[llvm-branch-commits] [llvm] [AMDGPU] Support Wave Reduction for i16 types - 2 (PR #194810)
via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Tue May 5 07:14:22 PDT 2026
https://github.com/easyonaadit updated https://github.com/llvm/llvm-project/pull/194810
>From 75703526d2137af70bef10fc0e3b3e655b188043 Mon Sep 17 00:00:00 2001
From: Aaditya <Aaditya.AlokDeshpande at amd.com>
Date: Tue, 28 Apr 2026 14:26:15 +0530
Subject: [PATCH] [AMDGPU] Support Wave Reduction for i16 types - 2
Supported Ops: `add`, `sub`.
Supports only the iterative stratergy, DPP is yet
to be supported.
Supports only Fake-16 versions of the lowering.
True-16 support is yet to be added.
---
llvm/lib/Target/AMDGPU/SIISelLowering.cpp | 10 +-
llvm/lib/Target/AMDGPU/SIInstructions.td | 2 +
.../CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll | 800 +++++++++++++----
.../CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll | 825 ++++++++++++++----
4 files changed, 1285 insertions(+), 352 deletions(-)
diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
index 8a5107c2b0d88..37e374646a8e6 100644
--- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -5685,7 +5685,9 @@ static bool is16bitWaveReduction(unsigned Opc) {
Opc == AMDGPU::WAVE_REDUCE_UMIN_PSEUDO_U16 ||
Opc == AMDGPU::WAVE_REDUCE_UMIN_PSEUDO_U16_t16 ||
Opc == AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I16_t16 ||
- Opc == AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I16;
+ Opc == AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I16 ||
+ Opc == AMDGPU::WAVE_REDUCE_ADD_PSEUDO_I16 ||
+ Opc == AMDGPU::WAVE_REDUCE_SUB_PSEUDO_I16;
}
static bool is32bitWaveReduceOperation(unsigned Opc) {
@@ -6066,7 +6068,9 @@ static MachineBasicBlock *lowerWaveReduce(MachineInstr &MI,
MI.getOpcode() == AMDGPU::WAVE_REDUCE_MAX_PSEUDO_I16_t16 ||
MI.getOpcode() == AMDGPU::WAVE_REDUCE_MAX_PSEUDO_I16 ||
MI.getOpcode() == AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I16_t16 ||
- MI.getOpcode() == AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I16;
+ MI.getOpcode() == AMDGPU::WAVE_REDUCE_MIN_PSEUDO_I16 ||
+ MI.getOpcode() == AMDGPU::WAVE_REDUCE_ADD_PSEUDO_I16 ||
+ MI.getOpcode() == AMDGPU::WAVE_REDUCE_SUB_PSEUDO_I16;
bool useRealTrue16 = ST.useRealTrue16Insts();
// Create virtual registers required for lowering.
const TargetRegisterClass *WaveMaskRegClass = TRI->getWaveMaskRegClass();
@@ -6766,6 +6770,7 @@ SITargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
ST.getGeneration() >= AMDGPUSubtarget::GFX12
? AMDGPU::V_MAX_NUM_F64_e64
: AMDGPU::V_MAX_F64_e64);
+ case AMDGPU::WAVE_REDUCE_ADD_PSEUDO_I16:
case AMDGPU::WAVE_REDUCE_ADD_PSEUDO_I32:
return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_ADD_I32);
case AMDGPU::WAVE_REDUCE_ADD_PSEUDO_U64:
@@ -6777,6 +6782,7 @@ SITargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
ST.getGeneration() >= AMDGPUSubtarget::GFX12
? AMDGPU::V_ADD_F64_pseudo_e64
: AMDGPU::V_ADD_F64_e64);
+ case AMDGPU::WAVE_REDUCE_SUB_PSEUDO_I16:
case AMDGPU::WAVE_REDUCE_SUB_PSEUDO_I32:
return lowerWaveReduce(MI, *BB, *getSubtarget(), AMDGPU::S_SUB_I32);
case AMDGPU::WAVE_REDUCE_SUB_PSEUDO_U64:
diff --git a/llvm/lib/Target/AMDGPU/SIInstructions.td b/llvm/lib/Target/AMDGPU/SIInstructions.td
index 2d0d9ce0ea1b7..f54e01a6b8656 100644
--- a/llvm/lib/Target/AMDGPU/SIInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SIInstructions.td
@@ -412,6 +412,8 @@ defvar Operations = [
WaveReduceOp<"min", "I16", i16, SGPR_32, VSrc_b16, NotUseRealTrue16Insts>,
WaveReduceOp<"umax", "U16", i16, SGPR_32, VSrc_b16, NotUseRealTrue16Insts>,
WaveReduceOp<"max", "I16", i16, SGPR_32, VSrc_b16, NotUseRealTrue16Insts>,
+ WaveReduceOp<"add", "I16", i16, SGPR_32, VSrc_b16, NotUseRealTrue16Insts>,
+ WaveReduceOp<"sub", "I16", i16, SGPR_32, VSrc_b16, NotUseRealTrue16Insts>,
WaveReduceOp<"umin", "U16_t16", i16, SGPR_32, VSrcT_b16, UseRealTrue16Insts>,
WaveReduceOp<"min", "I16_t16", i16, SGPR_32, VSrcT_b16, UseRealTrue16Insts>,
WaveReduceOp<"umax", "U16_t16", i16, SGPR_32, VSrcT_b16, UseRealTrue16Insts>,
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll
index eddd57c8da852..c92181a56b225 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.add.ll
@@ -7,11 +7,459 @@
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=1 -new-reg-bank-select -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1064GISEL %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=0 < %s | FileCheck -check-prefixes=GFX1032DAGISEL %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GFX1032GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164DAGISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 < %s | FileCheck -check-prefixes=GFX1132DAGISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GFX1132GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel=0 < %s | FileCheck -check-prefixes=GFX12DAGISEL %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=0 -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164DAGISEL,GFX1164DAGISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164GISEL,GFX1164GISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=0 < %s | FileCheck -check-prefixes=GFX1132DAGISEL,GFX1132DAGISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GFX1132GISEL,GFX1132GISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1200 -global-isel=0 < %s | FileCheck -check-prefixes=GFX12DAGISEL,GFX12DAGISEL-FAKE16 %s
+
+define amdgpu_kernel void @uniform_value_i16(ptr addrspace(1) %out, i16 %in) {
+; GFX8DAGISEL-LABEL: uniform_value_i16:
+; GFX8DAGISEL: ; %bb.0: ; %entry
+; GFX8DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8DAGISEL-NEXT: s_mul_i32 s2, s6, s2
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX8DAGISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8DAGISEL-NEXT: s_endpgm
+;
+; GFX8GISEL-LABEL: uniform_value_i16:
+; GFX8GISEL: ; %bb.0: ; %entry
+; GFX8GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8GISEL-NEXT: s_and_b32 s3, 0xffff, s6
+; GFX8GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0
+; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s1
+; GFX8GISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8GISEL-NEXT: s_endpgm
+;
+; GFX9DAGISEL-LABEL: uniform_value_i16:
+; GFX9DAGISEL: ; %bb.0: ; %entry
+; GFX9DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9DAGISEL-NEXT: s_mul_i32 s2, s6, s2
+; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX9DAGISEL-NEXT: global_store_short v0, v1, s[0:1]
+; GFX9DAGISEL-NEXT: s_endpgm
+;
+; GFX9GISEL-LABEL: uniform_value_i16:
+; GFX9GISEL: ; %bb.0: ; %entry
+; GFX9GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9GISEL-NEXT: s_and_b32 s3, 0xffff, s6
+; GFX9GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX9GISEL-NEXT: global_store_short v1, v0, s[0:1]
+; GFX9GISEL-NEXT: s_endpgm
+;
+; GFX1064DAGISEL-LABEL: uniform_value_i16:
+; GFX1064DAGISEL: ; %bb.0: ; %entry
+; GFX1064DAGISEL-NEXT: s_clause 0x1
+; GFX1064DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s6, s2
+; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX1064DAGISEL-NEXT: global_store_short v0, v1, s[0:1]
+; GFX1064DAGISEL-NEXT: s_endpgm
+;
+; GFX1064GISEL-LABEL: uniform_value_i16:
+; GFX1064GISEL: ; %bb.0: ; %entry
+; GFX1064GISEL-NEXT: s_clause 0x1
+; GFX1064GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1064GISEL-NEXT: s_and_b32 s3, 0xffff, s6
+; GFX1064GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1064GISEL-NEXT: global_store_short v1, v0, s[0:1]
+; GFX1064GISEL-NEXT: s_endpgm
+;
+; GFX1032DAGISEL-LABEL: uniform_value_i16:
+; GFX1032DAGISEL: ; %bb.0: ; %entry
+; GFX1032DAGISEL-NEXT: s_clause 0x1
+; GFX1032DAGISEL-NEXT: s_load_dword s2, s[4:5], 0x2c
+; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX1032DAGISEL-NEXT: global_store_short v0, v1, s[0:1]
+; GFX1032DAGISEL-NEXT: s_endpgm
+;
+; GFX1032GISEL-LABEL: uniform_value_i16:
+; GFX1032GISEL: ; %bb.0: ; %entry
+; GFX1032GISEL-NEXT: s_clause 0x1
+; GFX1032GISEL-NEXT: s_load_dword s2, s[4:5], 0x2c
+; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1032GISEL-NEXT: s_and_b32 s2, 0xffff, s2
+; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1032GISEL-NEXT: global_store_short v1, v0, s[0:1]
+; GFX1032GISEL-NEXT: s_endpgm
+;
+; GFX1164DAGISEL-LABEL: uniform_value_i16:
+; GFX1164DAGISEL: ; %bb.0: ; %entry
+; GFX1164DAGISEL-NEXT: s_clause 0x1
+; GFX1164DAGISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
+; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s6, s2
+; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX1164DAGISEL-NEXT: global_store_b16 v0, v1, s[0:1]
+; GFX1164DAGISEL-NEXT: s_endpgm
+;
+; GFX1164GISEL-LABEL: uniform_value_i16:
+; GFX1164GISEL: ; %bb.0: ; %entry
+; GFX1164GISEL-NEXT: s_clause 0x1
+; GFX1164GISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
+; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1164GISEL-NEXT: s_and_b32 s3, 0xffff, s6
+; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1164GISEL-NEXT: global_store_b16 v1, v0, s[0:1]
+; GFX1164GISEL-NEXT: s_endpgm
+;
+; GFX1132DAGISEL-LABEL: uniform_value_i16:
+; GFX1132DAGISEL: ; %bb.0: ; %entry
+; GFX1132DAGISEL-NEXT: s_clause 0x1
+; GFX1132DAGISEL-NEXT: s_load_b32 s2, s[4:5], 0x2c
+; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s2
+; GFX1132DAGISEL-NEXT: global_store_b16 v0, v1, s[0:1]
+; GFX1132DAGISEL-NEXT: s_endpgm
+;
+; GFX1132GISEL-LABEL: uniform_value_i16:
+; GFX1132GISEL: ; %bb.0: ; %entry
+; GFX1132GISEL-NEXT: s_clause 0x1
+; GFX1132GISEL-NEXT: s_load_b32 s2, s[4:5], 0x2c
+; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1132GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1132GISEL-NEXT: s_and_b32 s2, 0xffff, s2
+; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1132GISEL-NEXT: global_store_b16 v1, v0, s[0:1]
+; GFX1132GISEL-NEXT: s_endpgm
+;
+; GFX12DAGISEL-LABEL: uniform_value_i16:
+; GFX12DAGISEL: ; %bb.0: ; %entry
+; GFX12DAGISEL-NEXT: s_load_b96 s[0:2], s[4:5], 0x24
+; GFX12DAGISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX12DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX12DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
+; GFX12DAGISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX12DAGISEL-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s2
+; GFX12DAGISEL-NEXT: global_store_b16 v0, v1, s[0:1]
+; GFX12DAGISEL-NEXT: s_endpgm
+entry:
+ %result = call i16 @llvm.amdgcn.wave.reduce.add.i16(i16 %in, i32 1)
+ store i16 %result, ptr addrspace(1) %out
+ ret void
+}
+
+define void @divergent_value_i16(ptr addrspace(1) %out, i16 %in) {
+; GFX8DAGISEL-LABEL: divergent_value_i16:
+; GFX8DAGISEL: ; %bb.0: ; %entry
+; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX8DAGISEL-NEXT: s_mov_b32 s6, 0
+; GFX8DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX8DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX8DAGISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX8DAGISEL-NEXT: s_add_i32 s6, s6, s8
+; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8DAGISEL-NEXT: ; %bb.2:
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX8DAGISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX8GISEL-LABEL: divergent_value_i16:
+; GFX8GISEL: ; %bb.0: ; %entry
+; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8GISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX8GISEL-NEXT: s_mov_b32 s6, 0
+; GFX8GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX8GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX8GISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX8GISEL-NEXT: s_add_i32 s6, s6, s8
+; GFX8GISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8GISEL-NEXT: ; %bb.2:
+; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX8GISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8GISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX8GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX9DAGISEL-LABEL: divergent_value_i16:
+; GFX9DAGISEL: ; %bb.0: ; %entry
+; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX9DAGISEL-NEXT: s_mov_b32 s6, 0
+; GFX9DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX9DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX9DAGISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX9DAGISEL-NEXT: s_add_i32 s6, s6, s8
+; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9DAGISEL-NEXT: ; %bb.2:
+; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX9DAGISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX9GISEL-LABEL: divergent_value_i16:
+; GFX9GISEL: ; %bb.0: ; %entry
+; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX9GISEL-NEXT: s_mov_b32 s6, 0
+; GFX9GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX9GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX9GISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX9GISEL-NEXT: s_add_i32 s6, s6, s8
+; GFX9GISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9GISEL-NEXT: ; %bb.2:
+; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX9GISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX9GISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX9GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1064DAGISEL-LABEL: divergent_value_i16:
+; GFX1064DAGISEL: ; %bb.0: ; %entry
+; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1064DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX1064DAGISEL-NEXT: s_mov_b32 s6, 0
+; GFX1064DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX1064DAGISEL-NEXT: s_add_i32 s6, s6, s8
+; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064DAGISEL-NEXT: ; %bb.2:
+; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX1064DAGISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1064GISEL-LABEL: divergent_value_i16:
+; GFX1064GISEL: ; %bb.0: ; %entry
+; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1064GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX1064GISEL-NEXT: s_mov_b32 s6, 0
+; GFX1064GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX1064GISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX1064GISEL-NEXT: s_add_i32 s6, s6, s8
+; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064GISEL-NEXT: ; %bb.2:
+; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX1064GISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1032DAGISEL-LABEL: divergent_value_i16:
+; GFX1032DAGISEL: ; %bb.0: ; %entry
+; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1032DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1032DAGISEL-NEXT: s_mov_b32 s5, exec_lo
+; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0
+; GFX1032DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s6, s5
+; GFX1032DAGISEL-NEXT: v_readlane_b32 s7, v2, s6
+; GFX1032DAGISEL-NEXT: s_bitset0_b32 s5, s6
+; GFX1032DAGISEL-NEXT: s_add_i32 s4, s4, s7
+; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s5, 0
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032DAGISEL-NEXT: ; %bb.2:
+; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4
+; GFX1032DAGISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1032GISEL-LABEL: divergent_value_i16:
+; GFX1032GISEL: ; %bb.0: ; %entry
+; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1032GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1032GISEL-NEXT: s_mov_b32 s5, exec_lo
+; GFX1032GISEL-NEXT: s_mov_b32 s4, 0
+; GFX1032GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: s_ff1_i32_b32 s6, s5
+; GFX1032GISEL-NEXT: v_readlane_b32 s7, v2, s6
+; GFX1032GISEL-NEXT: s_bitset0_b32 s5, s6
+; GFX1032GISEL-NEXT: s_add_i32 s4, s4, s7
+; GFX1032GISEL-NEXT: s_cmp_lg_u32 s5, 0
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032GISEL-NEXT: ; %bb.2:
+; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4
+; GFX1032GISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1164DAGISEL-LABEL: divergent_value_i16:
+; GFX1164DAGISEL: ; %bb.0: ; %entry
+; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1164DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1164DAGISEL-NEXT: s_mov_b64 s[0:1], exec
+; GFX1164DAGISEL-NEXT: s_mov_b32 s2, 0
+; GFX1164DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s3, s[0:1]
+; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s3
+; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[0:1], s3
+; GFX1164DAGISEL-NEXT: s_add_i32 s2, s2, s4
+; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164DAGISEL-NEXT: ; %bb.2:
+; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX1164DAGISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1164GISEL-LABEL: divergent_value_i16:
+; GFX1164GISEL: ; %bb.0: ; %entry
+; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1164GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1164GISEL-NEXT: s_mov_b64 s[0:1], exec
+; GFX1164GISEL-NEXT: s_mov_b32 s2, 0
+; GFX1164GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: s_ctz_i32_b64 s3, s[0:1]
+; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s3
+; GFX1164GISEL-NEXT: s_bitset0_b64 s[0:1], s3
+; GFX1164GISEL-NEXT: s_add_i32 s2, s2, s4
+; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164GISEL-NEXT: ; %bb.2:
+; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX1164GISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1132DAGISEL-LABEL: divergent_value_i16:
+; GFX1132DAGISEL: ; %bb.0: ; %entry
+; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1132DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1132DAGISEL-NEXT: s_mov_b32 s1, exec_lo
+; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0
+; GFX1132DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s2, s1
+; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1132DAGISEL-NEXT: v_readlane_b32 s3, v2, s2
+; GFX1132DAGISEL-NEXT: s_bitset0_b32 s1, s2
+; GFX1132DAGISEL-NEXT: s_add_i32 s0, s0, s3
+; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132DAGISEL-NEXT: ; %bb.2:
+; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, s0
+; GFX1132DAGISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1132GISEL-LABEL: divergent_value_i16:
+; GFX1132GISEL: ; %bb.0: ; %entry
+; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1132GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1132GISEL-NEXT: s_mov_b32 s1, exec_lo
+; GFX1132GISEL-NEXT: s_mov_b32 s0, 0
+; GFX1132GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: s_ctz_i32_b32 s2, s1
+; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1132GISEL-NEXT: v_readlane_b32 s3, v2, s2
+; GFX1132GISEL-NEXT: s_bitset0_b32 s1, s2
+; GFX1132GISEL-NEXT: s_add_i32 s0, s0, s3
+; GFX1132GISEL-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132GISEL-NEXT: ; %bb.2:
+; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, s0
+; GFX1132GISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX12DAGISEL-LABEL: divergent_value_i16:
+; GFX12DAGISEL: ; %bb.0: ; %entry
+; GFX12DAGISEL-NEXT: s_wait_loadcnt_dscnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_expcnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_samplecnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_bvhcnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
+; GFX12DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX12DAGISEL-NEXT: s_mov_b32 s1, exec_lo
+; GFX12DAGISEL-NEXT: s_mov_b32 s0, 0
+; GFX12DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
+; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s2, s1
+; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
+; GFX12DAGISEL-NEXT: v_readlane_b32 s3, v2, s2
+; GFX12DAGISEL-NEXT: s_bitset0_b32 s1, s2
+; GFX12DAGISEL-NEXT: s_add_co_i32 s0, s0, s3
+; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
+; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX12DAGISEL-NEXT: ; %bb.2:
+; GFX12DAGISEL-NEXT: v_mov_b32_e32 v2, s0
+; GFX12DAGISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX12DAGISEL-NEXT: s_setpc_b64 s[30:31]
+entry:
+ %result = call i16 @llvm.amdgcn.wave.reduce.add.i16(i16 %in, i32 1)
+ store i16 %result, ptr addrspace(1) %out
+ ret void
+}
define amdgpu_kernel void @uniform_value(ptr addrspace(1) %out, i32 %in) {
; GFX8DAGISEL-LABEL: uniform_value:
@@ -206,13 +654,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX8DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX8DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX8DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX8DAGISEL-NEXT: s_add_i32 s4, s4, s6
; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX8DAGISEL-NEXT: ; %bb.2:
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0
@@ -226,13 +674,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8GISEL-NEXT: s_mov_b32 s4, 0
-; GFX8GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX8GISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX8GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX8GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX8GISEL-NEXT: s_add_i32 s4, s4, s6
; GFX8GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX8GISEL-NEXT: ; %bb.2:
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0
@@ -247,13 +695,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0
; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX9DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX9DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX9DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX9DAGISEL-NEXT: s_add_i32 s4, s4, s6
; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9DAGISEL-NEXT: ; %bb.2:
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -265,13 +713,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9GISEL-NEXT: s_mov_b32 s4, 0
-; GFX9GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX9GISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX9GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX9GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX9GISEL-NEXT: s_add_i32 s4, s4, s6
; GFX9GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9GISEL-NEXT: ; %bb.2:
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -285,13 +733,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, 0
; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX1064DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX1064DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1064DAGISEL-NEXT: s_add_i32 s4, s4, s6
; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064DAGISEL-NEXT: ; %bb.2:
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -303,13 +751,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064GISEL-NEXT: s_mov_b32 s4, 0
-; GFX1064GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1064GISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX1064GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1064GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1064GISEL-NEXT: s_add_i32 s4, s4, s6
; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064GISEL-NEXT: ; %bb.2:
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -323,13 +771,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, 0
; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1032DAGISEL-NEXT: s_mov_b32 s2, 0
-; GFX1032DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s4, s3
; GFX1032DAGISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1032DAGISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1032DAGISEL-NEXT: s_add_i32 s2, s2, s5
; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032DAGISEL-NEXT: ; %bb.2:
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -341,13 +789,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1032GISEL-NEXT: s_mov_b32 s2, 0
-; GFX1032GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1032GISEL-NEXT: s_ff1_i32_b32 s4, s3
; GFX1032GISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1032GISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1032GISEL-NEXT: s_add_i32 s2, s2, s5
; GFX1032GISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032GISEL-NEXT: ; %bb.2:
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -362,14 +810,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX1164DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s5, s[2:3]
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1164DAGISEL-NEXT: s_add_i32 s4, s4, s6
; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164DAGISEL-NEXT: ; %bb.2:
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -382,14 +830,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164GISEL-NEXT: s_mov_b32 s4, 0
-; GFX1164GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1164GISEL-NEXT: s_ctz_i32_b64 s5, s[2:3]
; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1164GISEL-NEXT: s_add_i32 s4, s4, s6
; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164GISEL-NEXT: ; %bb.2:
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -403,14 +851,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, 0
-; GFX1132DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s4, s3
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1132DAGISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1132DAGISEL-NEXT: s_add_i32 s2, s2, s5
; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132DAGISEL-NEXT: ; %bb.2:
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -423,14 +871,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0
; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1132GISEL-NEXT: s_mov_b32 s2, 0
-; GFX1132GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1132GISEL-NEXT: s_ctz_i32_b32 s4, s3
; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1132GISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1132GISEL-NEXT: s_add_i32 s2, s2, s5
; GFX1132GISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132GISEL-NEXT: ; %bb.2:
; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, 0
; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -443,14 +891,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX12DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX12DAGISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX12DAGISEL-NEXT: s_mov_b32 s2, 0
-; GFX12DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s4, s3
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX12DAGISEL-NEXT: s_bitset0_b32 s3, s4
; GFX12DAGISEL-NEXT: s_add_co_i32 s2, s2, s5
; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX12DAGISEL-NEXT: ; %bb.2:
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
@@ -2312,7 +2760,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr2
; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX8DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX8DAGISEL-NEXT: ; %bb.1: ; %else
; GFX8DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2320,24 +2768,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8DAGISEL-NEXT: ; implicit-def: $vgpr0
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8DAGISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX8DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX8DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX8DAGISEL-NEXT: ; %bb.3: ; %if
; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX8DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX8DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX8DAGISEL-NEXT: s_add_i32 s6, s6, s8
; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX8DAGISEL-NEXT: ; %bb.5:
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX8DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX8DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -2352,7 +2800,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8GISEL-NEXT: ; implicit-def: $sgpr2
; GFX8GISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX8GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX8GISEL-NEXT: ; %bb.1: ; %else
; GFX8GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2360,24 +2808,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8GISEL-NEXT: ; implicit-def: $vgpr0
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8GISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX8GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX8GISEL-NEXT: .LBB8_2: ; %Flow
; GFX8GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX8GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX8GISEL-NEXT: ; %bb.3: ; %if
; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8GISEL-NEXT: s_mov_b32 s6, 0
-; GFX8GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX8GISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX8GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX8GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX8GISEL-NEXT: s_add_i32 s6, s6, s8
; GFX8GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX8GISEL-NEXT: ; %bb.5:
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX8GISEL-NEXT: .LBB6_6: ; %endif
+; GFX8GISEL-NEXT: .LBB8_6: ; %endif
; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -2392,7 +2840,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr2
; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX9DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX9DAGISEL-NEXT: ; %bb.1: ; %else
; GFX9DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2400,24 +2848,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9DAGISEL-NEXT: ; implicit-def: $vgpr0
; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9DAGISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX9DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX9DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX9DAGISEL-NEXT: ; %bb.3: ; %if
; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX9DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX9DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX9DAGISEL-NEXT: s_add_i32 s6, s6, s8
; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX9DAGISEL-NEXT: ; %bb.5:
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX9DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX9DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2431,7 +2879,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9GISEL-NEXT: ; implicit-def: $sgpr2
; GFX9GISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX9GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX9GISEL-NEXT: ; %bb.1: ; %else
; GFX9GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2439,24 +2887,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9GISEL-NEXT: ; implicit-def: $vgpr0
; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9GISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX9GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX9GISEL-NEXT: .LBB8_2: ; %Flow
; GFX9GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX9GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX9GISEL-NEXT: ; %bb.3: ; %if
; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9GISEL-NEXT: s_mov_b32 s6, 0
-; GFX9GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX9GISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX9GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX9GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX9GISEL-NEXT: s_add_i32 s6, s6, s8
; GFX9GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX9GISEL-NEXT: ; %bb.5:
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX9GISEL-NEXT: .LBB6_6: ; %endif
+; GFX9GISEL-NEXT: .LBB8_6: ; %endif
; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2470,7 +2918,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr2
; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX1064DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1064DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2478,24 +2926,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX1064DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1064DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1064DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX1064DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1064DAGISEL-NEXT: s_add_i32 s6, s6, s8
; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1064DAGISEL-NEXT: ; %bb.5:
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1064DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1064DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2509,7 +2957,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064GISEL-NEXT: ; implicit-def: $sgpr2
; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX1064GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1064GISEL-NEXT: ; %bb.1: ; %else
; GFX1064GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2517,24 +2965,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064GISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX1064GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1064GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1064GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1064GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1064GISEL-NEXT: ; %bb.3: ; %if
; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064GISEL-NEXT: s_mov_b32 s6, 0
-; GFX1064GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1064GISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX1064GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1064GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1064GISEL-NEXT: s_add_i32 s6, s6, s8
; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1064GISEL-NEXT: ; %bb.5:
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1064GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1064GISEL-NEXT: .LBB8_6: ; %endif
; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2548,7 +2996,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr1
; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo
; GFX1032DAGISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1032DAGISEL-NEXT: s_load_dword s1, s[4:5], 0x2c
; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2556,24 +3004,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2
; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032DAGISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1032DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1032DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1032DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1032DAGISEL-NEXT: s_mov_b32 s1, 0
-; GFX1032DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032DAGISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1032DAGISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1032DAGISEL-NEXT: s_add_i32 s1, s1, s6
; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1032DAGISEL-NEXT: ; %bb.5:
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1032DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1032DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2587,7 +3035,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032GISEL-NEXT: ; implicit-def: $sgpr1
; GFX1032GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo
; GFX1032GISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1032GISEL-NEXT: ; %bb.1: ; %else
; GFX1032GISEL-NEXT: s_load_dword s1, s[4:5], 0x2c
; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2595,24 +3043,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s2, s2
; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032GISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1032GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1032GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1032GISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1032GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1032GISEL-NEXT: ; %bb.3: ; %if
; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1032GISEL-NEXT: s_mov_b32 s1, 0
-; GFX1032GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1032GISEL-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032GISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1032GISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1032GISEL-NEXT: s_add_i32 s1, s1, s6
; GFX1032GISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1032GISEL-NEXT: ; %bb.5:
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1032GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1032GISEL-NEXT: .LBB8_6: ; %endif
; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2628,7 +3076,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1164DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1164DAGISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2637,25 +3085,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX1164DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1164DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1164DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX1164DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s7, s[2:3]
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1164DAGISEL-NEXT: s_add_i32 s6, s6, s8
; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1164DAGISEL-NEXT: ; %bb.5:
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1164DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1164DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2671,7 +3119,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1164GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1164GISEL-NEXT: ; %bb.1: ; %else
; GFX1164GISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2680,25 +3128,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164GISEL-NEXT: s_mul_i32 s2, s6, s2
-; GFX1164GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1164GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1164GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1164GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1164GISEL-NEXT: ; %bb.3: ; %if
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164GISEL-NEXT: s_mov_b32 s6, 0
-; GFX1164GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1164GISEL-NEXT: s_ctz_i32_b64 s7, s[2:3]
; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1164GISEL-NEXT: s_add_i32 s6, s6, s8
; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1164GISEL-NEXT: ; %bb.5:
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1164GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1164GISEL-NEXT: .LBB8_6: ; %endif
; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2714,7 +3162,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1132DAGISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1132DAGISEL-NEXT: s_load_b32 s1, s[4:5], 0x2c
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2723,25 +3171,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2
; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132DAGISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1132DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1132DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1132DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1132DAGISEL-NEXT: s_mov_b32 s1, 0
-; GFX1132DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132DAGISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1132DAGISEL-NEXT: s_add_i32 s1, s1, s6
; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1132DAGISEL-NEXT: ; %bb.5:
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1132DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1132DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2757,7 +3205,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1132GISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1132GISEL-NEXT: ; %bb.1: ; %else
; GFX1132GISEL-NEXT: s_load_b32 s1, s[4:5], 0x2c
; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2766,25 +3214,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s2, s2
; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132GISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1132GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1132GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1132GISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1132GISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1132GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1132GISEL-NEXT: ; %bb.3: ; %if
; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1132GISEL-NEXT: s_mov_b32 s1, 0
-; GFX1132GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1132GISEL-NEXT: s_add_i32 s1, s1, s6
; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1132GISEL-NEXT: ; %bb.5:
; GFX1132GISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1132GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1132GISEL-NEXT: .LBB8_6: ; %endif
; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2800,7 +3248,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX12DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX12DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX12DAGISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX12DAGISEL-NEXT: ; %bb.1: ; %else
; GFX12DAGISEL-NEXT: s_load_b32 s1, s[4:5], 0x2c
; GFX12DAGISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2809,15 +3257,15 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX12DAGISEL-NEXT: s_bcnt1_i32_b32 s2, s2
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
; GFX12DAGISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX12DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX12DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX12DAGISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX12DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX12DAGISEL-NEXT: ; %bb.3: ; %if
; GFX12DAGISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX12DAGISEL-NEXT: s_mov_b32 s1, 0
-; GFX12DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: v_readlane_b32 s6, v0, s3
@@ -2825,10 +3273,10 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX12DAGISEL-NEXT: s_add_co_i32 s1, s1, s6
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX12DAGISEL-NEXT: ; %bb.5:
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX12DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX12DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX12DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX12DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -3076,7 +3524,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX8DAGISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3084,7 +3532,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX8DAGISEL-NEXT: s_addc_u32 s5, s5, s10
; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX8DAGISEL-NEXT: ; %bb.2:
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3097,7 +3545,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX8GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX8GISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX8GISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3105,7 +3553,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX8GISEL-NEXT: s_addc_u32 s5, s5, s10
; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX8GISEL-NEXT: ; %bb.2:
; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3118,7 +3566,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX9DAGISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3126,7 +3574,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX9DAGISEL-NEXT: s_addc_u32 s5, s5, s10
; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9DAGISEL-NEXT: ; %bb.2:
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3139,7 +3587,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX9GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX9GISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX9GISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3147,7 +3595,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX9GISEL-NEXT: s_addc_u32 s5, s5, s10
; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9GISEL-NEXT: ; %bb.2:
; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3160,7 +3608,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX1064DAGISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3168,7 +3616,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064DAGISEL-NEXT: s_add_u32 s4, s4, s9
; GFX1064DAGISEL-NEXT: s_addc_u32 s5, s5, s10
; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064DAGISEL-NEXT: ; %bb.2:
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3180,7 +3628,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1064GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX1064GISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX1064GISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3188,7 +3636,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064GISEL-NEXT: s_add_u32 s4, s4, s9
; GFX1064GISEL-NEXT: s_addc_u32 s5, s5, s10
; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064GISEL-NEXT: ; %bb.2:
; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3200,7 +3648,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1032DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo
-; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6
; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7
@@ -3208,7 +3656,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032DAGISEL-NEXT: s_add_u32 s4, s4, s8
; GFX1032DAGISEL-NEXT: s_addc_u32 s5, s5, s9
; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0
-; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032DAGISEL-NEXT: ; %bb.2:
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3220,7 +3668,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1032GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo
-; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6
; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7
; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7
@@ -3228,7 +3676,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032GISEL-NEXT: s_add_u32 s4, s4, s8
; GFX1032GISEL-NEXT: s_addc_u32 s5, s5, s9
; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0
-; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032GISEL-NEXT: ; %bb.2:
; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3240,7 +3688,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1164DAGISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
-; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s4, s[2:3]
; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v2, s4
@@ -3249,7 +3697,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164DAGISEL-NEXT: s_add_u32 s0, s0, s5
; GFX1164DAGISEL-NEXT: s_addc_u32 s1, s1, s6
; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164DAGISEL-NEXT: ; %bb.2:
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0
@@ -3261,7 +3709,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1164GISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
-; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164GISEL-NEXT: s_ctz_i32_b64 s4, s[2:3]
; GFX1164GISEL-NEXT: v_readlane_b32 s5, v2, s4
@@ -3270,7 +3718,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164GISEL-NEXT: s_add_u32 s0, s0, s5
; GFX1164GISEL-NEXT: s_addc_u32 s1, s1, s6
; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164GISEL-NEXT: ; %bb.2:
; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1
; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0
@@ -3282,7 +3730,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1132DAGISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo
-; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3
@@ -3291,7 +3739,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132DAGISEL-NEXT: s_add_u32 s0, s0, s4
; GFX1132DAGISEL-NEXT: s_addc_u32 s1, s1, s5
; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132DAGISEL-NEXT: ; %bb.2:
; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0
; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off
@@ -3302,7 +3750,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1132GISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo
-; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3
@@ -3311,7 +3759,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132GISEL-NEXT: s_add_u32 s0, s0, s4
; GFX1132GISEL-NEXT: s_addc_u32 s1, s1, s5
; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132GISEL-NEXT: ; %bb.2:
; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0
; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off
@@ -3326,7 +3774,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
; GFX12DAGISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX12DAGISEL-NEXT: s_mov_b32 s2, exec_lo
-; GFX12DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
@@ -3336,7 +3784,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX12DAGISEL-NEXT: s_add_nc_u64 s[0:1], s[0:1], s[4:5]
-; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX12DAGISEL-NEXT: ; %bb.2:
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0
@@ -3357,7 +3805,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX8DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX8DAGISEL-NEXT: ; %bb.1: ; %else
; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7]
@@ -3366,7 +3814,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s7
; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, s7
; GFX8DAGISEL-NEXT: s_add_u32 s7, s2, s3
-; GFX8DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX8DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6
@@ -3395,7 +3843,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX8GISEL-NEXT: ; %bb.1: ; %else
; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7]
@@ -3404,13 +3852,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8GISEL-NEXT: s_mul_hi_u32 s2, s2, s7
; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s7
; GFX8GISEL-NEXT: s_add_u32 s7, s2, s3
-; GFX8GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX8GISEL-NEXT: .LBB11_2: ; %Flow
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX8GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX8GISEL-NEXT: ; %bb.3: ; %if
; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34
; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec
@@ -3422,7 +3870,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8GISEL-NEXT: s_add_u32 s7, s4, s5
; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7
-; GFX8GISEL-NEXT: .LBB9_4: ; %endif
+; GFX8GISEL-NEXT: .LBB11_4: ; %endif
; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1
; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0
@@ -3437,7 +3885,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5
; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX9DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX9DAGISEL-NEXT: ; %bb.1: ; %else
; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec
; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s5, s[4:5]
@@ -3446,7 +3894,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9DAGISEL-NEXT: s_mul_hi_u32 s2, s2, s5
; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5
; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s3
-; GFX9DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX9DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4
@@ -3474,7 +3922,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX9GISEL-NEXT: ; %bb.1: ; %else
; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s7, s[6:7]
@@ -3483,13 +3931,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9GISEL-NEXT: s_mul_hi_u32 s2, s2, s7
; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s7
; GFX9GISEL-NEXT: s_add_u32 s7, s2, s3
-; GFX9GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX9GISEL-NEXT: .LBB11_2: ; %Flow
; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX9GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX9GISEL-NEXT: ; %bb.3: ; %if
; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec
@@ -3501,7 +3949,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9GISEL-NEXT: s_add_u32 s5, s6, s5
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX9GISEL-NEXT: .LBB9_4: ; %endif
+; GFX9GISEL-NEXT: .LBB11_4: ; %endif
; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3516,7 +3964,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9
; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc
; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5]
-; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1064DAGISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9]
@@ -3525,7 +3973,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064DAGISEL-NEXT: s_mul_i32 s3, s3, s8
; GFX1064DAGISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1064DAGISEL-NEXT: s_add_u32 s9, s9, s3
-; GFX1064DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1064DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[4:5]
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s8
@@ -3553,7 +4001,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064GISEL-NEXT: ; implicit-def: $sgpr8_sgpr9
; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc
; GFX1064GISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1064GISEL-NEXT: ; %bb.1: ; %else
; GFX1064GISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9]
@@ -3562,13 +4010,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064GISEL-NEXT: s_mul_i32 s3, s3, s8
; GFX1064GISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1064GISEL-NEXT: s_add_u32 s9, s9, s3
-; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1064GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7]
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s8
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s9
; GFX1064GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1064GISEL-NEXT: ; %bb.3: ; %if
; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec
@@ -3580,7 +4028,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064GISEL-NEXT: s_add_u32 s5, s5, s7
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX1064GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1064GISEL-NEXT: .LBB11_4: ; %endif
; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3595,7 +4043,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5
; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo
; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4
@@ -3604,7 +4052,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032DAGISEL-NEXT: s_mul_i32 s3, s3, s4
; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s2, s4
; GFX1032DAGISEL-NEXT: s_add_u32 s5, s5, s3
-; GFX1032DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1032DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4
@@ -3632,7 +4080,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1032GISEL-NEXT: ; %bb.1: ; %else
; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s6, s6
@@ -3641,13 +4089,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032GISEL-NEXT: s_mul_i32 s3, s3, s6
; GFX1032GISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX1032GISEL-NEXT: s_add_u32 s7, s7, s3
-; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1032GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032GISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX1032GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1032GISEL-NEXT: ; %bb.3: ; %if
; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo
@@ -3659,7 +4107,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032GISEL-NEXT: s_add_u32 s5, s5, s7
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX1032GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1032GISEL-NEXT: .LBB11_4: ; %endif
; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2
; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3676,7 +4124,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7]
-; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1164DAGISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
@@ -3686,7 +4134,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164DAGISEL-NEXT: s_mul_i32 s3, s3, s8
; GFX1164DAGISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1164DAGISEL-NEXT: s_add_u32 s9, s9, s3
-; GFX1164DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1164DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7]
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s8
@@ -3718,7 +4166,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1164GISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1164GISEL-NEXT: ; %bb.1: ; %else
; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
@@ -3728,13 +4176,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164GISEL-NEXT: s_mul_i32 s3, s3, s8
; GFX1164GISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1164GISEL-NEXT: s_add_u32 s9, s9, s3
-; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1164GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7]
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s8
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s9
; GFX1164GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1164GISEL-NEXT: ; %bb.3: ; %if
; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34
; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec
@@ -3747,7 +4195,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164GISEL-NEXT: s_add_u32 s5, s7, s5
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX1164GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1164GISEL-NEXT: .LBB11_4: ; %endif
; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1]
@@ -3764,7 +4212,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1132DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
@@ -3774,7 +4222,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132DAGISEL-NEXT: s_mul_i32 s3, s3, s6
; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX1132DAGISEL-NEXT: s_add_u32 s7, s7, s3
-; GFX1132DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1132DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7
@@ -3804,7 +4252,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1132GISEL-NEXT: ; %bb.1: ; %else
; GFX1132GISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
@@ -3814,12 +4262,12 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132GISEL-NEXT: s_mul_i32 s3, s3, s6
; GFX1132GISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX1132GISEL-NEXT: s_add_u32 s7, s7, s3
-; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1132GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132GISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7
; GFX1132GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1132GISEL-NEXT: ; %bb.3: ; %if
; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34
; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo
@@ -3832,7 +4280,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132GISEL-NEXT: s_add_u32 s5, s6, s5
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5
-; GFX1132GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1132GISEL-NEXT: .LBB11_4: ; %endif
; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2
; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1]
@@ -3849,7 +4297,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX12DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX12DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX12DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX12DAGISEL-NEXT: ; %bb.1: ; %else
; GFX12DAGISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX12DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
@@ -3859,7 +4307,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX12DAGISEL-NEXT: s_mul_i32 s3, s3, s6
; GFX12DAGISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX12DAGISEL-NEXT: s_add_co_u32 s7, s7, s3
-; GFX12DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX12DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
; GFX12DAGISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX12DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7
@@ -3898,3 +4346,9 @@ endif:
store i64 %combine, ptr addrspace(1) %out
ret void
}
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; GFX1132DAGISEL-FAKE16: {{.*}}
+; GFX1132GISEL-FAKE16: {{.*}}
+; GFX1164DAGISEL-FAKE16: {{.*}}
+; GFX1164GISEL-FAKE16: {{.*}}
+; GFX12DAGISEL-FAKE16: {{.*}}
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll
index 71b33f9ddffa2..01c212dcaee07 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.sub.ll
@@ -7,11 +7,476 @@
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=1 -new-reg-bank-select -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1064GISEL %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=0 < %s | FileCheck -check-prefixes=GFX1032DAGISEL %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GFX1032GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164DAGISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 < %s | FileCheck -check-prefixes=GFX1132DAGISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GFX1132GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel=0 < %s | FileCheck -check-prefixes=GFX12DAGISEL %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=0 -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164DAGISEL,GFX1164DAGISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX1164GISEL,GFX1164GISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=0 < %s | FileCheck -check-prefixes=GFX1132DAGISEL,GFX1132DAGISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1100 -global-isel=1 -new-reg-bank-select < %s | FileCheck -check-prefixes=GFX1132GISEL,GFX1132GISEL-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mattr=-real-true16 -mcpu=gfx1200 -global-isel=0 < %s | FileCheck -check-prefixes=GFX12DAGISEL,GFX12DAGISEL-FAKE16 %s
+
+define amdgpu_kernel void @uniform_value_i16(ptr addrspace(1) %out, i16 %in) {
+; GFX8DAGISEL-LABEL: uniform_value_i16:
+; GFX8DAGISEL: ; %bb.0: ; %entry
+; GFX8DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8DAGISEL-NEXT: s_sub_i32 s3, 0, s6
+; GFX8DAGISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s1
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX8DAGISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8DAGISEL-NEXT: s_endpgm
+;
+; GFX8GISEL-LABEL: uniform_value_i16:
+; GFX8GISEL: ; %bb.0: ; %entry
+; GFX8GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX8GISEL-NEXT: s_and_b32 s3, 0xffff, s6
+; GFX8GISEL-NEXT: s_sub_i32 s3, 0, s3
+; GFX8GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0
+; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s1
+; GFX8GISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8GISEL-NEXT: s_endpgm
+;
+; GFX9DAGISEL-LABEL: uniform_value_i16:
+; GFX9DAGISEL: ; %bb.0: ; %entry
+; GFX9DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9DAGISEL-NEXT: s_sub_i32 s3, 0, s6
+; GFX9DAGISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX9DAGISEL-NEXT: global_store_short v0, v1, s[0:1]
+; GFX9DAGISEL-NEXT: s_endpgm
+;
+; GFX9GISEL-LABEL: uniform_value_i16:
+; GFX9GISEL: ; %bb.0: ; %entry
+; GFX9GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX9GISEL-NEXT: s_and_b32 s3, 0xffff, s6
+; GFX9GISEL-NEXT: s_sub_i32 s3, 0, s3
+; GFX9GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX9GISEL-NEXT: global_store_short v1, v0, s[0:1]
+; GFX9GISEL-NEXT: s_endpgm
+;
+; GFX1064DAGISEL-LABEL: uniform_value_i16:
+; GFX1064DAGISEL: ; %bb.0: ; %entry
+; GFX1064DAGISEL-NEXT: s_clause 0x1
+; GFX1064DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1064DAGISEL-NEXT: s_sub_i32 s3, 0, s6
+; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX1064DAGISEL-NEXT: global_store_short v0, v1, s[0:1]
+; GFX1064DAGISEL-NEXT: s_endpgm
+;
+; GFX1064GISEL-LABEL: uniform_value_i16:
+; GFX1064GISEL: ; %bb.0: ; %entry
+; GFX1064GISEL-NEXT: s_clause 0x1
+; GFX1064GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
+; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1064GISEL-NEXT: s_and_b32 s4, 0xffff, s6
+; GFX1064GISEL-NEXT: s_sub_i32 s3, 0, s4
+; GFX1064GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1064GISEL-NEXT: global_store_short v1, v0, s[0:1]
+; GFX1064GISEL-NEXT: s_endpgm
+;
+; GFX1032DAGISEL-LABEL: uniform_value_i16:
+; GFX1032DAGISEL: ; %bb.0: ; %entry
+; GFX1032DAGISEL-NEXT: s_clause 0x1
+; GFX1032DAGISEL-NEXT: s_load_dword s2, s[4:5], 0x2c
+; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1032DAGISEL-NEXT: s_sub_i32 s2, 0, s2
+; GFX1032DAGISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX1032DAGISEL-NEXT: global_store_short v0, v1, s[0:1]
+; GFX1032DAGISEL-NEXT: s_endpgm
+;
+; GFX1032GISEL-LABEL: uniform_value_i16:
+; GFX1032GISEL: ; %bb.0: ; %entry
+; GFX1032GISEL-NEXT: s_clause 0x1
+; GFX1032GISEL-NEXT: s_load_dword s2, s[4:5], 0x2c
+; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
+; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1032GISEL-NEXT: s_and_b32 s2, 0xffff, s2
+; GFX1032GISEL-NEXT: s_sub_i32 s2, 0, s2
+; GFX1032GISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1032GISEL-NEXT: global_store_short v1, v0, s[0:1]
+; GFX1032GISEL-NEXT: s_endpgm
+;
+; GFX1164DAGISEL-LABEL: uniform_value_i16:
+; GFX1164DAGISEL: ; %bb.0: ; %entry
+; GFX1164DAGISEL-NEXT: s_clause 0x1
+; GFX1164DAGISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
+; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0
+; GFX1164DAGISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1164DAGISEL-NEXT: s_sub_i32 s3, 0, s6
+; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s2
+; GFX1164DAGISEL-NEXT: global_store_b16 v0, v1, s[0:1]
+; GFX1164DAGISEL-NEXT: s_endpgm
+;
+; GFX1164GISEL-LABEL: uniform_value_i16:
+; GFX1164GISEL: ; %bb.0: ; %entry
+; GFX1164GISEL-NEXT: s_clause 0x1
+; GFX1164GISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
+; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
+; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1164GISEL-NEXT: s_bcnt1_i32_b64 s2, s[2:3]
+; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1164GISEL-NEXT: s_and_b32 s4, 0xffff, s6
+; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164GISEL-NEXT: s_sub_i32 s3, 0, s4
+; GFX1164GISEL-NEXT: s_mul_i32 s2, s3, s2
+; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1164GISEL-NEXT: global_store_b16 v1, v0, s[0:1]
+; GFX1164GISEL-NEXT: s_endpgm
+;
+; GFX1132DAGISEL-LABEL: uniform_value_i16:
+; GFX1132DAGISEL: ; %bb.0: ; %entry
+; GFX1132DAGISEL-NEXT: s_clause 0x1
+; GFX1132DAGISEL-NEXT: s_load_b32 s2, s[4:5], 0x2c
+; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX1132DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1132DAGISEL-NEXT: s_sub_i32 s2, 0, s2
+; GFX1132DAGISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s2
+; GFX1132DAGISEL-NEXT: global_store_b16 v0, v1, s[0:1]
+; GFX1132DAGISEL-NEXT: s_endpgm
+;
+; GFX1132GISEL-LABEL: uniform_value_i16:
+; GFX1132GISEL: ; %bb.0: ; %entry
+; GFX1132GISEL-NEXT: s_clause 0x1
+; GFX1132GISEL-NEXT: s_load_b32 s2, s[4:5], 0x2c
+; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX1132GISEL-NEXT: v_mov_b32_e32 v1, 0
+; GFX1132GISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
+; GFX1132GISEL-NEXT: s_and_b32 s2, 0xffff, s2
+; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132GISEL-NEXT: s_sub_i32 s2, 0, s2
+; GFX1132GISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, s2
+; GFX1132GISEL-NEXT: global_store_b16 v1, v0, s[0:1]
+; GFX1132GISEL-NEXT: s_endpgm
+;
+; GFX12DAGISEL-LABEL: uniform_value_i16:
+; GFX12DAGISEL: ; %bb.0: ; %entry
+; GFX12DAGISEL-NEXT: s_load_b96 s[0:2], s[4:5], 0x24
+; GFX12DAGISEL-NEXT: s_mov_b32 s3, exec_lo
+; GFX12DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX12DAGISEL-NEXT: s_bcnt1_i32_b32 s3, s3
+; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
+; GFX12DAGISEL-NEXT: s_sub_co_i32 s2, 0, s2
+; GFX12DAGISEL-NEXT: s_mul_i32 s2, s2, s3
+; GFX12DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
+; GFX12DAGISEL-NEXT: v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, s2
+; GFX12DAGISEL-NEXT: global_store_b16 v0, v1, s[0:1]
+; GFX12DAGISEL-NEXT: s_endpgm
+entry:
+ %result = call i16 @llvm.amdgcn.wave.reduce.sub.i16(i16 %in, i32 1)
+ store i16 %result, ptr addrspace(1) %out
+ ret void
+}
+
+define void @divergent_value_i16(ptr addrspace(1) %out, i16 %in) {
+; GFX8DAGISEL-LABEL: divergent_value_i16:
+; GFX8DAGISEL: ; %bb.0: ; %entry
+; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX8DAGISEL-NEXT: s_mov_b32 s6, 0
+; GFX8DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX8DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX8DAGISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX8DAGISEL-NEXT: s_sub_i32 s6, s6, s8
+; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8DAGISEL-NEXT: ; %bb.2:
+; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX8DAGISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX8DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX8GISEL-LABEL: divergent_value_i16:
+; GFX8GISEL: ; %bb.0: ; %entry
+; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8GISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX8GISEL-NEXT: s_mov_b32 s6, 0
+; GFX8GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX8GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX8GISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX8GISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX8GISEL-NEXT: s_sub_i32 s6, s6, s8
+; GFX8GISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8GISEL-NEXT: ; %bb.2:
+; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX8GISEL-NEXT: flat_store_short v[0:1], v2
+; GFX8GISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX8GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX9DAGISEL-LABEL: divergent_value_i16:
+; GFX9DAGISEL: ; %bb.0: ; %entry
+; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX9DAGISEL-NEXT: s_mov_b32 s6, 0
+; GFX9DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX9DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX9DAGISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX9DAGISEL-NEXT: s_sub_i32 s6, s6, s8
+; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9DAGISEL-NEXT: ; %bb.2:
+; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX9DAGISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX9DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX9GISEL-LABEL: divergent_value_i16:
+; GFX9GISEL: ; %bb.0: ; %entry
+; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX9GISEL-NEXT: s_mov_b32 s6, 0
+; GFX9GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX9GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX9GISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX9GISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX9GISEL-NEXT: s_sub_i32 s6, s6, s8
+; GFX9GISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9GISEL-NEXT: ; %bb.2:
+; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX9GISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX9GISEL-NEXT: s_waitcnt vmcnt(0)
+; GFX9GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1064DAGISEL-LABEL: divergent_value_i16:
+; GFX1064DAGISEL: ; %bb.0: ; %entry
+; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1064DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX1064DAGISEL-NEXT: s_mov_b32 s6, 0
+; GFX1064DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX1064DAGISEL-NEXT: s_sub_i32 s6, s6, s8
+; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064DAGISEL-NEXT: ; %bb.2:
+; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX1064DAGISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1064DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1064GISEL-LABEL: divergent_value_i16:
+; GFX1064GISEL: ; %bb.0: ; %entry
+; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1064GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec
+; GFX1064GISEL-NEXT: s_mov_b32 s6, 0
+; GFX1064GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: s_ff1_i32_b64 s7, s[4:5]
+; GFX1064GISEL-NEXT: v_readlane_b32 s8, v2, s7
+; GFX1064GISEL-NEXT: s_bitset0_b64 s[4:5], s7
+; GFX1064GISEL-NEXT: s_sub_i32 s6, s6, s8
+; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[4:5], 0
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064GISEL-NEXT: ; %bb.2:
+; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s6
+; GFX1064GISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1064GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1032DAGISEL-LABEL: divergent_value_i16:
+; GFX1032DAGISEL: ; %bb.0: ; %entry
+; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1032DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1032DAGISEL-NEXT: s_mov_b32 s5, exec_lo
+; GFX1032DAGISEL-NEXT: s_mov_b32 s4, 0
+; GFX1032DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s6, s5
+; GFX1032DAGISEL-NEXT: v_readlane_b32 s7, v2, s6
+; GFX1032DAGISEL-NEXT: s_bitset0_b32 s5, s6
+; GFX1032DAGISEL-NEXT: s_sub_i32 s4, s4, s7
+; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s5, 0
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032DAGISEL-NEXT: ; %bb.2:
+; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4
+; GFX1032DAGISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1032DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1032GISEL-LABEL: divergent_value_i16:
+; GFX1032GISEL: ; %bb.0: ; %entry
+; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1032GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1032GISEL-NEXT: s_mov_b32 s5, exec_lo
+; GFX1032GISEL-NEXT: s_mov_b32 s4, 0
+; GFX1032GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: s_ff1_i32_b32 s6, s5
+; GFX1032GISEL-NEXT: v_readlane_b32 s7, v2, s6
+; GFX1032GISEL-NEXT: s_bitset0_b32 s5, s6
+; GFX1032GISEL-NEXT: s_sub_i32 s4, s4, s7
+; GFX1032GISEL-NEXT: s_cmp_lg_u32 s5, 0
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032GISEL-NEXT: ; %bb.2:
+; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4
+; GFX1032GISEL-NEXT: global_store_short v[0:1], v2, off
+; GFX1032GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1164DAGISEL-LABEL: divergent_value_i16:
+; GFX1164DAGISEL: ; %bb.0: ; %entry
+; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1164DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1164DAGISEL-NEXT: s_mov_b64 s[0:1], exec
+; GFX1164DAGISEL-NEXT: s_mov_b32 s2, 0
+; GFX1164DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s3, s[0:1]
+; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1164DAGISEL-NEXT: v_readlane_b32 s4, v2, s3
+; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[0:1], s3
+; GFX1164DAGISEL-NEXT: s_sub_i32 s2, s2, s4
+; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164DAGISEL-NEXT: ; %bb.2:
+; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX1164DAGISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1164DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1164GISEL-LABEL: divergent_value_i16:
+; GFX1164GISEL: ; %bb.0: ; %entry
+; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1164GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1164GISEL-NEXT: s_mov_b64 s[0:1], exec
+; GFX1164GISEL-NEXT: s_mov_b32 s2, 0
+; GFX1164GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: s_ctz_i32_b64 s3, s[0:1]
+; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1164GISEL-NEXT: v_readlane_b32 s4, v2, s3
+; GFX1164GISEL-NEXT: s_bitset0_b64 s[0:1], s3
+; GFX1164GISEL-NEXT: s_sub_i32 s2, s2, s4
+; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164GISEL-NEXT: ; %bb.2:
+; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s2
+; GFX1164GISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1164GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1132DAGISEL-LABEL: divergent_value_i16:
+; GFX1132DAGISEL: ; %bb.0: ; %entry
+; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1132DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1132DAGISEL-NEXT: s_mov_b32 s1, exec_lo
+; GFX1132DAGISEL-NEXT: s_mov_b32 s0, 0
+; GFX1132DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s2, s1
+; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1132DAGISEL-NEXT: v_readlane_b32 s3, v2, s2
+; GFX1132DAGISEL-NEXT: s_bitset0_b32 s1, s2
+; GFX1132DAGISEL-NEXT: s_sub_i32 s0, s0, s3
+; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132DAGISEL-NEXT: ; %bb.2:
+; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v2, s0
+; GFX1132DAGISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1132DAGISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX1132GISEL-LABEL: divergent_value_i16:
+; GFX1132GISEL: ; %bb.0: ; %entry
+; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX1132GISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX1132GISEL-NEXT: s_mov_b32 s1, exec_lo
+; GFX1132GISEL-NEXT: s_mov_b32 s0, 0
+; GFX1132GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: s_ctz_i32_b32 s2, s1
+; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
+; GFX1132GISEL-NEXT: v_readlane_b32 s3, v2, s2
+; GFX1132GISEL-NEXT: s_bitset0_b32 s1, s2
+; GFX1132GISEL-NEXT: s_sub_i32 s0, s0, s3
+; GFX1132GISEL-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132GISEL-NEXT: ; %bb.2:
+; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, s0
+; GFX1132GISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX1132GISEL-NEXT: s_setpc_b64 s[30:31]
+;
+; GFX12DAGISEL-LABEL: divergent_value_i16:
+; GFX12DAGISEL: ; %bb.0: ; %entry
+; GFX12DAGISEL-NEXT: s_wait_loadcnt_dscnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_expcnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_samplecnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_bvhcnt 0x0
+; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
+; GFX12DAGISEL-NEXT: v_bfe_i32 v2, v2, 0, 16
+; GFX12DAGISEL-NEXT: s_mov_b32 s1, exec_lo
+; GFX12DAGISEL-NEXT: s_mov_b32 s0, 0
+; GFX12DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
+; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s2, s1
+; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
+; GFX12DAGISEL-NEXT: v_readlane_b32 s3, v2, s2
+; GFX12DAGISEL-NEXT: s_bitset0_b32 s1, s2
+; GFX12DAGISEL-NEXT: s_sub_co_i32 s0, s0, s3
+; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
+; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX12DAGISEL-NEXT: ; %bb.2:
+; GFX12DAGISEL-NEXT: v_mov_b32_e32 v2, s0
+; GFX12DAGISEL-NEXT: global_store_b16 v[0:1], v2, off
+; GFX12DAGISEL-NEXT: s_setpc_b64 s[30:31]
+entry:
+ %result = call i16 @llvm.amdgcn.wave.reduce.sub.i16(i16 %in, i32 1)
+ store i16 %result, ptr addrspace(1) %out
+ ret void
+}
define amdgpu_kernel void @uniform_value(ptr addrspace(1) %out, i32 %in) {
; GFX8DAGISEL-LABEL: uniform_value:
@@ -221,13 +686,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX8DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX8DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX8DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX8DAGISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX8DAGISEL-NEXT: ; %bb.2:
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s0
@@ -241,13 +706,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8GISEL-NEXT: s_mov_b32 s4, 0
-; GFX8GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX8GISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX8GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX8GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX8GISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX8GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX8GISEL-NEXT: ; %bb.2:
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s0
@@ -262,13 +727,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, 0
; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX9DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX9DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX9DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX9DAGISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9DAGISEL-NEXT: ; %bb.2:
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -280,13 +745,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9GISEL-NEXT: s_mov_b32 s4, 0
-; GFX9GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX9GISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX9GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX9GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX9GISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX9GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9GISEL-NEXT: ; %bb.2:
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -300,13 +765,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, 0
; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX1064DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX1064DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1064DAGISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064DAGISEL-NEXT: ; %bb.2:
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -318,13 +783,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064GISEL-NEXT: s_mov_b32 s4, 0
-; GFX1064GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1064GISEL-NEXT: s_ff1_i32_b64 s5, s[2:3]
; GFX1064GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1064GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1064GISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064GISEL-NEXT: ; %bb.2:
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -338,13 +803,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, 0
; GFX1032DAGISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1032DAGISEL-NEXT: s_mov_b32 s2, 0
-; GFX1032DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s4, s3
; GFX1032DAGISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1032DAGISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1032DAGISEL-NEXT: s_sub_i32 s2, s2, s5
; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032DAGISEL-NEXT: ; %bb.2:
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -356,13 +821,13 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1032GISEL-NEXT: s_mov_b32 s2, 0
-; GFX1032GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1032GISEL-NEXT: s_ff1_i32_b32 s4, s3
; GFX1032GISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1032GISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1032GISEL-NEXT: s_sub_i32 s2, s2, s5
; GFX1032GISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032GISEL-NEXT: ; %bb.2:
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -377,14 +842,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1164DAGISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164DAGISEL-NEXT: s_mov_b32 s4, 0
-; GFX1164DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s5, s[2:3]
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164DAGISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1164DAGISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164DAGISEL-NEXT: ; %bb.2:
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -397,14 +862,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1164GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164GISEL-NEXT: s_mov_b32 s4, 0
-; GFX1164GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1164GISEL-NEXT: s_ctz_i32_b64 s5, s[2:3]
; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164GISEL-NEXT: v_readlane_b32 s6, v0, s5
; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s5
; GFX1164GISEL-NEXT: s_sub_i32 s4, s4, s6
; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164GISEL-NEXT: ; %bb.2:
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, 0
@@ -418,14 +883,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX1132DAGISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, 0
-; GFX1132DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s4, s3
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132DAGISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1132DAGISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1132DAGISEL-NEXT: s_sub_i32 s2, s2, s5
; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132DAGISEL-NEXT: ; %bb.2:
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -438,14 +903,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX1132GISEL-NEXT: v_and_b32_e32 v0, 0x3ff, v0
; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX1132GISEL-NEXT: s_mov_b32 s2, 0
-; GFX1132GISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX1132GISEL-NEXT: s_ctz_i32_b32 s4, s3
; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX1132GISEL-NEXT: s_bitset0_b32 s3, s4
; GFX1132GISEL-NEXT: s_sub_i32 s2, s2, s5
; GFX1132GISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132GISEL-NEXT: ; %bb.2:
; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, 0
; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -458,14 +923,14 @@ define amdgpu_kernel void @divergent_value(ptr addrspace(1) %out) {
; GFX12DAGISEL-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX12DAGISEL-NEXT: s_mov_b32 s3, exec_lo
; GFX12DAGISEL-NEXT: s_mov_b32 s2, 0
-; GFX12DAGISEL-NEXT: .LBB1_1: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: .LBB3_1: ; =>This Inner Loop Header: Depth=1
; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s4, s3
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: v_readlane_b32 s5, v0, s4
; GFX12DAGISEL-NEXT: s_bitset0_b32 s3, s4
; GFX12DAGISEL-NEXT: s_sub_co_i32 s2, s2, s5
; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s3, 0
-; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB1_1
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB3_1
; GFX12DAGISEL-NEXT: ; %bb.2:
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v0, s2
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
@@ -2402,7 +2867,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr2
; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX8DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX8DAGISEL-NEXT: ; %bb.1: ; %else
; GFX8DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2411,24 +2876,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8DAGISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX8DAGISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX8DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX8DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX8DAGISEL-NEXT: ; %bb.3: ; %if
; GFX8DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX8DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX8DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX8DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX8DAGISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX8DAGISEL-NEXT: ; %bb.5:
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX8DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX8DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX8DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -2443,7 +2908,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8GISEL-NEXT: ; implicit-def: $sgpr2
; GFX8GISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX8GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX8GISEL-NEXT: ; %bb.1: ; %else
; GFX8GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2452,24 +2917,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8GISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX8GISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX8GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX8GISEL-NEXT: .LBB8_2: ; %Flow
; GFX8GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX8GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX8GISEL-NEXT: ; %bb.3: ; %if
; GFX8GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX8GISEL-NEXT: s_mov_b32 s6, 0
-; GFX8GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX8GISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX8GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX8GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX8GISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX8GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX8GISEL-NEXT: ; %bb.5:
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX8GISEL-NEXT: .LBB6_6: ; %endif
+; GFX8GISEL-NEXT: .LBB8_6: ; %endif
; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX8GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
@@ -2484,7 +2949,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr2
; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX9DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX9DAGISEL-NEXT: ; %bb.1: ; %else
; GFX9DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2493,24 +2958,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9DAGISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX9DAGISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX9DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX9DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX9DAGISEL-NEXT: ; %bb.3: ; %if
; GFX9DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX9DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX9DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX9DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX9DAGISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX9DAGISEL-NEXT: ; %bb.5:
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX9DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX9DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX9DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2524,7 +2989,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9GISEL-NEXT: ; implicit-def: $sgpr2
; GFX9GISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX9GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX9GISEL-NEXT: ; %bb.1: ; %else
; GFX9GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2533,24 +2998,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9GISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX9GISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX9GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX9GISEL-NEXT: .LBB8_2: ; %Flow
; GFX9GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX9GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX9GISEL-NEXT: ; %bb.3: ; %if
; GFX9GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX9GISEL-NEXT: s_mov_b32 s6, 0
-; GFX9GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX9GISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX9GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX9GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX9GISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX9GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX9GISEL-NEXT: ; %bb.5:
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX9GISEL-NEXT: .LBB6_6: ; %endif
+; GFX9GISEL-NEXT: .LBB8_6: ; %endif
; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX9GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2564,7 +3029,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr2
; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX1064DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1064DAGISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2573,24 +3038,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064DAGISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX1064DAGISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX1064DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1064DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1064DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1064DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1064DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX1064DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX1064DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1064DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1064DAGISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1064DAGISEL-NEXT: ; %bb.5:
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1064DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1064DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1064DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1064DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2604,7 +3069,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064GISEL-NEXT: ; implicit-def: $sgpr2
; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[0:1], vcc
; GFX1064GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1064GISEL-NEXT: ; %bb.1: ; %else
; GFX1064GISEL-NEXT: s_load_dword s6, s[4:5], 0x2c
; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2613,24 +3078,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064GISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX1064GISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX1064GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1064GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1064GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1064GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1064GISEL-NEXT: ; %bb.3: ; %if
; GFX1064GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1064GISEL-NEXT: s_mov_b32 s6, 0
-; GFX1064GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1064GISEL-NEXT: s_ff1_i32_b64 s7, s[2:3]
; GFX1064GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1064GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1064GISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1064GISEL-NEXT: ; %bb.5:
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1064GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1064GISEL-NEXT: .LBB8_6: ; %endif
; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1064GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2644,7 +3109,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr1
; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo
; GFX1032DAGISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1032DAGISEL-NEXT: s_load_dword s1, s[4:5], 0x2c
; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2653,24 +3118,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032DAGISEL-NEXT: s_sub_i32 s1, 0, s1
; GFX1032DAGISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1032DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1032DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1032DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1032DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1032DAGISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1032DAGISEL-NEXT: s_mov_b32 s1, 0
-; GFX1032DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032DAGISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1032DAGISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1032DAGISEL-NEXT: s_sub_i32 s1, s1, s6
; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1032DAGISEL-NEXT: ; %bb.5:
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1032DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1032DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1032DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1032DAGISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2684,7 +3149,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032GISEL-NEXT: ; implicit-def: $sgpr1
; GFX1032GISEL-NEXT: s_and_saveexec_b32 s0, vcc_lo
; GFX1032GISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1032GISEL-NEXT: ; %bb.1: ; %else
; GFX1032GISEL-NEXT: s_load_dword s1, s[4:5], 0x2c
; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2693,24 +3158,24 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032GISEL-NEXT: s_sub_i32 s1, 0, s1
; GFX1032GISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1032GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1032GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1032GISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1032GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1032GISEL-NEXT: ; %bb.3: ; %if
; GFX1032GISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1032GISEL-NEXT: s_mov_b32 s1, 0
-; GFX1032GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1032GISEL-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032GISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1032GISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1032GISEL-NEXT: s_sub_i32 s1, s1, s6
; GFX1032GISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1032GISEL-NEXT: ; %bb.5:
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1032GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1032GISEL-NEXT: .LBB8_6: ; %endif
; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1032GISEL-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2726,7 +3191,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1164DAGISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1164DAGISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2736,25 +3201,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164DAGISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX1164DAGISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX1164DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1164DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1164DAGISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1164DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164DAGISEL-NEXT: s_mov_b32 s6, 0
-; GFX1164DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s7, s[2:3]
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164DAGISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1164DAGISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1164DAGISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1164DAGISEL-NEXT: ; %bb.5:
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1164DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1164DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1164DAGISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1164DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2770,7 +3235,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1164GISEL-NEXT: s_xor_b64 s[0:1], exec, s[0:1]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1164GISEL-NEXT: ; %bb.1: ; %else
; GFX1164GISEL-NEXT: s_load_b32 s6, s[4:5], 0x2c
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
@@ -2780,25 +3245,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164GISEL-NEXT: s_sub_i32 s3, 0, s6
; GFX1164GISEL-NEXT: s_mul_i32 s2, s3, s2
-; GFX1164GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1164GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1164GISEL-NEXT: s_or_saveexec_b64 s[0:1], s[0:1]
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s2
; GFX1164GISEL-NEXT: s_xor_b64 exec, exec, s[0:1]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1164GISEL-NEXT: ; %bb.3: ; %if
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
; GFX1164GISEL-NEXT: s_mov_b32 s6, 0
-; GFX1164GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1164GISEL-NEXT: s_ctz_i32_b64 s7, s[2:3]
; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164GISEL-NEXT: v_readlane_b32 s8, v0, s7
; GFX1164GISEL-NEXT: s_bitset0_b64 s[2:3], s7
; GFX1164GISEL-NEXT: s_sub_i32 s6, s6, s8
; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1164GISEL-NEXT: ; %bb.5:
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s6
-; GFX1164GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1164GISEL-NEXT: .LBB8_6: ; %endif
; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[0:1]
; GFX1164GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2814,7 +3279,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1132DAGISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1132DAGISEL-NEXT: s_load_b32 s1, s[4:5], 0x2c
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2824,25 +3289,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132DAGISEL-NEXT: s_sub_i32 s1, 0, s1
; GFX1132DAGISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1132DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1132DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1132DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1132DAGISEL-NEXT: ; %bb.3: ; %if
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1132DAGISEL-NEXT: s_mov_b32 s1, 0
-; GFX1132DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132DAGISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1132DAGISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1132DAGISEL-NEXT: s_sub_i32 s1, s1, s6
; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1132DAGISEL-NEXT: ; %bb.5:
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1132DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX1132DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX1132DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1132DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1132DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2858,7 +3323,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1132GISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX1132GISEL-NEXT: ; %bb.1: ; %else
; GFX1132GISEL-NEXT: s_load_b32 s1, s[4:5], 0x2c
; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2868,25 +3333,25 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132GISEL-NEXT: s_sub_i32 s1, 0, s1
; GFX1132GISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX1132GISEL-NEXT: .LBB6_2: ; %Flow
+; GFX1132GISEL-NEXT: .LBB8_2: ; %Flow
; GFX1132GISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX1132GISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX1132GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX1132GISEL-NEXT: ; %bb.3: ; %if
; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX1132GISEL-NEXT: s_mov_b32 s1, 0
-; GFX1132GISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: v_readlane_b32 s6, v0, s3
; GFX1132GISEL-NEXT: s_bitset0_b32 s2, s3
; GFX1132GISEL-NEXT: s_sub_i32 s1, s1, s6
; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX1132GISEL-NEXT: ; %bb.5:
; GFX1132GISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX1132GISEL-NEXT: .LBB6_6: ; %endif
+; GFX1132GISEL-NEXT: .LBB8_6: ; %endif
; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX1132GISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX1132GISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -2902,7 +3367,7 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX12DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX12DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX12DAGISEL-NEXT: s_xor_b32 s0, exec_lo, s0
-; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB6_2
+; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB8_2
; GFX12DAGISEL-NEXT: ; %bb.1: ; %else
; GFX12DAGISEL-NEXT: s_load_b32 s1, s[4:5], 0x2c
; GFX12DAGISEL-NEXT: s_mov_b32 s2, exec_lo
@@ -2912,15 +3377,15 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
; GFX12DAGISEL-NEXT: s_sub_co_i32 s1, 0, s1
; GFX12DAGISEL-NEXT: s_mul_i32 s1, s1, s2
-; GFX12DAGISEL-NEXT: .LBB6_2: ; %Flow
+; GFX12DAGISEL-NEXT: .LBB8_2: ; %Flow
; GFX12DAGISEL-NEXT: s_or_saveexec_b32 s0, s0
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v1, s1
; GFX12DAGISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s0
-; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB6_6
+; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB8_6
; GFX12DAGISEL-NEXT: ; %bb.3: ; %if
; GFX12DAGISEL-NEXT: s_mov_b32 s2, exec_lo
; GFX12DAGISEL-NEXT: s_mov_b32 s1, 0
-; GFX12DAGISEL-NEXT: .LBB6_4: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: .LBB8_4: ; =>This Inner Loop Header: Depth=1
; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: v_readlane_b32 s6, v0, s3
@@ -2928,10 +3393,10 @@ define amdgpu_kernel void @divergent_cfg(ptr addrspace(1) %out, i32 %in) {
; GFX12DAGISEL-NEXT: s_sub_co_i32 s1, s1, s6
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB6_4
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB8_4
; GFX12DAGISEL-NEXT: ; %bb.5:
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v1, s1
-; GFX12DAGISEL-NEXT: .LBB6_6: ; %endif
+; GFX12DAGISEL-NEXT: .LBB8_6: ; %endif
; GFX12DAGISEL-NEXT: s_or_b32 exec_lo, exec_lo, s0
; GFX12DAGISEL-NEXT: s_load_b64 s[0:1], s[4:5], 0x24
; GFX12DAGISEL-NEXT: v_mov_b32_e32 v0, 0
@@ -3236,7 +3701,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX8DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX8DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX8DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX8DAGISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX8DAGISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3244,7 +3709,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX8DAGISEL-NEXT: s_subb_u32 s5, s5, s10
; GFX8DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX8DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX8DAGISEL-NEXT: ; %bb.2:
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3257,7 +3722,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX8GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX8GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX8GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX8GISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX8GISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3265,7 +3730,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX8GISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX8GISEL-NEXT: s_subb_u32 s5, s5, s10
; GFX8GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX8GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX8GISEL-NEXT: ; %bb.2:
; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3278,7 +3743,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX9DAGISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX9DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX9DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX9DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX9DAGISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX9DAGISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3286,7 +3751,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9DAGISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX9DAGISEL-NEXT: s_subb_u32 s5, s5, s10
; GFX9DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX9DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9DAGISEL-NEXT: ; %bb.2:
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3299,7 +3764,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX9GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX9GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX9GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX9GISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX9GISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3307,7 +3772,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX9GISEL-NEXT: s_bitset0_b64 s[6:7], s8
; GFX9GISEL-NEXT: s_subb_u32 s5, s5, s10
; GFX9GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX9GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9GISEL-NEXT: ; %bb.2:
; GFX9GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX9GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3320,7 +3785,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1064DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1064DAGISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX1064DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1064DAGISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX1064DAGISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX1064DAGISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3328,7 +3793,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064DAGISEL-NEXT: s_sub_u32 s4, s4, s9
; GFX1064DAGISEL-NEXT: s_subb_u32 s5, s5, s10
; GFX1064DAGISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1064DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064DAGISEL-NEXT: ; %bb.2:
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3340,7 +3805,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1064GISEL-NEXT: s_mov_b64 s[6:7], exec
-; GFX1064GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1064GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1064GISEL-NEXT: s_ff1_i32_b64 s8, s[6:7]
; GFX1064GISEL-NEXT: v_readlane_b32 s9, v2, s8
; GFX1064GISEL-NEXT: v_readlane_b32 s10, v3, s8
@@ -3348,7 +3813,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1064GISEL-NEXT: s_sub_u32 s4, s4, s9
; GFX1064GISEL-NEXT: s_subb_u32 s5, s5, s10
; GFX1064GISEL-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1064GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064GISEL-NEXT: ; %bb.2:
; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1064GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3360,7 +3825,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1032DAGISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1032DAGISEL-NEXT: s_mov_b32 s6, exec_lo
-; GFX1032DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1032DAGISEL-NEXT: s_ff1_i32_b32 s7, s6
; GFX1032DAGISEL-NEXT: v_readlane_b32 s8, v2, s7
; GFX1032DAGISEL-NEXT: v_readlane_b32 s9, v3, s7
@@ -3368,7 +3833,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032DAGISEL-NEXT: s_sub_u32 s4, s4, s8
; GFX1032DAGISEL-NEXT: s_subb_u32 s5, s5, s9
; GFX1032DAGISEL-NEXT: s_cmp_lg_u32 s6, 0
-; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1032DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032DAGISEL-NEXT: ; %bb.2:
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3380,7 +3845,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1032GISEL-NEXT: s_mov_b64 s[4:5], 0
; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo
-; GFX1032GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1032GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1032GISEL-NEXT: s_ff1_i32_b32 s7, s6
; GFX1032GISEL-NEXT: v_readlane_b32 s8, v2, s7
; GFX1032GISEL-NEXT: v_readlane_b32 s9, v3, s7
@@ -3388,7 +3853,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1032GISEL-NEXT: s_sub_u32 s4, s4, s8
; GFX1032GISEL-NEXT: s_subb_u32 s5, s5, s9
; GFX1032GISEL-NEXT: s_cmp_lg_u32 s6, 0
-; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1032GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032GISEL-NEXT: ; %bb.2:
; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, s4
; GFX1032GISEL-NEXT: v_mov_b32_e32 v3, s5
@@ -3400,7 +3865,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1164DAGISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1164DAGISEL-NEXT: s_mov_b64 s[2:3], exec
-; GFX1164DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164DAGISEL-NEXT: s_ctz_i32_b64 s4, s[2:3]
; GFX1164DAGISEL-NEXT: v_readlane_b32 s5, v2, s4
@@ -3409,7 +3874,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164DAGISEL-NEXT: s_sub_u32 s0, s0, s5
; GFX1164DAGISEL-NEXT: s_subb_u32 s1, s1, s6
; GFX1164DAGISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1164DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164DAGISEL-NEXT: ; %bb.2:
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v3, s1
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v2, s0
@@ -3421,7 +3886,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1164GISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1164GISEL-NEXT: s_mov_b64 s[2:3], exec
-; GFX1164GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1164GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164GISEL-NEXT: s_ctz_i32_b64 s4, s[2:3]
; GFX1164GISEL-NEXT: v_readlane_b32 s5, v2, s4
@@ -3430,7 +3895,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1164GISEL-NEXT: s_sub_u32 s0, s0, s5
; GFX1164GISEL-NEXT: s_subb_u32 s1, s1, s6
; GFX1164GISEL-NEXT: s_cmp_lg_u64 s[2:3], 0
-; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1164GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164GISEL-NEXT: ; %bb.2:
; GFX1164GISEL-NEXT: v_mov_b32_e32 v3, s1
; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, s0
@@ -3442,7 +3907,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132DAGISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1132DAGISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1132DAGISEL-NEXT: s_mov_b32 s2, exec_lo
-; GFX1132DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1132DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132DAGISEL-NEXT: v_readlane_b32 s4, v2, s3
@@ -3451,7 +3916,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132DAGISEL-NEXT: s_sub_u32 s0, s0, s4
; GFX1132DAGISEL-NEXT: s_subb_u32 s1, s1, s5
; GFX1132DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1132DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132DAGISEL-NEXT: ; %bb.2:
; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0
; GFX1132DAGISEL-NEXT: global_store_b64 v[0:1], v[2:3], off
@@ -3462,7 +3927,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132GISEL-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX1132GISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX1132GISEL-NEXT: s_mov_b32 s2, exec_lo
-; GFX1132GISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX1132GISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX1132GISEL-NEXT: v_readlane_b32 s4, v2, s3
@@ -3471,7 +3936,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX1132GISEL-NEXT: s_sub_u32 s0, s0, s4
; GFX1132GISEL-NEXT: s_subb_u32 s1, s1, s5
; GFX1132GISEL-NEXT: s_cmp_lg_u32 s2, 0
-; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX1132GISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132GISEL-NEXT: ; %bb.2:
; GFX1132GISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0
; GFX1132GISEL-NEXT: global_store_b64 v[0:1], v[2:3], off
@@ -3486,7 +3951,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
; GFX12DAGISEL-NEXT: s_mov_b64 s[0:1], 0
; GFX12DAGISEL-NEXT: s_mov_b32 s2, exec_lo
-; GFX12DAGISEL-NEXT: .LBB8_1: ; =>This Inner Loop Header: Depth=1
+; GFX12DAGISEL-NEXT: .LBB10_1: ; =>This Inner Loop Header: Depth=1
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: s_ctz_i32_b32 s3, s2
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
@@ -3496,7 +3961,7 @@ define void @divergent_value_i64(ptr addrspace(1) %out, i64 %id.x) {
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX12DAGISEL-NEXT: s_sub_nc_u64 s[0:1], s[0:1], s[4:5]
-; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB8_1
+; GFX12DAGISEL-NEXT: s_cbranch_scc1 .LBB10_1
; GFX12DAGISEL-NEXT: ; %bb.2:
; GFX12DAGISEL-NEXT: s_wait_alu depctr_sa_sdst(0)
; GFX12DAGISEL-NEXT: v_dual_mov_b32 v3, s1 :: v_dual_mov_b32 v2, s0
@@ -3517,7 +3982,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8DAGISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX8DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX8DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX8DAGISEL-NEXT: ; %bb.1: ; %else
; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7]
@@ -3530,13 +3995,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8DAGISEL-NEXT: s_mul_i32 s3, s3, s7
; GFX8DAGISEL-NEXT: s_add_u32 s2, s2, s3
; GFX8DAGISEL-NEXT: s_add_u32 s7, s2, s10
-; GFX8DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX8DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX8DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX8DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX8DAGISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX8DAGISEL-NEXT: ; %bb.3: ; %if
; GFX8DAGISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX8DAGISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7]
@@ -3550,7 +4015,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8DAGISEL-NEXT: s_add_u32 s7, s4, s8
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v1, s7
-; GFX8DAGISEL-NEXT: .LBB9_4: ; %endif
+; GFX8DAGISEL-NEXT: .LBB11_4: ; %endif
; GFX8DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v2, s0
; GFX8DAGISEL-NEXT: v_mov_b32_e32 v3, s1
@@ -3564,7 +4029,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX8GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX8GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX8GISEL-NEXT: ; %bb.1: ; %else
; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX8GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7]
@@ -3577,13 +4042,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8GISEL-NEXT: s_mul_i32 s3, s3, s7
; GFX8GISEL-NEXT: s_add_u32 s2, s2, s3
; GFX8GISEL-NEXT: s_add_u32 s7, s2, s10
-; GFX8GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX8GISEL-NEXT: .LBB11_2: ; %Flow
; GFX8GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX8GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX8GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX8GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX8GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX8GISEL-NEXT: ; %bb.3: ; %if
; GFX8GISEL-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x34
; GFX8GISEL-NEXT: s_mov_b64 s[6:7], exec
@@ -3599,7 +4064,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX8GISEL-NEXT: s_add_u32 s7, s4, s8
; GFX8GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX8GISEL-NEXT: v_mov_b32_e32 v1, s7
-; GFX8GISEL-NEXT: .LBB9_4: ; %endif
+; GFX8GISEL-NEXT: .LBB11_4: ; %endif
; GFX8GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX8GISEL-NEXT: v_mov_b32_e32 v3, s1
; GFX8GISEL-NEXT: v_mov_b32_e32 v2, s0
@@ -3614,7 +4079,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5
; GFX9DAGISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX9DAGISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX9DAGISEL-NEXT: ; %bb.1: ; %else
; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec
; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5]
@@ -3627,13 +4092,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9DAGISEL-NEXT: s_mul_i32 s3, s3, s5
; GFX9DAGISEL-NEXT: s_add_u32 s2, s2, s3
; GFX9DAGISEL-NEXT: s_add_u32 s5, s2, s10
-; GFX9DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX9DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX9DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5
; GFX9DAGISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX9DAGISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX9DAGISEL-NEXT: ; %bb.3: ; %if
; GFX9DAGISEL-NEXT: s_mov_b64 s[4:5], exec
; GFX9DAGISEL-NEXT: s_bcnt1_i32_b64 s4, s[4:5]
@@ -3647,7 +4112,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9DAGISEL-NEXT: s_add_u32 s5, s5, s8
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX9DAGISEL-NEXT: .LBB9_4: ; %endif
+; GFX9DAGISEL-NEXT: .LBB11_4: ; %endif
; GFX9DAGISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX9DAGISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX9DAGISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3660,7 +4125,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX9GISEL-NEXT: s_and_saveexec_b64 s[8:9], vcc
; GFX9GISEL-NEXT: s_xor_b64 s[8:9], exec, s[8:9]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX9GISEL-NEXT: ; %bb.1: ; %else
; GFX9GISEL-NEXT: s_mov_b64 s[6:7], exec
; GFX9GISEL-NEXT: s_bcnt1_i32_b64 s6, s[6:7]
@@ -3673,13 +4138,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9GISEL-NEXT: s_mul_i32 s3, s3, s7
; GFX9GISEL-NEXT: s_add_u32 s2, s2, s3
; GFX9GISEL-NEXT: s_add_u32 s7, s2, s10
-; GFX9GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX9GISEL-NEXT: .LBB11_2: ; %Flow
; GFX9GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX9GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[8:9]
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX9GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX9GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX9GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX9GISEL-NEXT: ; %bb.3: ; %if
; GFX9GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX9GISEL-NEXT: s_mov_b64 s[4:5], exec
@@ -3695,7 +4160,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX9GISEL-NEXT: s_add_u32 s5, s5, s8
; GFX9GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX9GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX9GISEL-NEXT: .LBB9_4: ; %endif
+; GFX9GISEL-NEXT: .LBB11_4: ; %endif
; GFX9GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX9GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX9GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3710,7 +4175,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064DAGISEL-NEXT: ; implicit-def: $sgpr8_sgpr9
; GFX1064DAGISEL-NEXT: s_and_saveexec_b64 s[4:5], vcc
; GFX1064DAGISEL-NEXT: s_xor_b64 s[4:5], exec, s[4:5]
-; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1064DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1064DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1064DAGISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1064DAGISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9]
@@ -3723,7 +4188,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064DAGISEL-NEXT: s_add_u32 s3, s10, s3
; GFX1064DAGISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1064DAGISEL-NEXT: s_add_u32 s9, s3, s9
-; GFX1064DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1064DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1064DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[4:5]
; GFX1064DAGISEL-NEXT: v_mov_b32_e32 v0, s8
@@ -3755,7 +4220,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064GISEL-NEXT: ; implicit-def: $sgpr8_sgpr9
; GFX1064GISEL-NEXT: s_and_saveexec_b64 s[6:7], vcc
; GFX1064GISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1064GISEL-NEXT: ; %bb.1: ; %else
; GFX1064GISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1064GISEL-NEXT: s_bcnt1_i32_b64 s8, s[8:9]
@@ -3768,13 +4233,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064GISEL-NEXT: s_add_u32 s3, s10, s3
; GFX1064GISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1064GISEL-NEXT: s_add_u32 s9, s3, s9
-; GFX1064GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1064GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1064GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1064GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7]
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s8
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s9
; GFX1064GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX1064GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1064GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1064GISEL-NEXT: ; %bb.3: ; %if
; GFX1064GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX1064GISEL-NEXT: s_mov_b64 s[4:5], exec
@@ -3790,7 +4255,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1064GISEL-NEXT: s_add_u32 s5, s7, s5
; GFX1064GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1064GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX1064GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1064GISEL-NEXT: .LBB11_4: ; %endif
; GFX1064GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX1064GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1064GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3805,7 +4270,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032DAGISEL-NEXT: ; implicit-def: $sgpr4_sgpr5
; GFX1032DAGISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1032DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1032DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1032DAGISEL-NEXT: s_mov_b32 s4, exec_lo
; GFX1032DAGISEL-NEXT: s_bcnt1_i32_b32 s4, s4
@@ -3818,7 +4283,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032DAGISEL-NEXT: s_add_u32 s3, s9, s3
; GFX1032DAGISEL-NEXT: s_mul_i32 s4, s2, s4
; GFX1032DAGISEL-NEXT: s_add_u32 s5, s3, s5
-; GFX1032DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1032DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1032DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032DAGISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1032DAGISEL-NEXT: v_mov_b32_e32 v0, s4
@@ -3850,7 +4315,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032GISEL-NEXT: ; implicit-def: $sgpr6_sgpr7
; GFX1032GISEL-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032GISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1032GISEL-NEXT: ; %bb.1: ; %else
; GFX1032GISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX1032GISEL-NEXT: s_bcnt1_i32_b32 s6, s6
@@ -3863,13 +4328,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032GISEL-NEXT: s_add_u32 s3, s9, s3
; GFX1032GISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX1032GISEL-NEXT: s_add_u32 s7, s3, s7
-; GFX1032GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1032GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1032GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1032GISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s6
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s7
; GFX1032GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2
-; GFX1032GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1032GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1032GISEL-NEXT: ; %bb.3: ; %if
; GFX1032GISEL-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX1032GISEL-NEXT: s_mov_b32 s3, exec_lo
@@ -3885,7 +4350,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1032GISEL-NEXT: s_add_u32 s5, s5, s8
; GFX1032GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1032GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX1032GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1032GISEL-NEXT: .LBB11_4: ; %endif
; GFX1032GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2
; GFX1032GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1032GISEL-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
@@ -3902,7 +4367,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1164DAGISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7]
-; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1164DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1164DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1164DAGISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1164DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
@@ -3917,7 +4382,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164DAGISEL-NEXT: s_add_u32 s3, s10, s3
; GFX1164DAGISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1164DAGISEL-NEXT: s_add_u32 s9, s3, s9
-; GFX1164DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1164DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1164DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164DAGISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7]
; GFX1164DAGISEL-NEXT: v_mov_b32_e32 v0, s8
@@ -3953,7 +4418,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1164GISEL-NEXT: s_xor_b64 s[6:7], exec, s[6:7]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1164GISEL-NEXT: ; %bb.1: ; %else
; GFX1164GISEL-NEXT: s_mov_b64 s[8:9], exec
; GFX1164GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
@@ -3968,13 +4433,13 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164GISEL-NEXT: s_add_u32 s3, s10, s3
; GFX1164GISEL-NEXT: s_mul_i32 s8, s2, s8
; GFX1164GISEL-NEXT: s_add_u32 s9, s3, s9
-; GFX1164GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1164GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1164GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1164GISEL-NEXT: s_or_saveexec_b64 s[2:3], s[6:7]
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s8
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s9
; GFX1164GISEL-NEXT: s_xor_b64 exec, exec, s[2:3]
-; GFX1164GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1164GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1164GISEL-NEXT: ; %bb.3: ; %if
; GFX1164GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34
; GFX1164GISEL-NEXT: s_mov_b64 s[6:7], exec
@@ -3992,7 +4457,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1164GISEL-NEXT: s_add_u32 s5, s5, s7
; GFX1164GISEL-NEXT: v_mov_b32_e32 v0, s4
; GFX1164GISEL-NEXT: v_mov_b32_e32 v1, s5
-; GFX1164GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1164GISEL-NEXT: .LBB11_4: ; %endif
; GFX1164GISEL-NEXT: s_or_b64 exec, exec, s[2:3]
; GFX1164GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1164GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1]
@@ -4009,7 +4474,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX1132DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1132DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1132DAGISEL-NEXT: ; %bb.1: ; %else
; GFX1132DAGISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX1132DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
@@ -4024,7 +4489,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132DAGISEL-NEXT: s_add_u32 s3, s9, s3
; GFX1132DAGISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX1132DAGISEL-NEXT: s_add_u32 s7, s3, s7
-; GFX1132DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1132DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX1132DAGISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132DAGISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1132DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7
@@ -4059,7 +4524,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132GISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132GISEL-NEXT: v_cmpx_le_u32_e32 16, v0
; GFX1132GISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX1132GISEL-NEXT: ; %bb.1: ; %else
; GFX1132GISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
@@ -4074,12 +4539,12 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132GISEL-NEXT: s_add_u32 s3, s9, s3
; GFX1132GISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX1132GISEL-NEXT: s_add_u32 s7, s3, s7
-; GFX1132GISEL-NEXT: .LBB9_2: ; %Flow
+; GFX1132GISEL-NEXT: .LBB11_2: ; %Flow
; GFX1132GISEL-NEXT: s_waitcnt lgkmcnt(0)
; GFX1132GISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7
; GFX1132GISEL-NEXT: s_xor_b32 exec_lo, exec_lo, s2
-; GFX1132GISEL-NEXT: s_cbranch_execz .LBB9_4
+; GFX1132GISEL-NEXT: s_cbranch_execz .LBB11_4
; GFX1132GISEL-NEXT: ; %bb.3: ; %if
; GFX1132GISEL-NEXT: s_load_b64 s[4:5], s[4:5], 0x34
; GFX1132GISEL-NEXT: s_mov_b32 s3, exec_lo
@@ -4097,7 +4562,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX1132GISEL-NEXT: s_add_u32 s5, s5, s6
; GFX1132GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132GISEL-NEXT: v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v1, s5
-; GFX1132GISEL-NEXT: .LBB9_4: ; %endif
+; GFX1132GISEL-NEXT: .LBB11_4: ; %endif
; GFX1132GISEL-NEXT: s_or_b32 exec_lo, exec_lo, s2
; GFX1132GISEL-NEXT: v_mov_b32_e32 v2, 0
; GFX1132GISEL-NEXT: global_store_b64 v2, v[0:1], s[0:1]
@@ -4114,7 +4579,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX12DAGISEL-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX12DAGISEL-NEXT: v_cmpx_lt_u32_e32 15, v0
; GFX12DAGISEL-NEXT: s_xor_b32 s8, exec_lo, s8
-; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB9_2
+; GFX12DAGISEL-NEXT: s_cbranch_execz .LBB11_2
; GFX12DAGISEL-NEXT: ; %bb.1: ; %else
; GFX12DAGISEL-NEXT: s_mov_b32 s6, exec_lo
; GFX12DAGISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
@@ -4129,7 +4594,7 @@ define amdgpu_kernel void @divergent_cfg_i64(ptr addrspace(1) %out, i64 %in, i64
; GFX12DAGISEL-NEXT: s_add_co_u32 s3, s9, s3
; GFX12DAGISEL-NEXT: s_mul_i32 s6, s2, s6
; GFX12DAGISEL-NEXT: s_add_co_u32 s7, s3, s7
-; GFX12DAGISEL-NEXT: .LBB9_2: ; %Flow
+; GFX12DAGISEL-NEXT: .LBB11_2: ; %Flow
; GFX12DAGISEL-NEXT: s_wait_kmcnt 0x0
; GFX12DAGISEL-NEXT: s_or_saveexec_b32 s2, s8
; GFX12DAGISEL-NEXT: v_dual_mov_b32 v0, s6 :: v_dual_mov_b32 v1, s7
@@ -4174,3 +4639,9 @@ endif:
store i64 %combine, ptr addrspace(1) %out
ret void
}
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; GFX1132DAGISEL-FAKE16: {{.*}}
+; GFX1132GISEL-FAKE16: {{.*}}
+; GFX1164DAGISEL-FAKE16: {{.*}}
+; GFX1164GISEL-FAKE16: {{.*}}
+; GFX12DAGISEL-FAKE16: {{.*}}
More information about the llvm-branch-commits
mailing list