Address instruction pass 108; GCN-O0-NEXT: Basic Alias Analysis (stateless AA impl) 109; GCN-O0-NEXT: Function Alias Analysis Results 110; GCN-O0-NEXT: MachineDominator Tree Construction 111; GCN-O0-NEXT: Slot index numbering 112; GCN-O0-NEXT: Live Interval Analysis 113; GCN-O0-NEXT: MachinePostDominator Tree Construction 114; GCN-O0-NEXT: SI Whole Quad Mode 115; GCN-O0-NEXT: Virtual Register Map 116; GCN-O0-NEXT: Live Register Matrix 117; GCN-O0-NEXT: SI Pre-allocate WWM Registers 118; GCN-O0-NEXT: Fast Register Allocator 119; GCN-O0-NEXT: SI lower SGPR spill instructions 120; GCN-O0-NEXT: Fast Register Allocator 121; GCN-O0-NEXT: SI Fix VGPR copies 122; GCN-O0-NEXT: Remove Redundant DEBUG_VALUE analysis 123; GCN-O0-NEXT: Fixup Statepoint Caller Saved 124; GCN-O0-NEXT: Lazy Machine Block Frequency Analysis 125; GCN-O0-NEXT: Machine Optimization Remark Emitter 126; GCN-O0-NEXT: Prologue/Epilogue Insertion & Frame Finalization 127; GCN-O0-NEXT: Post-RA pseudo instruction expansion pass 128; GCN-O0-NEXT: SI post-RA bundler 129; GCN-O0-NEXT: Insert fentry calls 130; GCN-O0-NEXT: Insert XRay ops 131; GCN-O0-NEXT: SI Memory Legalizer 132; GCN-O0-NEXT: MachinePostDominator Tree Construction 133; GCN-O0-NEXT: SI insert wait instructions 134; GCN-O0-NEXT: Insert required mode register values 135; GCN-O0-NEXT: MachineDominator Tree Construction 136; GCN-O0-NEXT: SI Final Branch Preparation 137; GCN-O0-NEXT: Post RA hazard recognizer 138; GCN-O0-NEXT: Branch relaxation pass 139; GCN-O0-NEXT: Register Usage Information Collector Pass 140; GCN-O0-NEXT: Live DEBUG_VALUE analysis 141; GCN-O0-NEXT: Function register usage analysis 142; GCN-O0-NEXT: FunctionPass Manager 143; GCN-O0-NEXT: Lazy Machine Block Frequency Analysis 144; GCN-O0-NEXT: Machine Optimization Remark Emitter 145; GCN-O0-NEXT: AMDGPU Assembly Printer 146; GCN-O0-NEXT: Free MachineFunction 147; GCN-O0-NEXT:Pass Arguments: -domtree 148; GCN-O0-NEXT: FunctionPass Manager 149; GCN-O0-NEXT: Dominator Tree Construction 150 151; GCN-O1:Target Library Information 152; GCN-O1-NEXT:Target Pass Configuration 153; GCN-O1-NEXT:Machine Module Information 154; GCN-O1-NEXT:Target Transform Information 155; GCN-O1-NEXT:Assumption Cache Tracker 156; GCN-O1-NEXT:Profile summary info 157; GCN-O1-NEXT:AMDGPU Address space based Alias Analysis 158; GCN-O1-NEXT:External Alias Analysis 159; GCN-O1-NEXT:Type-Based Alias Analysis 160; GCN-O1-NEXT:Scoped NoAlias Alias Analysis 161; GCN-O1-NEXT:Argument Register Usage Information Storage 162; GCN-O1-NEXT:Create Garbage Collector Module Metadata 163; GCN-O1-NEXT:Machine Branch Probability Analysis 164; GCN-O1-NEXT:Register Usage Information Storage 165; GCN-O1-NEXT: ModulePass Manager 166; GCN-O1-NEXT: Pre-ISel Intrinsic Lowering 167; GCN-O1-NEXT: AMDGPU Printf lowering 168; GCN-O1-NEXT: FunctionPass Manager 169; GCN-O1-NEXT: Dominator Tree Construction 170; GCN-O1-NEXT: Lower ctors and dtors for AMDGPU 171; GCN-O1-NEXT: Fix function bitcasts for AMDGPU 172; GCN-O1-NEXT: FunctionPass Manager 173; GCN-O1-NEXT: Early propagate attributes from kernels to functions 174; GCN-O1-NEXT: AMDGPU Lower Intrinsics 175; GCN-O1-NEXT: AMDGPU Inline All Functions 176; GCN-O1-NEXT: CallGraph Construction 177; GCN-O1-NEXT: Call Graph SCC Pass Manager 178; GCN-O1-NEXT: Inliner for always_inline functions 179; GCN-O1-NEXT: A No-Op Barrier Pass 180; GCN-O1-NEXT: Lower OpenCL enqueued blocks 181; GCN-O1-NEXT: Lower uses of LDS variables from non-kernel functions 182; GCN-O1-NEXT: FunctionPass Manager 183; GCN-O1-NEXT: Infer address spaces 184; GCN-O1-NEXT: Expand Atomic instructions 185; GCN-O1-NEXT: AMDGPU Promote Alloca 186; GCN-O1-NEXT: Dominator Tree Construction 187; GCN-O1-NEXT: SROA 188; GCN-O1-NEXT: Post-Dominator Tree Construction 189; GCN-O1-NEXT: Natural Loop Information 190; GCN-O1-NEXT: Legacy Divergence Analysis 191; GCN-O1-NEXT: AMDGPU IR optimizations 192; GCN-O1-NEXT: Basic Alias Analysis (stateless AA impl) 193; GCN-O1-NEXT: Canonicalize natural loops 194; GCN-O1-NEXT: Scalar Evolution Analysis 195; GCN-O1-NEXT: Loop Pass Manager 196; GCN-O1-NEXT: Canonicalize Freeze Instructions in Loops 197; GCN-O1-NEXT: Induction Variable Users 198; GCN-O1-NEXT: Loop Strength Reduction 199; GCN-O1-NEXT: Basic Alias Analysis (stateless AA impl) 200; GCN-O1-NEXT: Function Alias Analysis Results 201; GCN-O1-NEXT: Merge contiguous icmps into a memcmp 202; GCN-O1-NEXT: Natural Loop Information 203; GCN-O1-NEXT: Lazy Branch Probability Analysis 204; GCN-O1-NEXT: Lazy Block Frequency Analysis 205; GCN-O1-NEXT: Expand memcmp() to load/stores 206; GCN-O1-NEXT: Lower constant intrinsics 207; GCN-O1-NEXT: Remove unreachable blocks from the CFG 208; GCN-O1-NEXT: Natural Loop Information 209; GCN-O1-NEXT: Post-Dominator Tree Construction 210; GCN-O1-NEXT: Branch Probability Analysis 211; GCN-O1-NEXT: Block Frequency Analysis 212; GCN-O1-NEXT: Constant Hoisting 213; GCN-O1-NEXT: Replace intrinsics with calls to vector library 214; GCN-O1-NEXT: Partially inline calls to library functions 215; GCN-O1-NEXT: Expand vector predication intrinsics 216; GCN-O1-NEXT: Scalarize Masked Memory Intrinsics 217; GCN-O1-NEXT: Expand reduction intrinsics 218; GCN-O1-NEXT: AMDGPU Attributor 219; GCN-O1-NEXT: CallGraph Construction 220; GCN-O1-NEXT: Call Graph SCC Pass Manager 221; GCN-O1-NEXT: AMDGPU Annotate Kernel Features 222; GCN-O1-NEXT: FunctionPass Manager 223; GCN-O1-NEXT: AMDGPU Lower Kernel Arguments 224; GCN-O1-NEXT: Dominator Tree Construction 225; GCN-O1-NEXT: Natural Loop Information 226; GCN-O1-NEXT: CodeGen Prepare 227; GCN-O1-NEXT: Lazy Value Information Analysis 228; GCN-O1-NEXT: Lower SwitchInst's to branches 229; GCN-O1-NEXT: Lower invoke and unwind, for unwindless code generators 230; GCN-O1-NEXT: Remove unreachable blocks from the CFG 231; GCN-O1-NEXT: Dominator Tree Construction 232; GCN-O1-NEXT: Basic Alias Analysis (stateless AA impl) 233; GCN-O1-NEXT: Function Alias Analysis Results 234; GCN-O1-NEXT: Flatten the CFG 235; GCN-O1-NEXT: Dominator Tree Construction 236; GCN-O1-NEXT: Post-Dominator Tree Construction 237; GCN-O1-NEXT: Natural Loop Information 238; GCN-O1-NEXT: Legacy Divergence Analysis 239; GCN-O1-NEXT: AMDGPU IR late optimizations 240; GCN-O1-NEXT: Basic Alias Analysis (stateless AA impl) 241; GCN-O1-NEXT: Function Alias Analysis Results 242; GCN-O1-NEXT: Code sinking 243; GCN-O1-NEXT: Legacy Divergence Analysis 244; GCN-O1-NEXT: Unify divergent function exit nodes 245; GCN-O1-NEXT: Lazy Value Information Analysis 246; GCN-O1-NEXT: Lower SwitchInst's to branches 247; GCN-O1-NEXT: Dominator Tree Construction 248; GCN-O1-NEXT: Natural Loop Information 249; GCN-O1-NEXT: Convert irreducible control-flow into natural loops 250; GCN-O1-NEXT: Fixup each natural loop to have a single exit block 251; GCN-O1-NEXT: Post-Dominator Tree Construction 252; GCN-O1-NEXT: Dominance Frontier Construction 253; GCN-O1-NEXT: Detect single entry single exit regions 254; GCN-O1-NEXT: Region Pass Manager 255; GCN-O1-NEXT: Structurize control flow 256; GCN-O1-NEXT: Post-Dominator Tree Construction 257; GCN-O1-NEXT: Natural Loop Information 258; GCN-O1-NEXT: Legacy Divergence Analysis 259; GCN-O1-NEXT: Basic Alias Analysis (stateless AA impl) 260; GCN-O1-NEXT: Function Alias Analysis Results 261; GCN-O1-NEXT: Memory SSA 262; GCN-O1-NEXT: AMDGPU Annotate Uniform Values 263; GCN-O1-NEXT: SI annotate control flow 264; GCN-O1-NEXT: LCSSA Verifier 265; GCN-O1-NEXT: Loop-Closed SSA Form Pass 266; GCN-O1-NEXT: DummyCGSCCPass 267; GCN-O1-NEXT: FunctionPass Manager 268; GCN-O1-NEXT: Safe Stack instrumentation pass 269; GCN-O1-NEXT: Insert stack protectors 270; GCN-O1-NEXT: Dominator Tree Construction 271; GCN-O1-NEXT: Post-Dominator Tree Construction 272; GCN-O1-NEXT: Natural Loop Information 273; GCN-O1-NEXT: Legacy Divergence Analysis 274; GCN-O1-NEXT: Basic Alias Analysis (stateless AA impl) 275; GCN-O1-NEXT: Function Alias Analysis Results 276; GCN-O1-NEXT: Branch Probability Analysis 277; GCN-O1-NEXT: Lazy Branch Probability Analysis 278; GCN-O1-NEXT: Lazy Block Frequency Analysis 279; GCN-O1-NEXT: AMDGPU DAG->DAG Pattern Instruction Selection 280; GCN-O1-NEXT: MachineDominator Tree Construction 281; GCN-O1-NEXT: SI Fix SGPR copies 282; GCN-O1-NEXT: MachinePostDominator Tree Construction 283; GCN-O1-NEXT: SI Lower i1 Copies 284; GCN-O1-NEXT: Finalize ISel and expand pseudo-instructions 285; GCN-O1-NEXT: Lazy Machine Block Frequency Analysis 286; GCN-O1-NEXT: Early Tail Duplication 287; GCN-O1-NEXT: Optimize machine instruction PHIs 288; GCN-O1-NEXT: Slot index numbering 289; GCN-O1-NEXT: Merge disjoint stack slots 290; GCN-O1-NEXT: Local Stack Slot Allocation 291; GCN-O1-NEXT: Remove dead machine instructions 292; GCN-O1-NEXT: MachineDominator Tree Construction 293; GCN-O1-NEXT: Machine Natural Loop Construction 294; GCN-O1-NEXT: Machine Block Frequency Analysis 295; GCN-O1-NEXT: Early Machine Loop Invariant Code Motion 296; GCN-O1-NEXT: MachineDominator Tree Construction 297; GCN-O1-NEXT: Machine Block Frequency Analysis 298; GCN-O1-NEXT: Machine Common Subexpression Elimination 299; GCN-O1-NEXT: MachinePostDominator Tree Construction 300; GCN-O1-NEXT: Machine code sinking 301; GCN-O1-NEXT: Peephole Optimizations 302; GCN-O1-NEXT: Remove dead machine instructions 303; GCN-O1-NEXT: SI Fold Operands 304; GCN-O1-NEXT: GCN DPP Combine 305; GCN-O1-NEXT: SI Load Store Optimizer 306; GCN-O1-NEXT: Remove dead machine instructions 307; GCN-O1-NEXT: SI Shrink Instructions 308; GCN-O1-NEXT: Register Usage Information Propagation 309; GCN-O1-NEXT: Detect Dead Lanes 310; GCN-O1-NEXT: Remove dead machine instructions 311; GCN-O1-NEXT: Process Implicit Definitions 312; GCN-O1-NEXT: Remove unreachable machine basic blocks 313; GCN-O1-NEXT: Live Variable Analysis 314; GCN-O1-NEXT: MachineDominator Tree Construction 315; GCN-O1-NEXT: SI Optimize VGPR LiveRange 316; GCN-O1-NEXT: Eliminate PHI nodes for register allocation 317; GCN-O1-NEXT: SI Lower control flow pseudo instructions 318; GCN-O1-NEXT: Two-Address instruction pass 319; GCN-O1-NEXT: Slot index numbering 320; GCN-O1-NEXT: Live Interval Analysis 321; GCN-O1-NEXT: Machine Natural Loop Construction 322; GCN-O1-NEXT: Simple Register Coalescing 323; GCN-O1-NEXT: Rename Disconnected Subregister Components 324; GCN-O1-NEXT: Machine Instruction Scheduler 325; GCN-O1-NEXT: MachinePostDominator Tree Construction 326; GCN-O1-NEXT: SI Whole Quad Mode 327; GCN-O1-NEXT: Virtual Register Map 328; GCN-O1-NEXT: Live Register Matrix 329; GCN-O1-NEXT: SI Pre-allocate WWM Registers 330; GCN-O1-NEXT: SI optimize exec mask operations pre-RA 331; GCN-O1-NEXT: Machine Natural Loop Construction 332; GCN-O1-NEXT: Machine Block Frequency Analysis 333; GCN-O1-NEXT: Debug Variable Analysis 334; GCN-O1-NEXT: Live Stack Slot Analysis 335; GCN-O1-NEXT: Virtual Register Map 336; GCN-O1-NEXT: Live Register Matrix 337; GCN-O1-NEXT: Bundle Machine CFG Edges 338; GCN-O1-NEXT: Spill Code Placement Analysis 339; GCN-O1-NEXT: Lazy Machine Block Frequency Analysis 340; GCN-O1-NEXT: Machine Optimization Remark Emitter 341; GCN-O1-NEXT: Greedy Register Allocator 342; GCN-O1-NEXT: Virtual Register Rewriter 343; GCN-O1-NEXT: SI lower SGPR spill instructions 344; GCN-O1-NEXT: Virtual Register Map 345; GCN-O1-NEXT: Live Register Matrix 346; GCN-O1-NEXT: Greedy Register Allocator 347; GCN-O1-NEXT: GCN NSA Reassign 348; GCN-O1-NEXT: Virtual Register Rewriter 349; GCN-O1-NEXT: Stack Slot Coloring 350; GCN-O1-NEXT: Machine Copy Propagation Pass 351; GCN-O1-NEXT: Machine Loop Invariant Code Motion 352; GCN-O1-NEXT: SI Fix VGPR copies 353; GCN-O1-NEXT: SI optimize exec mask operations 354; GCN-O1-NEXT: Remove Redundant DEBUG_VALUE analysis 355; GCN-O1-NEXT: Fixup Statepoint Caller Saved 356; GCN-O1-NEXT: PostRA Machine Sink 357; GCN-O1-NEXT: MachineDominator Tree Construction 358; GCN-O1-NEXT: Machine Natural Loop Construction 359; GCN-O1-NEXT: Machine Block Frequency Analysis 360; GCN-O1-NEXT: MachinePostDominator Tree Construction 361; GCN-O1-NEXT: Lazy Machine Block Frequency Analysis 362; GCN-O1-NEXT: Machine Optimization Remark Emitter 363; GCN-O1-NEXT: Shrink Wrapping analysis 364; GCN-O1-NEXT: Prologue/Epilogue Insertion & Frame Finalization 365; GCN-O1-NEXT: Control Flow Optimizer 366; GCN-O1-NEXT: Lazy Machine Block Frequency Analysis 367; GCN-O1-NEXT: Tail Duplication 368; GCN-O1-NEXT: Machine Copy Propagation Pass 369; GCN-O1-NEXT: Post-RA pseudo instruction expansion pass 370; GCN-O1-NEXT: SI Shrink Instructions 371; GCN-O1-NEXT: SI post-RA bundler 372; GCN-O1-NEXT: MachineDominator Tree Construction 373; GCN-O1-NEXT: Machine Natural Loop Construction 374; GCN-O1-NEXT: PostRA Machine Instruction Scheduler 375; GCN-O1-NEXT: Machine Block Frequency Analysis 376; GCN-O1-NEXT: MachinePostDominator Tree Construction 377; GCN-O1-NEXT: Branch Probability Basic Block Placement 378; GCN-O1-NEXT: Insert fentry calls 379; GCN-O1-NEXT: Insert XRay ops 380; GCN-O1-NEXT: SI Memory Legalizer 381; GCN-O1-NEXT: MachinePostDominator Tree Construction 382; GCN-O1-NEXT: SI insert wait instructions 383; GCN-O1-NEXT: Insert required mode register values 384; GCN-O1-NEXT: SI Insert Hard Clauses 385; GCN-O1-NEXT: MachineDominator Tree Construction 386; GCN-O1-NEXT: SI Final Branch Preparation 387; GCN-O1-NEXT: SI peephole optimizations 388; GCN-O1-NEXT: Post RA hazard recognizer 389; GCN-O1-NEXT: Branch relaxation pass 390; GCN-O1-NEXT: Register Usage Information Collector Pass 391; GCN-O1-NEXT: Live DEBUG_VALUE analysis 392; GCN-O1-NEXT: Function register usage analysis 393; GCN-O1-NEXT: FunctionPass Manager 394; GCN-O1-NEXT: Lazy Machine Block Frequency Analysis 395; GCN-O1-NEXT: Machine Optimization Remark Emitter 396; GCN-O1-NEXT: AMDGPU Assembly Printer 397; GCN-O1-NEXT: Free MachineFunction 398; GCN-O1-NEXT:Pass Arguments: -domtree 399; GCN-O1-NEXT: FunctionPass Manager 400; GCN-O1-NEXT: Dominator Tree Construction 401 402; GCN-O1-OPTS:Target Library Information 403; GCN-O1-OPTS-NEXT:Target Pass Configuration 404; GCN-O1-OPTS-NEXT:Machine Module Information 405; GCN-O1-OPTS-NEXT:Target Transform Information 406; GCN-O1-OPTS-NEXT:Assumption Cache Tracker 407; GCN-O1-OPTS-NEXT:Profile summary info 408; GCN-O1-OPTS-NEXT:AMDGPU Address space based Alias Analysis 409; GCN-O1-OPTS-NEXT:External Alias Analysis 410; GCN-O1-OPTS-NEXT:Type-Based Alias Analysis 411; GCN-O1-OPTS-NEXT:Scoped NoAlias Alias Analysis 412; GCN-O1-OPTS-NEXT:Argument Register Usage Information Storage 413; GCN-O1-OPTS-NEXT:Create Garbage Collector Module Metadata 414; GCN-O1-OPTS-NEXT:Machine Branch Probability Analysis 415; GCN-O1-OPTS-NEXT:Register Usage Information Storage 416; GCN-O1-OPTS-NEXT: ModulePass Manager 417; GCN-O1-OPTS-NEXT: Pre-ISel Intrinsic Lowering 418; GCN-O1-OPTS-NEXT: AMDGPU Printf lowering 419; GCN-O1-OPTS-NEXT: FunctionPass Manager 420; GCN-O1-OPTS-NEXT: Dominator Tree Construction 421; GCN-O1-OPTS-NEXT: Lower ctors and dtors for AMDGPU 422; GCN-O1-OPTS-NEXT: Fix function bitcasts for AMDGPU 423; GCN-O1-OPTS-NEXT: FunctionPass Manager 424; GCN-O1-OPTS-NEXT: Early propagate attributes from kernels to functions 425; GCN-O1-OPTS-NEXT: AMDGPU Lower Intrinsics 426; GCN-O1-OPTS-NEXT: AMDGPU Inline All Functions 427; GCN-O1-OPTS-NEXT: CallGraph Construction 428; GCN-O1-OPTS-NEXT: Call Graph SCC Pass Manager 429; GCN-O1-OPTS-NEXT: Inliner for always_inline functions 430; GCN-O1-OPTS-NEXT: A No-Op Barrier Pass 431; GCN-O1-OPTS-NEXT: Lower OpenCL enqueued blocks 432; GCN-O1-OPTS-NEXT: Lower uses of LDS variables from non-kernel functions 433; GCN-O1-OPTS-NEXT: FunctionPass Manager 434; GCN-O1-OPTS-NEXT: Infer address spaces 435; GCN-O1-OPTS-NEXT: Expand Atomic instructions 436; GCN-O1-OPTS-NEXT: AMDGPU Promote Alloca 437; GCN-O1-OPTS-NEXT: Dominator Tree Construction 438; GCN-O1-OPTS-NEXT: SROA 439; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 440; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 441; GCN-O1-OPTS-NEXT: Memory SSA 442; GCN-O1-OPTS-NEXT: Natural Loop Information 443; GCN-O1-OPTS-NEXT: Canonicalize natural loops 444; GCN-O1-OPTS-NEXT: LCSSA Verifier 445; GCN-O1-OPTS-NEXT: Loop-Closed SSA Form Pass 446; GCN-O1-OPTS-NEXT: Scalar Evolution Analysis 447; GCN-O1-OPTS-NEXT: Lazy Branch Probability Analysis 448; GCN-O1-OPTS-NEXT: Lazy Block Frequency Analysis 449; GCN-O1-OPTS-NEXT: Loop Pass Manager 450; GCN-O1-OPTS-NEXT: Loop Invariant Code Motion 451; GCN-O1-OPTS-NEXT: Split GEPs to a variadic base and a constant offset for better CSE 452; GCN-O1-OPTS-NEXT: Speculatively execute instructions 453; GCN-O1-OPTS-NEXT: Scalar Evolution Analysis 454; GCN-O1-OPTS-NEXT: Straight line strength reduction 455; GCN-O1-OPTS-NEXT: Early CSE 456; GCN-O1-OPTS-NEXT: Scalar Evolution Analysis 457; GCN-O1-OPTS-NEXT: Nary reassociation 458; GCN-O1-OPTS-NEXT: Early CSE 459; GCN-O1-OPTS-NEXT: Post-Dominator Tree Construction 460; GCN-O1-OPTS-NEXT: Legacy Divergence Analysis 461; GCN-O1-OPTS-NEXT: AMDGPU IR optimizations 462; GCN-O1-OPTS-NEXT: Canonicalize natural loops 463; GCN-O1-OPTS-NEXT: Scalar Evolution Analysis 464; GCN-O1-OPTS-NEXT: Loop Pass Manager 465; GCN-O1-OPTS-NEXT: Canonicalize Freeze Instructions in Loops 466; GCN-O1-OPTS-NEXT: Induction Variable Users 467; GCN-O1-OPTS-NEXT: Loop Strength Reduction 468; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 469; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 470; GCN-O1-OPTS-NEXT: Merge contiguous icmps into a memcmp 471; GCN-O1-OPTS-NEXT: Natural Loop Information 472; GCN-O1-OPTS-NEXT: Lazy Branch Probability Analysis 473; GCN-O1-OPTS-NEXT: Lazy Block Frequency Analysis 474; GCN-O1-OPTS-NEXT: Expand memcmp() to load/stores 475; GCN-O1-OPTS-NEXT: Lower constant intrinsics 476; GCN-O1-OPTS-NEXT: Remove unreachable blocks from the CFG 477; GCN-O1-OPTS-NEXT: Natural Loop Information 478; GCN-O1-OPTS-NEXT: Post-Dominator Tree Construction 479; GCN-O1-OPTS-NEXT: Branch Probability Analysis 480; GCN-O1-OPTS-NEXT: Block Frequency Analysis 481; GCN-O1-OPTS-NEXT: Constant Hoisting 482; GCN-O1-OPTS-NEXT: Replace intrinsics with calls to vector library 483; GCN-O1-OPTS-NEXT: Partially inline calls to library functions 484; GCN-O1-OPTS-NEXT: Expand vector predication intrinsics 485; GCN-O1-OPTS-NEXT: Scalarize Masked Memory Intrinsics 486; GCN-O1-OPTS-NEXT: Expand reduction intrinsics 487; GCN-O1-OPTS-NEXT: Early CSE 488; GCN-O1-OPTS-NEXT: AMDGPU Attributor 489; GCN-O1-OPTS-NEXT: CallGraph Construction 490; GCN-O1-OPTS-NEXT: Call Graph SCC Pass Manager 491; GCN-O1-OPTS-NEXT: AMDGPU Annotate Kernel Features 492; GCN-O1-OPTS-NEXT: FunctionPass Manager 493; GCN-O1-OPTS-NEXT: AMDGPU Lower Kernel Arguments 494; GCN-O1-OPTS-NEXT: Dominator Tree Construction 495; GCN-O1-OPTS-NEXT: Natural Loop Information 496; GCN-O1-OPTS-NEXT: CodeGen Prepare 497; GCN-O1-OPTS-NEXT: Dominator Tree Construction 498; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 499; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 500; GCN-O1-OPTS-NEXT: Natural Loop Information 501; GCN-O1-OPTS-NEXT: Scalar Evolution Analysis 502; GCN-O1-OPTS-NEXT: GPU Load and Store Vectorizer 503; GCN-O1-OPTS-NEXT: Lazy Value Information Analysis 504; GCN-O1-OPTS-NEXT: Lower SwitchInst's to branches 505; GCN-O1-OPTS-NEXT: Lower invoke and unwind, for unwindless code generators 506; GCN-O1-OPTS-NEXT: Remove unreachable blocks from the CFG 507; GCN-O1-OPTS-NEXT: Dominator Tree Construction 508; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 509; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 510; GCN-O1-OPTS-NEXT: Flatten the CFG 511; GCN-O1-OPTS-NEXT: Dominator Tree Construction 512; GCN-O1-OPTS-NEXT: Post-Dominator Tree Construction 513; GCN-O1-OPTS-NEXT: Natural Loop Information 514; GCN-O1-OPTS-NEXT: Legacy Divergence Analysis 515; GCN-O1-OPTS-NEXT: AMDGPU IR late optimizations 516; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 517; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 518; GCN-O1-OPTS-NEXT: Code sinking 519; GCN-O1-OPTS-NEXT: Legacy Divergence Analysis 520; GCN-O1-OPTS-NEXT: Unify divergent function exit nodes 521; GCN-O1-OPTS-NEXT: Lazy Value Information Analysis 522; GCN-O1-OPTS-NEXT: Lower SwitchInst's to branches 523; GCN-O1-OPTS-NEXT: Dominator Tree Construction 524; GCN-O1-OPTS-NEXT: Natural Loop Information 525; GCN-O1-OPTS-NEXT: Convert irreducible control-flow into natural loops 526; GCN-O1-OPTS-NEXT: Fixup each natural loop to have a single exit block 527; GCN-O1-OPTS-NEXT: Post-Dominator Tree Construction 528; GCN-O1-OPTS-NEXT: Dominance Frontier Construction 529; GCN-O1-OPTS-NEXT: Detect single entry single exit regions 530; GCN-O1-OPTS-NEXT: Region Pass Manager 531; GCN-O1-OPTS-NEXT: Structurize control flow 532; GCN-O1-OPTS-NEXT: Post-Dominator Tree Construction 533; GCN-O1-OPTS-NEXT: Natural Loop Information 534; GCN-O1-OPTS-NEXT: Legacy Divergence Analysis 535; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 536; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 537; GCN-O1-OPTS-NEXT: Memory SSA 538; GCN-O1-OPTS-NEXT: AMDGPU Annotate Uniform Values 539; GCN-O1-OPTS-NEXT: SI annotate control flow 540; GCN-O1-OPTS-NEXT: LCSSA Verifier 541; GCN-O1-OPTS-NEXT: Loop-Closed SSA Form Pass 542; GCN-O1-OPTS-NEXT: DummyCGSCCPass 543; GCN-O1-OPTS-NEXT: FunctionPass Manager 544; GCN-O1-OPTS-NEXT: Safe Stack instrumentation pass 545; GCN-O1-OPTS-NEXT: Insert stack protectors 546; GCN-O1-OPTS-NEXT: Dominator Tree Construction 547; GCN-O1-OPTS-NEXT: Post-Dominator Tree Construction 548; GCN-O1-OPTS-NEXT: Natural Loop Information 549; GCN-O1-OPTS-NEXT: Legacy Divergence Analysis 550; GCN-O1-OPTS-NEXT: Basic Alias Analysis (stateless AA impl) 551; GCN-O1-OPTS-NEXT: Function Alias Analysis Results 552; GCN-O1-OPTS-NEXT: Branch Probability Analysis 553; GCN-O1-OPTS-NEXT: Lazy Branch Probability Analysis 554; GCN-O1-OPTS-NEXT: Lazy Block Frequency Analysis 555; GCN-O1-OPTS-NEXT: AMDGPU DAG->DAG Pattern Instruction Selection 556; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 557; GCN-O1-OPTS-NEXT: SI Fix SGPR copies 558; GCN-O1-OPTS-NEXT: MachinePostDominator Tree Construction 559; GCN-O1-OPTS-NEXT: SI Lower i1 Copies 560; GCN-O1-OPTS-NEXT: Finalize ISel and expand pseudo-instructions 561; GCN-O1-OPTS-NEXT: Lazy Machine Block Frequency Analysis 562; GCN-O1-OPTS-NEXT: Early Tail Duplication 563; GCN-O1-OPTS-NEXT: Optimize machine instruction PHIs 564; GCN-O1-OPTS-NEXT: Slot index numbering 565; GCN-O1-OPTS-NEXT: Merge disjoint stack slots 566; GCN-O1-OPTS-NEXT: Local Stack Slot Allocation 567; GCN-O1-OPTS-NEXT: Remove dead machine instructions 568; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 569; GCN-O1-OPTS-NEXT: Machine Natural Loop Construction 570; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 571; GCN-O1-OPTS-NEXT: Early Machine Loop Invariant Code Motion 572; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 573; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 574; GCN-O1-OPTS-NEXT: Machine Common Subexpression Elimination 575; GCN-O1-OPTS-NEXT: MachinePostDominator Tree Construction 576; GCN-O1-OPTS-NEXT: Machine code sinking 577; GCN-O1-OPTS-NEXT: Peephole Optimizations 578; GCN-O1-OPTS-NEXT: Remove dead machine instructions 579; GCN-O1-OPTS-NEXT: SI Fold Operands 580; GCN-O1-OPTS-NEXT: GCN DPP Combine 581; GCN-O1-OPTS-NEXT: SI Load Store Optimizer 582; GCN-O1-OPTS-NEXT: SI Peephole SDWA 583; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 584; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 585; GCN-O1-OPTS-NEXT: Early Machine Loop Invariant Code Motion 586; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 587; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 588; GCN-O1-OPTS-NEXT: Machine Common Subexpression Elimination 589; GCN-O1-OPTS-NEXT: SI Fold Operands 590; GCN-O1-OPTS-NEXT: Remove dead machine instructions 591; GCN-O1-OPTS-NEXT: SI Shrink Instructions 592; GCN-O1-OPTS-NEXT: Register Usage Information Propagation 593; GCN-O1-OPTS-NEXT: Detect Dead Lanes 594; GCN-O1-OPTS-NEXT: Remove dead machine instructions 595; GCN-O1-OPTS-NEXT: Process Implicit Definitions 596; GCN-O1-OPTS-NEXT: Remove unreachable machine basic blocks 597; GCN-O1-OPTS-NEXT: Live Variable Analysis 598; GCN-O1-OPTS-NEXT: SI Optimize VGPR LiveRange 599; GCN-O1-OPTS-NEXT: Eliminate PHI nodes for register allocation 600; GCN-O1-OPTS-NEXT: SI Lower control flow pseudo instructions 601; GCN-O1-OPTS-NEXT: Two-Address instruction pass 602; GCN-O1-OPTS-NEXT: Slot index numbering 603; GCN-O1-OPTS-NEXT: Live Interval Analysis 604; GCN-O1-OPTS-NEXT: Machine Natural Loop Construction 605; GCN-O1-OPTS-NEXT: Simple Register Coalescing 606; GCN-O1-OPTS-NEXT: Rename Disconnected Subregister Components 607; GCN-O1-OPTS-NEXT: AMDGPU Pre-RA optimizations 608; GCN-O1-OPTS-NEXT: Machine Instruction Scheduler 609; GCN-O1-OPTS-NEXT: MachinePostDominator Tree Construction 610; GCN-O1-OPTS-NEXT: SI Whole Quad Mode 611; GCN-O1-OPTS-NEXT: Virtual Register Map 612; GCN-O1-OPTS-NEXT: Live Register Matrix 613; GCN-O1-OPTS-NEXT: SI Pre-allocate WWM Registers 614; GCN-O1-OPTS-NEXT: SI optimize exec mask operations pre-RA 615; GCN-O1-OPTS-NEXT: Machine Natural Loop Construction 616; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 617; GCN-O1-OPTS-NEXT: Debug Variable Analysis 618; GCN-O1-OPTS-NEXT: Live Stack Slot Analysis 619; GCN-O1-OPTS-NEXT: Virtual Register Map 620; GCN-O1-OPTS-NEXT: Live Register Matrix 621; GCN-O1-OPTS-NEXT: Bundle Machine CFG Edges 622; GCN-O1-OPTS-NEXT: Spill Code Placement Analysis 623; GCN-O1-OPTS-NEXT: Lazy Machine Block Frequency Analysis 624; GCN-O1-OPTS-NEXT: Machine Optimization Remark Emitter 625; GCN-O1-OPTS-NEXT: Greedy Register Allocator 626; GCN-O1-OPTS-NEXT: Virtual Register Rewriter 627; GCN-O1-OPTS-NEXT: SI lower SGPR spill instructions 628; GCN-O1-OPTS-NEXT: Virtual Register Map 629; GCN-O1-OPTS-NEXT: Live Register Matrix 630; GCN-O1-OPTS-NEXT: Greedy Register Allocator 631; GCN-O1-OPTS-NEXT: GCN NSA Reassign 632; GCN-O1-OPTS-NEXT: Virtual Register Rewriter 633; GCN-O1-OPTS-NEXT: Stack Slot Coloring 634; GCN-O1-OPTS-NEXT: Machine Copy Propagation Pass 635; GCN-O1-OPTS-NEXT: Machine Loop Invariant Code Motion 636; GCN-O1-OPTS-NEXT: SI Fix VGPR copies 637; GCN-O1-OPTS-NEXT: SI optimize exec mask operations 638; GCN-O1-OPTS-NEXT: Remove Redundant DEBUG_VALUE analysis 639; GCN-O1-OPTS-NEXT: Fixup Statepoint Caller Saved 640; GCN-O1-OPTS-NEXT: PostRA Machine Sink 641; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 642; GCN-O1-OPTS-NEXT: Machine Natural Loop Construction 643; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 644; GCN-O1-OPTS-NEXT: MachinePostDominator Tree Construction 645; GCN-O1-OPTS-NEXT: Lazy Machine Block Frequency Analysis 646; GCN-O1-OPTS-NEXT: Machine Optimization Remark Emitter 647; GCN-O1-OPTS-NEXT: Shrink Wrapping analysis 648; GCN-O1-OPTS-NEXT: Prologue/Epilogue Insertion & Frame Finalization 649; GCN-O1-OPTS-NEXT: Control Flow Optimizer 650; GCN-O1-OPTS-NEXT: Lazy Machine Block Frequency Analysis 651; GCN-O1-OPTS-NEXT: Tail Duplication 652; GCN-O1-OPTS-NEXT: Machine Copy Propagation Pass 653; GCN-O1-OPTS-NEXT: Post-RA pseudo instruction expansion pass 654; GCN-O1-OPTS-NEXT: SI Shrink Instructions 655; GCN-O1-OPTS-NEXT: SI post-RA bundler 656; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 657; GCN-O1-OPTS-NEXT: Machine Natural Loop Construction 658; GCN-O1-OPTS-NEXT: PostRA Machine Instruction Scheduler 659; GCN-O1-OPTS-NEXT: Machine Block Frequency Analysis 660; GCN-O1-OPTS-NEXT: MachinePostDominator Tree Construction 661; GCN-O1-OPTS-NEXT: Branch Probability Basic Block Placement 662; GCN-O1-OPTS-NEXT: Insert fentry calls 663; GCN-O1-OPTS-NEXT: Insert XRay ops 664; GCN-O1-OPTS-NEXT: SI Memory Legalizer 665; GCN-O1-OPTS-NEXT: MachinePostDominator Tree Construction 666; GCN-O1-OPTS-NEXT: SI insert wait instructions 667; GCN-O1-OPTS-NEXT: Insert required mode register values 668; GCN-O1-OPTS-NEXT: SI Insert Hard Clauses 669; GCN-O1-OPTS-NEXT: MachineDominator Tree Construction 670; GCN-O1-OPTS-NEXT: SI Final Branch Preparation 671; GCN-O1-OPTS-NEXT: SI peephole optimizations 672; GCN-O1-OPTS-NEXT: Post RA hazard recognizer 673; GCN-O1-OPTS-NEXT: Branch relaxation pass 674; GCN-O1-OPTS-NEXT: Register Usage Information Collector Pass 675; GCN-O1-OPTS-NEXT: Live DEBUG_VALUE analysis 676; GCN-O1-OPTS-NEXT: Function register usage analysis 677; GCN-O1-OPTS-NEXT: FunctionPass Manager 678; GCN-O1-OPTS-NEXT: Lazy Machine Block Frequency Analysis 679; GCN-O1-OPTS-NEXT: Machine Optimization Remark Emitter 680; GCN-O1-OPTS-NEXT: AMDGPU Assembly Printer 681; GCN-O1-OPTS-NEXT: Free MachineFunction 682; GCN-O1-OPTS-NEXT:Pass Arguments: -domtree 683; GCN-O1-OPTS-NEXT: FunctionPass Manager 684; GCN-O1-OPTS-NEXT: Dominator Tree Construction 685 686; GCN-O2:Target Library Information 687; GCN-O2-NEXT:Target Pass Configuration 688; GCN-O2-NEXT:Machine Module Information 689; GCN-O2-NEXT:Target Transform Information 690; GCN-O2-NEXT:Assumption Cache Tracker 691; GCN-O2-NEXT:Profile summary info 692; GCN-O2-NEXT:AMDGPU Address space based Alias Analysis 693; GCN-O2-NEXT:External Alias Analysis 694; GCN-O2-NEXT:Type-Based Alias Analysis 695; GCN-O2-NEXT:Scoped NoAlias Alias Analysis 696; GCN-O2-NEXT:Argument Register Usage Information Storage 697; GCN-O2-NEXT:Create Garbage Collector Module Metadata 698; GCN-O2-NEXT:Machine Branch Probability Analysis 699; GCN-O2-NEXT:Register Usage Information Storage 700; GCN-O2-NEXT: ModulePass Manager 701; GCN-O2-NEXT: Pre-ISel Intrinsic Lowering 702; GCN-O2-NEXT: AMDGPU Printf lowering 703; GCN-O2-NEXT: FunctionPass Manager 704; GCN-O2-NEXT: Dominator Tree Construction 705; GCN-O2-NEXT: Lower ctors and dtors for AMDGPU 706; GCN-O2-NEXT: Fix function bitcasts for AMDGPU 707; GCN-O2-NEXT: FunctionPass Manager 708; GCN-O2-NEXT: Early propagate attributes from kernels to functions 709; GCN-O2-NEXT: AMDGPU Lower Intrinsics 710; GCN-O2-NEXT: AMDGPU Inline All Functions 711; GCN-O2-NEXT: CallGraph Construction 712; GCN-O2-NEXT: Call Graph SCC Pass Manager 713; GCN-O2-NEXT: Inliner for always_inline functions 714; GCN-O2-NEXT: A No-Op Barrier Pass 715; GCN-O2-NEXT: Lower OpenCL enqueued blocks 716; GCN-O2-NEXT: Lower uses of LDS variables from non-kernel functions 717; GCN-O2-NEXT: FunctionPass Manager 718; GCN-O2-NEXT: Infer address spaces 719; GCN-O2-NEXT: Expand Atomic instructions 720; GCN-O2-NEXT: AMDGPU Promote Alloca 721; GCN-O2-NEXT: Dominator Tree Construction 722; GCN-O2-NEXT: SROA 723; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 724; GCN-O2-NEXT: Function Alias Analysis Results 725; GCN-O2-NEXT: Memory SSA 726; GCN-O2-NEXT: Natural Loop Information 727; GCN-O2-NEXT: Canonicalize natural loops 728; GCN-O2-NEXT: LCSSA Verifier 729; GCN-O2-NEXT: Loop-Closed SSA Form Pass 730; GCN-O2-NEXT: Scalar Evolution Analysis 731; GCN-O2-NEXT: Lazy Branch Probability Analysis 732; GCN-O2-NEXT: Lazy Block Frequency Analysis 733; GCN-O2-NEXT: Loop Pass Manager 734; GCN-O2-NEXT: Loop Invariant Code Motion 735; GCN-O2-NEXT: Split GEPs to a variadic base and a constant offset for better CSE 736; GCN-O2-NEXT: Speculatively execute instructions 737; GCN-O2-NEXT: Scalar Evolution Analysis 738; GCN-O2-NEXT: Straight line strength reduction 739; GCN-O2-NEXT: Early CSE 740; GCN-O2-NEXT: Scalar Evolution Analysis 741; GCN-O2-NEXT: Nary reassociation 742; GCN-O2-NEXT: Early CSE 743; GCN-O2-NEXT: Post-Dominator Tree Construction 744; GCN-O2-NEXT: Legacy Divergence Analysis 745; GCN-O2-NEXT: AMDGPU IR optimizations 746; GCN-O2-NEXT: Canonicalize natural loops 747; GCN-O2-NEXT: Scalar Evolution Analysis 748; GCN-O2-NEXT: Loop Pass Manager 749; GCN-O2-NEXT: Canonicalize Freeze Instructions in Loops 750; GCN-O2-NEXT: Induction Variable Users 751; GCN-O2-NEXT: Loop Strength Reduction 752; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 753; GCN-O2-NEXT: Function Alias Analysis Results 754; GCN-O2-NEXT: Merge contiguous icmps into a memcmp 755; GCN-O2-NEXT: Natural Loop Information 756; GCN-O2-NEXT: Lazy Branch Probability Analysis 757; GCN-O2-NEXT: Lazy Block Frequency Analysis 758; GCN-O2-NEXT: Expand memcmp() to load/stores 759; GCN-O2-NEXT: Lower constant intrinsics 760; GCN-O2-NEXT: Remove unreachable blocks from the CFG 761; GCN-O2-NEXT: Natural Loop Information 762; GCN-O2-NEXT: Post-Dominator Tree Construction 763; GCN-O2-NEXT: Branch Probability Analysis 764; GCN-O2-NEXT: Block Frequency Analysis 765; GCN-O2-NEXT: Constant Hoisting 766; GCN-O2-NEXT: Replace intrinsics with calls to vector library 767; GCN-O2-NEXT: Partially inline calls to library functions 768; GCN-O2-NEXT: Expand vector predication intrinsics 769; GCN-O2-NEXT: Scalarize Masked Memory Intrinsics 770; GCN-O2-NEXT: Expand reduction intrinsics 771; GCN-O2-NEXT: Early CSE 772; GCN-O2-NEXT: AMDGPU Attributor 773; GCN-O2-NEXT: CallGraph Construction 774; GCN-O2-NEXT: Call Graph SCC Pass Manager 775; GCN-O2-NEXT: AMDGPU Annotate Kernel Features 776; GCN-O2-NEXT: FunctionPass Manager 777; GCN-O2-NEXT: AMDGPU Lower Kernel Arguments 778; GCN-O2-NEXT: Dominator Tree Construction 779; GCN-O2-NEXT: Natural Loop Information 780; GCN-O2-NEXT: CodeGen Prepare 781; GCN-O2-NEXT: Dominator Tree Construction 782; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 783; GCN-O2-NEXT: Function Alias Analysis Results 784; GCN-O2-NEXT: Natural Loop Information 785; GCN-O2-NEXT: Scalar Evolution Analysis 786; GCN-O2-NEXT: GPU Load and Store Vectorizer 787; GCN-O2-NEXT: Lazy Value Information Analysis 788; GCN-O2-NEXT: Lower SwitchInst's to branches 789; GCN-O2-NEXT: Lower invoke and unwind, for unwindless code generators 790; GCN-O2-NEXT: Remove unreachable blocks from the CFG 791; GCN-O2-NEXT: Dominator Tree Construction 792; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 793; GCN-O2-NEXT: Function Alias Analysis Results 794; GCN-O2-NEXT: Flatten the CFG 795; GCN-O2-NEXT: Dominator Tree Construction 796; GCN-O2-NEXT: Post-Dominator Tree Construction 797; GCN-O2-NEXT: Natural Loop Information 798; GCN-O2-NEXT: Legacy Divergence Analysis 799; GCN-O2-NEXT: AMDGPU IR late optimizations 800; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 801; GCN-O2-NEXT: Function Alias Analysis Results 802; GCN-O2-NEXT: Code sinking 803; GCN-O2-NEXT: Legacy Divergence Analysis 804; GCN-O2-NEXT: Unify divergent function exit nodes 805; GCN-O2-NEXT: Lazy Value Information Analysis 806; GCN-O2-NEXT: Lower SwitchInst's to branches 807; GCN-O2-NEXT: Dominator Tree Construction 808; GCN-O2-NEXT: Natural Loop Information 809; GCN-O2-NEXT: Convert irreducible control-flow into natural loops 810; GCN-O2-NEXT: Fixup each natural loop to have a single exit block 811; GCN-O2-NEXT: Post-Dominator Tree Construction 812; GCN-O2-NEXT: Dominance Frontier Construction 813; GCN-O2-NEXT: Detect single entry single exit regions 814; GCN-O2-NEXT: Region Pass Manager 815; GCN-O2-NEXT: Structurize control flow 816; GCN-O2-NEXT: Post-Dominator Tree Construction 817; GCN-O2-NEXT: Natural Loop Information 818; GCN-O2-NEXT: Legacy Divergence Analysis 819; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 820; GCN-O2-NEXT: Function Alias Analysis Results 821; GCN-O2-NEXT: Memory SSA 822; GCN-O2-NEXT: AMDGPU Annotate Uniform Values 823; GCN-O2-NEXT: SI annotate control flow 824; GCN-O2-NEXT: LCSSA Verifier 825; GCN-O2-NEXT: Loop-Closed SSA Form Pass 826; GCN-O2-NEXT: Analysis if a function is memory bound 827; GCN-O2-NEXT: DummyCGSCCPass 828; GCN-O2-NEXT: FunctionPass Manager 829; GCN-O2-NEXT: Safe Stack instrumentation pass 830; GCN-O2-NEXT: Insert stack protectors 831; GCN-O2-NEXT: Dominator Tree Construction 832; GCN-O2-NEXT: Post-Dominator Tree Construction 833; GCN-O2-NEXT: Natural Loop Information 834; GCN-O2-NEXT: Legacy Divergence Analysis 835; GCN-O2-NEXT: Basic Alias Analysis (stateless AA impl) 836; GCN-O2-NEXT: Function Alias Analysis Results 837; GCN-O2-NEXT: Branch Probability Analysis 838; GCN-O2-NEXT: Lazy Branch Probability Analysis 839; GCN-O2-NEXT: Lazy Block Frequency Analysis 840; GCN-O2-NEXT: AMDGPU DAG->DAG Pattern Instruction Selection 841; GCN-O2-NEXT: MachineDominator Tree Construction 842; GCN-O2-NEXT: SI Fix SGPR copies 843; GCN-O2-NEXT: MachinePostDominator Tree Construction 844; GCN-O2-NEXT: SI Lower i1 Copies 845; GCN-O2-NEXT: Finalize ISel and expand pseudo-instructions 846; GCN-O2-NEXT: Lazy Machine Block Frequency Analysis 847; GCN-O2-NEXT: Early Tail Duplication 848; GCN-O2-NEXT: Optimize machine instruction PHIs 849; GCN-O2-NEXT: Slot index numbering 850; GCN-O2-NEXT: Merge disjoint stack slots 851; GCN-O2-NEXT: Local Stack Slot Allocation 852; GCN-O2-NEXT: Remove dead machine instructions 853; GCN-O2-NEXT: MachineDominator Tree Construction 854; GCN-O2-NEXT: Machine Natural Loop Construction 855; GCN-O2-NEXT: Machine Block Frequency Analysis 856; GCN-O2-NEXT: Early Machine Loop Invariant Code Motion 857; GCN-O2-NEXT: MachineDominator Tree Construction 858; GCN-O2-NEXT: Machine Block Frequency Analysis 859; GCN-O2-NEXT: Machine Common Subexpression Elimination 860; GCN-O2-NEXT: MachinePostDominator Tree Construction 861; GCN-O2-NEXT: Machine code sinking 862; GCN-O2-NEXT: Peephole Optimizations 863; GCN-O2-NEXT: Remove dead machine instructions 864; GCN-O2-NEXT: SI Fold Operands 865; GCN-O2-NEXT: GCN DPP Combine 866; GCN-O2-NEXT: SI Load Store Optimizer 867; GCN-O2-NEXT: SI Peephole SDWA 868; GCN-O2-NEXT: Machine Block Frequency Analysis 869; GCN-O2-NEXT: MachineDominator Tree Construction 870; GCN-O2-NEXT: Early Machine Loop Invariant Code Motion 871; GCN-O2-NEXT: MachineDominator Tree Construction 872; GCN-O2-NEXT: Machine Block Frequency Analysis 873; GCN-O2-NEXT: Machine Common Subexpression Elimination 874; GCN-O2-NEXT: SI Fold Operands 875; GCN-O2-NEXT: Remove dead machine instructions 876; GCN-O2-NEXT: SI Shrink Instructions 877; GCN-O2-NEXT: Register Usage Information Propagation 878; GCN-O2-NEXT: Detect Dead Lanes 879; GCN-O2-NEXT: Remove dead machine instructions 880; GCN-O2-NEXT: Process Implicit Definitions 881; GCN-O2-NEXT: Remove unreachable machine basic blocks 882; GCN-O2-NEXT: Live Variable Analysis 883; GCN-O2-NEXT: SI Optimize VGPR LiveRange 884; GCN-O2-NEXT: Eliminate PHI nodes for register allocation 885; GCN-O2-NEXT: SI Lower control flow pseudo instructions 886; GCN-O2-NEXT: Two-Address instruction pass 887; GCN-O2-NEXT: Slot index numbering 888; GCN-O2-NEXT: Live Interval Analysis 889; GCN-O2-NEXT: Machine Natural Loop Construction 890; GCN-O2-NEXT: Simple Register Coalescing 891; GCN-O2-NEXT: Rename Disconnected Subregister Components 892; GCN-O2-NEXT: AMDGPU Pre-RA optimizations 893; GCN-O2-NEXT: Machine Instruction Scheduler 894; GCN-O2-NEXT: MachinePostDominator Tree Construction 895; GCN-O2-NEXT: SI Whole Quad Mode 896; GCN-O2-NEXT: Virtual Register Map 897; GCN-O2-NEXT: Live Register Matrix 898; GCN-O2-NEXT: SI Pre-allocate WWM Registers 899; GCN-O2-NEXT: SI optimize exec mask operations pre-RA 900; GCN-O2-NEXT: SI Form memory clauses 901; GCN-O2-NEXT: Machine Natural Loop Construction 902; GCN-O2-NEXT: Machine Block Frequency Analysis 903; GCN-O2-NEXT: Debug Variable Analysis 904; GCN-O2-NEXT: Live Stack Slot Analysis 905; GCN-O2-NEXT: Virtual Register Map 906; GCN-O2-NEXT: Live Register Matrix 907; GCN-O2-NEXT: Bundle Machine CFG Edges 908; GCN-O2-NEXT: Spill Code Placement Analysis 909; GCN-O2-NEXT: Lazy Machine Block Frequency Analysis 910; GCN-O2-NEXT: Machine Optimization Remark Emitter 911; GCN-O2-NEXT: Greedy Register Allocator 912; GCN-O2-NEXT: Virtual Register Rewriter 913; GCN-O2-NEXT: SI lower SGPR spill instructions 914; GCN-O2-NEXT: Virtual Register Map 915; GCN-O2-NEXT: Live Register Matrix 916; GCN-O2-NEXT: Greedy Register Allocator 917; GCN-O2-NEXT: GCN NSA Reassign 918; GCN-O2-NEXT: Virtual Register Rewriter 919; GCN-O2-NEXT: Stack Slot Coloring 920; GCN-O2-NEXT: Machine Copy Propagation Pass 921; GCN-O2-NEXT: Machine Loop Invariant Code Motion 922; GCN-O2-NEXT: SI Fix VGPR copies 923; GCN-O2-NEXT: SI optimize exec mask operations 924; GCN-O2-NEXT: Remove Redundant DEBUG_VALUE analysis 925; GCN-O2-NEXT: Fixup Statepoint Caller Saved 926; GCN-O2-NEXT: PostRA Machine Sink 927; GCN-O2-NEXT: MachineDominator Tree Construction 928; GCN-O2-NEXT: Machine Natural Loop Construction 929; GCN-O2-NEXT: Machine Block Frequency Analysis 930; GCN-O2-NEXT: MachinePostDominator Tree Construction 931; GCN-O2-NEXT: Lazy Machine Block Frequency Analysis 932; GCN-O2-NEXT: Machine Optimization Remark Emitter 933; GCN-O2-NEXT: Shrink Wrapping analysis 934; GCN-O2-NEXT: Prologue/Epilogue Insertion & Frame Finalization 935; GCN-O2-NEXT: Control Flow Optimizer 936; GCN-O2-NEXT: Lazy Machine Block Frequency Analysis 937; GCN-O2-NEXT: Tail Duplication 938; GCN-O2-NEXT: Machine Copy Propagation Pass 939; GCN-O2-NEXT: Post-RA pseudo instruction expansion pass 940; GCN-O2-NEXT: SI Shrink Instructions 941; GCN-O2-NEXT: SI post-RA bundler 942; GCN-O2-NEXT: MachineDominator Tree Construction 943; GCN-O2-NEXT: Machine Natural Loop Construction 944; GCN-O2-NEXT: PostRA Machine Instruction Scheduler 945; GCN-O2-NEXT: Machine Block Frequency Analysis 946; GCN-O2-NEXT: MachinePostDominator Tree Construction 947; GCN-O2-NEXT: Branch Probability Basic Block Placement 948; GCN-O2-NEXT: Insert fentry calls 949; GCN-O2-NEXT: Insert XRay ops 950; GCN-O2-NEXT: SI Memory Legalizer 951; GCN-O2-NEXT: MachinePostDominator Tree Construction 952; GCN-O2-NEXT: SI insert wait instructions 953; GCN-O2-NEXT: Insert required mode register values 954; GCN-O2-NEXT: SI Insert Hard Clauses 955; GCN-O2-NEXT: MachineDominator Tree Construction 956; GCN-O2-NEXT: SI Final Branch Preparation 957; GCN-O2-NEXT: SI peephole optimizations 958; GCN-O2-NEXT: Post RA hazard recognizer 959; GCN-O2-NEXT: Branch relaxation pass 960; GCN-O2-NEXT: Register Usage Information Collector Pass 961; GCN-O2-NEXT: Live DEBUG_VALUE analysis 962; GCN-O2-NEXT: Function register usage analysis 963; GCN-O2-NEXT: FunctionPass Manager 964; GCN-O2-NEXT: Lazy Machine Block Frequency Analysis 965; GCN-O2-NEXT: Machine Optimization Remark Emitter 966; GCN-O2-NEXT: AMDGPU Assembly Printer 967; GCN-O2-NEXT: Free MachineFunction 968; GCN-O2-NEXT:Pass Arguments: -domtree 969; GCN-O2-NEXT: FunctionPass Manager 970; GCN-O2-NEXT: Dominator Tree Construction 971 972; GCN-O3:Target Library Information 973; GCN-O3-NEXT:Target Pass Configuration 974; GCN-O3-NEXT:Machine Module Information 975; GCN-O3-NEXT:Target Transform Information 976; GCN-O3-NEXT:Assumption Cache Tracker 977; GCN-O3-NEXT:Profile summary info 978; GCN-O3-NEXT:AMDGPU Address space based Alias Analysis 979; GCN-O3-NEXT:External Alias Analysis 980; GCN-O3-NEXT:Type-Based Alias Analysis 981; GCN-O3-NEXT:Scoped NoAlias Alias Analysis 982; GCN-O3-NEXT:Argument Register Usage Information Storage 983; GCN-O3-NEXT:Create Garbage Collector Module Metadata 984; GCN-O3-NEXT:Machine Branch Probability Analysis 985; GCN-O3-NEXT:Register Usage Information Storage 986; GCN-O3-NEXT: ModulePass Manager 987; GCN-O3-NEXT: Pre-ISel Intrinsic Lowering 988; GCN-O3-NEXT: AMDGPU Printf lowering 989; GCN-O3-NEXT: FunctionPass Manager 990; GCN-O3-NEXT: Dominator Tree Construction 991; GCN-O3-NEXT: Lower ctors and dtors for AMDGPU 992; GCN-O3-NEXT: Fix function bitcasts for AMDGPU 993; GCN-O3-NEXT: FunctionPass Manager 994; GCN-O3-NEXT: Early propagate attributes from kernels to functions 995; GCN-O3-NEXT: AMDGPU Lower Intrinsics 996; GCN-O3-NEXT: AMDGPU Inline All Functions 997; GCN-O3-NEXT: CallGraph Construction 998; GCN-O3-NEXT: Call Graph SCC Pass Manager 999; GCN-O3-NEXT: Inliner for always_inline functions 1000; GCN-O3-NEXT: A No-Op Barrier Pass 1001; GCN-O3-NEXT: Lower OpenCL enqueued blocks 1002; GCN-O3-NEXT: Lower uses of LDS variables from non-kernel functions 1003; GCN-O3-NEXT: FunctionPass Manager 1004; GCN-O3-NEXT: Infer address spaces 1005; GCN-O3-NEXT: Expand Atomic instructions 1006; GCN-O3-NEXT: AMDGPU Promote Alloca 1007; GCN-O3-NEXT: Dominator Tree Construction 1008; GCN-O3-NEXT: SROA 1009; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1010; GCN-O3-NEXT: Function Alias Analysis Results 1011; GCN-O3-NEXT: Memory SSA 1012; GCN-O3-NEXT: Natural Loop Information 1013; GCN-O3-NEXT: Canonicalize natural loops 1014; GCN-O3-NEXT: LCSSA Verifier 1015; GCN-O3-NEXT: Loop-Closed SSA Form Pass 1016; GCN-O3-NEXT: Scalar Evolution Analysis 1017; GCN-O3-NEXT: Lazy Branch Probability Analysis 1018; GCN-O3-NEXT: Lazy Block Frequency Analysis 1019; GCN-O3-NEXT: Loop Pass Manager 1020; GCN-O3-NEXT: Loop Invariant Code Motion 1021; GCN-O3-NEXT: Split GEPs to a variadic base and a constant offset for better CSE 1022; GCN-O3-NEXT: Speculatively execute instructions 1023; GCN-O3-NEXT: Scalar Evolution Analysis 1024; GCN-O3-NEXT: Straight line strength reduction 1025; GCN-O3-NEXT: Phi Values Analysis 1026; GCN-O3-NEXT: Function Alias Analysis Results 1027; GCN-O3-NEXT: Memory Dependence Analysis 1028; GCN-O3-NEXT: Optimization Remark Emitter 1029; GCN-O3-NEXT: Global Value Numbering 1030; GCN-O3-NEXT: Scalar Evolution Analysis 1031; GCN-O3-NEXT: Nary reassociation 1032; GCN-O3-NEXT: Early CSE 1033; GCN-O3-NEXT: Post-Dominator Tree Construction 1034; GCN-O3-NEXT: Legacy Divergence Analysis 1035; GCN-O3-NEXT: AMDGPU IR optimizations 1036; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1037; GCN-O3-NEXT: Canonicalize natural loops 1038; GCN-O3-NEXT: Scalar Evolution Analysis 1039; GCN-O3-NEXT: Loop Pass Manager 1040; GCN-O3-NEXT: Canonicalize Freeze Instructions in Loops 1041; GCN-O3-NEXT: Induction Variable Users 1042; GCN-O3-NEXT: Loop Strength Reduction 1043; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1044; GCN-O3-NEXT: Function Alias Analysis Results 1045; GCN-O3-NEXT: Merge contiguous icmps into a memcmp 1046; GCN-O3-NEXT: Natural Loop Information 1047; GCN-O3-NEXT: Lazy Branch Probability Analysis 1048; GCN-O3-NEXT: Lazy Block Frequency Analysis 1049; GCN-O3-NEXT: Expand memcmp() to load/stores 1050; GCN-O3-NEXT: Lower constant intrinsics 1051; GCN-O3-NEXT: Remove unreachable blocks from the CFG 1052; GCN-O3-NEXT: Natural Loop Information 1053; GCN-O3-NEXT: Post-Dominator Tree Construction 1054; GCN-O3-NEXT: Branch Probability Analysis 1055; GCN-O3-NEXT: Block Frequency Analysis 1056; GCN-O3-NEXT: Constant Hoisting 1057; GCN-O3-NEXT: Replace intrinsics with calls to vector library 1058; GCN-O3-NEXT: Partially inline calls to library functions 1059; GCN-O3-NEXT: Expand vector predication intrinsics 1060; GCN-O3-NEXT: Scalarize Masked Memory Intrinsics 1061; GCN-O3-NEXT: Expand reduction intrinsics 1062; GCN-O3-NEXT: Natural Loop Information 1063; GCN-O3-NEXT: Phi Values Analysis 1064; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1065; GCN-O3-NEXT: Function Alias Analysis Results 1066; GCN-O3-NEXT: Memory Dependence Analysis 1067; GCN-O3-NEXT: Lazy Branch Probability Analysis 1068; GCN-O3-NEXT: Lazy Block Frequency Analysis 1069; GCN-O3-NEXT: Optimization Remark Emitter 1070; GCN-O3-NEXT: Global Value Numbering 1071; GCN-O3-NEXT: AMDGPU Attributor 1072; GCN-O3-NEXT: CallGraph Construction 1073; GCN-O3-NEXT: Call Graph SCC Pass Manager 1074; GCN-O3-NEXT: AMDGPU Annotate Kernel Features 1075; GCN-O3-NEXT: FunctionPass Manager 1076; GCN-O3-NEXT: AMDGPU Lower Kernel Arguments 1077; GCN-O3-NEXT: Dominator Tree Construction 1078; GCN-O3-NEXT: Natural Loop Information 1079; GCN-O3-NEXT: CodeGen Prepare 1080; GCN-O3-NEXT: Dominator Tree Construction 1081; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1082; GCN-O3-NEXT: Function Alias Analysis Results 1083; GCN-O3-NEXT: Natural Loop Information 1084; GCN-O3-NEXT: Scalar Evolution Analysis 1085; GCN-O3-NEXT: GPU Load and Store Vectorizer 1086; GCN-O3-NEXT: Lazy Value Information Analysis 1087; GCN-O3-NEXT: Lower SwitchInst's to branches 1088; GCN-O3-NEXT: Lower invoke and unwind, for unwindless code generators 1089; GCN-O3-NEXT: Remove unreachable blocks from the CFG 1090; GCN-O3-NEXT: Dominator Tree Construction 1091; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1092; GCN-O3-NEXT: Function Alias Analysis Results 1093; GCN-O3-NEXT: Flatten the CFG 1094; GCN-O3-NEXT: Dominator Tree Construction 1095; GCN-O3-NEXT: Post-Dominator Tree Construction 1096; GCN-O3-NEXT: Natural Loop Information 1097; GCN-O3-NEXT: Legacy Divergence Analysis 1098; GCN-O3-NEXT: AMDGPU IR late optimizations 1099; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1100; GCN-O3-NEXT: Function Alias Analysis Results 1101; GCN-O3-NEXT: Code sinking 1102; GCN-O3-NEXT: Legacy Divergence Analysis 1103; GCN-O3-NEXT: Unify divergent function exit nodes 1104; GCN-O3-NEXT: Lazy Value Information Analysis 1105; GCN-O3-NEXT: Lower SwitchInst's to branches 1106; GCN-O3-NEXT: Dominator Tree Construction 1107; GCN-O3-NEXT: Natural Loop Information 1108; GCN-O3-NEXT: Convert irreducible control-flow into natural loops 1109; GCN-O3-NEXT: Fixup each natural loop to have a single exit block 1110; GCN-O3-NEXT: Post-Dominator Tree Construction 1111; GCN-O3-NEXT: Dominance Frontier Construction 1112; GCN-O3-NEXT: Detect single entry single exit regions 1113; GCN-O3-NEXT: Region Pass Manager 1114; GCN-O3-NEXT: Structurize control flow 1115; GCN-O3-NEXT: Post-Dominator Tree Construction 1116; GCN-O3-NEXT: Natural Loop Information 1117; GCN-O3-NEXT: Legacy Divergence Analysis 1118; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1119; GCN-O3-NEXT: Function Alias Analysis Results 1120; GCN-O3-NEXT: Memory SSA 1121; GCN-O3-NEXT: AMDGPU Annotate Uniform Values 1122; GCN-O3-NEXT: SI annotate control flow 1123; GCN-O3-NEXT: LCSSA Verifier 1124; GCN-O3-NEXT: Loop-Closed SSA Form Pass 1125; GCN-O3-NEXT: Analysis if a function is memory bound 1126; GCN-O3-NEXT: DummyCGSCCPass 1127; GCN-O3-NEXT: FunctionPass Manager 1128; GCN-O3-NEXT: Safe Stack instrumentation pass 1129; GCN-O3-NEXT: Insert stack protectors 1130; GCN-O3-NEXT: Dominator Tree Construction 1131; GCN-O3-NEXT: Post-Dominator Tree Construction 1132; GCN-O3-NEXT: Natural Loop Information 1133; GCN-O3-NEXT: Legacy Divergence Analysis 1134; GCN-O3-NEXT: Basic Alias Analysis (stateless AA impl) 1135; GCN-O3-NEXT: Function Alias Analysis Results 1136; GCN-O3-NEXT: Branch Probability Analysis 1137; GCN-O3-NEXT: Lazy Branch Probability Analysis 1138; GCN-O3-NEXT: Lazy Block Frequency Analysis 1139; GCN-O3-NEXT: AMDGPU DAG->DAG Pattern Instruction Selection 1140; GCN-O3-NEXT: MachineDominator Tree Construction 1141; GCN-O3-NEXT: SI Fix SGPR copies 1142; GCN-O3-NEXT: MachinePostDominator Tree Construction 1143; GCN-O3-NEXT: SI Lower i1 Copies 1144; GCN-O3-NEXT: Finalize ISel and expand pseudo-instructions 1145; GCN-O3-NEXT: Lazy Machine Block Frequency Analysis 1146; GCN-O3-NEXT: Early Tail Duplication 1147; GCN-O3-NEXT: Optimize machine instruction PHIs 1148; GCN-O3-NEXT: Slot index numbering 1149; GCN-O3-NEXT: Merge disjoint stack slots 1150; GCN-O3-NEXT: Local Stack Slot Allocation 1151; GCN-O3-NEXT: Remove dead machine instructions 1152; GCN-O3-NEXT: MachineDominator Tree Construction 1153; GCN-O3-NEXT: Machine Natural Loop Construction 1154; GCN-O3-NEXT: Machine Block Frequency Analysis 1155; GCN-O3-NEXT: Early Machine Loop Invariant Code Motion 1156; GCN-O3-NEXT: MachineDominator Tree Construction 1157; GCN-O3-NEXT: Machine Block Frequency Analysis 1158; GCN-O3-NEXT: Machine Common Subexpression Elimination 1159; GCN-O3-NEXT: MachinePostDominator Tree Construction 1160; GCN-O3-NEXT: Machine code sinking 1161; GCN-O3-NEXT: Peephole Optimizations 1162; GCN-O3-NEXT: Remove dead machine instructions 1163; GCN-O3-NEXT: SI Fold Operands 1164; GCN-O3-NEXT: GCN DPP Combine 1165; GCN-O3-NEXT: SI Load Store Optimizer 1166; GCN-O3-NEXT: SI Peephole SDWA 1167; GCN-O3-NEXT: Machine Block Frequency Analysis 1168; GCN-O3-NEXT: MachineDominator Tree Construction 1169; GCN-O3-NEXT: Early Machine Loop Invariant Code Motion 1170; GCN-O3-NEXT: MachineDominator Tree Construction 1171; GCN-O3-NEXT: Machine Block Frequency Analysis 1172; GCN-O3-NEXT: Machine Common Subexpression Elimination 1173; GCN-O3-NEXT: SI Fold Operands 1174; GCN-O3-NEXT: Remove dead machine instructions 1175; GCN-O3-NEXT: SI Shrink Instructions 1176; GCN-O3-NEXT: Register Usage Information Propagation 1177; GCN-O3-NEXT: Detect Dead Lanes 1178; GCN-O3-NEXT: Remove dead machine instructions 1179; GCN-O3-NEXT: Process Implicit Definitions 1180; GCN-O3-NEXT: Remove unreachable machine basic blocks 1181; GCN-O3-NEXT: Live Variable Analysis 1182; GCN-O3-NEXT: SI Optimize VGPR LiveRange 1183; GCN-O3-NEXT: Eliminate PHI nodes for register allocation 1184; GCN-O3-NEXT: SI Lower control flow pseudo instructions 1185; GCN-O3-NEXT: Two-Address instruction pass 1186; GCN-O3-NEXT: Slot index numbering 1187; GCN-O3-NEXT: Live Interval Analysis 1188; GCN-O3-NEXT: Machine Natural Loop Construction 1189; GCN-O3-NEXT: Simple Register Coalescing 1190; GCN-O3-NEXT: Rename Disconnected Subregister Components 1191; GCN-O3-NEXT: AMDGPU Pre-RA optimizations 1192; GCN-O3-NEXT: Machine Instruction Scheduler 1193; GCN-O3-NEXT: MachinePostDominator Tree Construction 1194; GCN-O3-NEXT: SI Whole Quad Mode 1195; GCN-O3-NEXT: Virtual Register Map 1196; GCN-O3-NEXT: Live Register Matrix 1197; GCN-O3-NEXT: SI Pre-allocate WWM Registers 1198; GCN-O3-NEXT: SI optimize exec mask operations pre-RA 1199; GCN-O3-NEXT: SI Form memory clauses 1200; GCN-O3-NEXT: Machine Natural Loop Construction 1201; GCN-O3-NEXT: Machine Block Frequency Analysis 1202; GCN-O3-NEXT: Debug Variable Analysis 1203; GCN-O3-NEXT: Live Stack Slot Analysis 1204; GCN-O3-NEXT: Virtual Register Map 1205; GCN-O3-NEXT: Live Register Matrix 1206; GCN-O3-NEXT: Bundle Machine CFG Edges 1207; GCN-O3-NEXT: Spill Code Placement Analysis 1208; GCN-O3-NEXT: Lazy Machine Block Frequency Analysis 1209; GCN-O3-NEXT: Machine Optimization Remark Emitter 1210; GCN-O3-NEXT: Greedy Register Allocator 1211; GCN-O3-NEXT: Virtual Register Rewriter 1212; GCN-O3-NEXT: SI lower SGPR spill instructions 1213; GCN-O3-NEXT: Virtual Register Map 1214; GCN-O3-NEXT: Live Register Matrix 1215; GCN-O3-NEXT: Greedy Register Allocator 1216; GCN-O3-NEXT: GCN NSA Reassign 1217; GCN-O3-NEXT: Virtual Register Rewriter 1218; GCN-O3-NEXT: Stack Slot Coloring 1219; GCN-O3-NEXT: Machine Copy Propagation Pass 1220; GCN-O3-NEXT: Machine Loop Invariant Code Motion 1221; GCN-O3-NEXT: SI Fix VGPR copies 1222; GCN-O3-NEXT: SI optimize exec mask operations 1223; GCN-O3-NEXT: Remove Redundant DEBUG_VALUE analysis 1224; GCN-O3-NEXT: Fixup Statepoint Caller Saved 1225; GCN-O3-NEXT: PostRA Machine Sink 1226; GCN-O3-NEXT: MachineDominator Tree Construction 1227; GCN-O3-NEXT: Machine Natural Loop Construction 1228; GCN-O3-NEXT: Machine Block Frequency Analysis 1229; GCN-O3-NEXT: MachinePostDominator Tree Construction 1230; GCN-O3-NEXT: Lazy Machine Block Frequency Analysis 1231; GCN-O3-NEXT: Machine Optimization Remark Emitter 1232; GCN-O3-NEXT: Shrink Wrapping analysis 1233; GCN-O3-NEXT: Prologue/Epilogue Insertion & Frame Finalization 1234; GCN-O3-NEXT: Control Flow Optimizer 1235; GCN-O3-NEXT: Lazy Machine Block Frequency Analysis 1236; GCN-O3-NEXT: Tail Duplication 1237; GCN-O3-NEXT: Machine Copy Propagation Pass 1238; GCN-O3-NEXT: Post-RA pseudo instruction expansion pass 1239; GCN-O3-NEXT: SI Shrink Instructions 1240; GCN-O3-NEXT: SI post-RA bundler 1241; GCN-O3-NEXT: MachineDominator Tree Construction 1242; GCN-O3-NEXT: Machine Natural Loop Construction 1243; GCN-O3-NEXT: PostRA Machine Instruction Scheduler 1244; GCN-O3-NEXT: Machine Block Frequency Analysis 1245; GCN-O3-NEXT: MachinePostDominator Tree Construction 1246; GCN-O3-NEXT: Branch Probability Basic Block Placement 1247; GCN-O3-NEXT: Insert fentry calls 1248; GCN-O3-NEXT: Insert XRay ops 1249; GCN-O3-NEXT: SI Memory Legalizer 1250; GCN-O3-NEXT: MachinePostDominator Tree Construction 1251; GCN-O3-NEXT: SI insert wait instructions 1252; GCN-O3-NEXT: Insert required mode register values 1253; GCN-O3-NEXT: SI Insert Hard Clauses 1254; GCN-O3-NEXT: MachineDominator Tree Construction 1255; GCN-O3-NEXT: SI Final Branch Preparation 1256; GCN-O3-NEXT: SI peephole optimizations 1257; GCN-O3-NEXT: Post RA hazard recognizer 1258; GCN-O3-NEXT: Branch relaxation pass 1259; GCN-O3-NEXT: Register Usage Information Collector Pass 1260; GCN-O3-NEXT: Live DEBUG_VALUE analysis 1261; GCN-O3-NEXT: Function register usage analysis 1262; GCN-O3-NEXT: FunctionPass Manager 1263; GCN-O3-NEXT: Lazy Machine Block Frequency Analysis 1264; GCN-O3-NEXT: Machine Optimization Remark Emitter 1265; GCN-O3-NEXT: AMDGPU Assembly Printer 1266; GCN-O3-NEXT: Free MachineFunction 1267; GCN-O3-NEXT:Pass Arguments: -domtree 1268; GCN-O3-NEXT: FunctionPass Manager 1269; GCN-O3-NEXT: Dominator Tree Construction 1270 1271define void @empty() { 1272 ret void 1273} 1274