[llvm] AMDGPU/GlobalISel: add RegBankLegalize rules for bitfield extract (PR #132381)
Petar Avramovic via llvm-commits
llvm-commits at lists.llvm.org
Fri Mar 21 05:15:01 PDT 2025
https://github.com/petar-avramovic created https://github.com/llvm/llvm-project/pull/132381
Divergent S32 instruction is available, for S64 need to lower to S32.
Uniform instructions available for both S32 and S64 but need to pack
bitfield offset and size of bitfield into S32. Uniform instruction is
straight up selected since there is no available isel pattern.
>From a0088e75160e03b5ef50645f7e8bcc4aa88513cb Mon Sep 17 00:00:00 2001
From: Petar Avramovic <Petar.Avramovic at amd.com>
Date: Fri, 21 Mar 2025 11:14:05 +0100
Subject: [PATCH] AMDGPU/GlobalISel: add RegBankLegalize rules for bitfield
extract
Divergent S32 instruction is available, for S64 need to lower to S32.
Uniform instructions available for both S32 and S64 but need to pack
bitfield offset and size of bitfield into S32. Uniform instruction is
straight up selected since there is no available isel pattern.
---
.../AMDGPU/AMDGPURegBankLegalizeHelper.cpp | 100 ++++++++++++++++++
.../AMDGPU/AMDGPURegBankLegalizeRules.cpp | 13 ++-
.../AMDGPU/AMDGPURegBankLegalizeRules.h | 2 +
.../AMDGPU/GlobalISel/regbankselect-sbfx.mir | 66 ++++++------
.../AMDGPU/GlobalISel/regbankselect-ubfx.mir | 68 ++++++------
llvm/test/CodeGen/AMDGPU/GlobalISel/sbfx.ll | 6 +-
llvm/test/CodeGen/AMDGPU/GlobalISel/ubfx.ll | 6 +-
7 files changed, 193 insertions(+), 68 deletions(-)
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeHelper.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeHelper.cpp
index 3383175fc1bdb..e4eaa01951a7f 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeHelper.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeHelper.cpp
@@ -15,9 +15,12 @@
#include "AMDGPUGlobalISelUtils.h"
#include "AMDGPUInstrInfo.h"
#include "AMDGPURegisterBankInfo.h"
+#include "GCNSubtarget.h"
#include "MCTargetDesc/AMDGPUMCTargetDesc.h"
+#include "llvm/CodeGen/GlobalISel/GenericMachineInstrs.h"
#include "llvm/CodeGen/GlobalISel/MachineIRBuilder.h"
#include "llvm/CodeGen/MachineUniformityAnalysis.h"
+#include "llvm/IR/IntrinsicsAMDGPU.h"
#define DEBUG_TYPE "amdgpu-regbanklegalize"
@@ -225,6 +228,103 @@ void RegBankLegalizeHelper::lower(MachineInstr &MI,
MI.eraseFromParent();
break;
}
+ case Div_BFE: {
+ Register Dst = MI.getOperand(0).getReg();
+ assert(MRI.getType(Dst) == LLT::scalar(64));
+ bool Signed = isa<GIntrinsic>(MI) ? MI.getOperand(1).getIntrinsicID() ==
+ Intrinsic::amdgcn_sbfe
+ : MI.getOpcode() == AMDGPU::G_SBFX;
+ unsigned FirstOpnd = isa<GIntrinsic>(MI) ? 2 : 1;
+ // Extract bitfield from Src, LSBit is the least-significant bit for the
+ // extraction (field offset) and Width is size of bitfield.
+ Register Src = MI.getOperand(FirstOpnd).getReg();
+ Register LSBit = MI.getOperand(FirstOpnd + 1).getReg();
+ Register Width = MI.getOperand(FirstOpnd + 2).getReg();
+ // Comments are for signed bitfield extract, similar for unsigned. x is sign
+ // bit. s is sign, l is LSB and y are remaining bits of bitfield to extract.
+
+ // Src >> LSBit Hi|Lo: x?????syyyyyyl??? -> xxxx?????syyyyyyl
+ unsigned SHROpc = Signed ? AMDGPU::G_ASHR : AMDGPU::G_LSHR;
+ auto SHRSrc = B.buildInstr(SHROpc, {{VgprRB, S64}}, {Src, LSBit});
+
+ auto ConstWidth = getIConstantVRegValWithLookThrough(Width, MRI);
+
+ // Expand to Src >> LSBit << (64 - Width) >> (64 - Width)
+ // << (64 - Width): Hi|Lo: xxxx?????syyyyyyl -> syyyyyyl000000000
+ // >> (64 - Width): Hi|Lo: syyyyyyl000000000 -> ssssssssssyyyyyyl
+ if (!ConstWidth) {
+ auto Amt = B.buildSub(VgprRB_S32, B.buildConstant(SgprRB_S32, 64), Width);
+ auto SignBit = B.buildShl({VgprRB, S64}, SHRSrc, Amt);
+ B.buildInstr(SHROpc, {Dst}, {SignBit, Amt});
+ MI.eraseFromParent();
+ return;
+ }
+
+ auto WidthImm = ConstWidth->Value.getZExtValue();
+ auto UnmergeSHRSrc = B.buildUnmerge(VgprRB_S32, SHRSrc);
+ Register SHRSrcLo = UnmergeSHRSrc.getReg(0);
+ Register SHRSrcHi = UnmergeSHRSrc.getReg(1);
+ auto Zero = B.buildConstant({VgprRB, S32}, 0);
+ unsigned BFXOpc = Signed ? AMDGPU::G_SBFX : AMDGPU::G_UBFX;
+
+ if (WidthImm <= 32) {
+ // SHRSrc Hi|Lo: ????????|???syyyl -> ????????|ssssyyyl
+ auto Lo = B.buildInstr(BFXOpc, {VgprRB_S32}, {SHRSrcLo, Zero, Width});
+ MachineInstrBuilder Hi;
+ if (Signed) {
+ // SHRSrc Hi|Lo: ????????|ssssyyyl -> ssssssss|ssssyyyl
+ Hi = B.buildAShr(VgprRB_S32, Lo, B.buildConstant(VgprRB_S32, 31));
+ } else {
+ // SHRSrc Hi|Lo: ????????|000syyyl -> 00000000|000syyyl
+ Hi = Zero;
+ }
+ B.buildMergeLikeInstr(Dst, {Lo, Hi});
+ } else {
+ auto Amt = B.buildConstant(VgprRB_S32, WidthImm - 32);
+ // SHRSrc Hi|Lo: ??????sy|yyyyyyyl -> sssssssy|yyyyyyyl
+ auto Hi = B.buildInstr(BFXOpc, {VgprRB_S32}, {SHRSrcHi, Zero, Amt});
+ B.buildMergeLikeInstr(Dst, {SHRSrcLo, Hi});
+ }
+
+ MI.eraseFromParent();
+ return;
+ }
+ case Uni_BFE: {
+ Register DstReg = MI.getOperand(0).getReg();
+ LLT Ty = MRI.getType(DstReg);
+ bool Signed = isa<GIntrinsic>(MI) ? MI.getOperand(1).getIntrinsicID() ==
+ Intrinsic::amdgcn_sbfe
+ : MI.getOpcode() == AMDGPU::G_SBFX;
+ unsigned FirstOpnd = isa<GIntrinsic>(MI) ? 2 : 1;
+ Register Src = MI.getOperand(FirstOpnd).getReg();
+ Register LSBit = MI.getOperand(FirstOpnd + 1).getReg();
+ Register Width = MI.getOperand(FirstOpnd + 2).getReg();
+ // For uniform bit field extract there are 4 available instructions, but
+ // LSBit(field offset) and Width(size of bitfield) need to be packed in S32,
+ // field offset in low and size in high 16 bits.
+
+ // Src1 Hi16|Lo16 = Size|FieldOffset
+ auto Mask = B.buildConstant(SgprRB_S32, maskTrailingOnes<unsigned>(6));
+ auto FieldOffset = B.buildAnd(SgprRB_S32, LSBit, Mask);
+ auto Size = B.buildShl(SgprRB_S32, Width, B.buildConstant(SgprRB_S32, 16));
+ auto Src1 = B.buildOr(SgprRB_S32, FieldOffset, Size);
+ unsigned Opc32 = Signed ? AMDGPU::S_BFE_I32 : AMDGPU::S_BFE_U32;
+ unsigned Opc64 = Signed ? AMDGPU::S_BFE_I64 : AMDGPU::S_BFE_U64;
+ unsigned Opc = Ty == S32 ? Opc32 : Opc64;
+
+ // Select machine instruction, because of reg class constraining, insert
+ // copies from reg class to reg bank.
+ auto S_BFE = B.buildInstr(Opc, {{SgprRB, Ty}},
+ {B.buildCopy(Ty, Src), B.buildCopy(S32, Src1)});
+ const GCNSubtarget &ST = B.getMF().getSubtarget<GCNSubtarget>();
+ if (!constrainSelectedInstRegOperands(*S_BFE, *ST.getInstrInfo(),
+ *ST.getRegisterInfo(), RBI))
+ llvm_unreachable("failed to constrain BFE");
+
+ B.buildCopy(DstReg, S_BFE->getOperand(0).getReg());
+ MI.eraseFromParent();
+ return;
+ }
case SplitLoad: {
LLT DstTy = MRI.getType(MI.getOperand(0).getReg());
unsigned Size = DstTy.getSizeInBits();
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
index 6ee15709d2fa6..7959bf30ca27d 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.cpp
@@ -450,6 +450,14 @@ RegBankLegalizeRules::RegBankLegalizeRules(const GCNSubtarget &_ST,
.Uni(S64, {{Sgpr64}, {Sgpr64, Sgpr32}})
.Div(S64, {{Vgpr64}, {Vgpr64, Vgpr32}});
+ addRulesForGOpcs({G_LSHR}, Standard).Uni(S32, {{Sgpr32}, {Sgpr32, Sgpr32}});
+
+ addRulesForGOpcs({G_UBFX, G_SBFX}, Standard)
+ .Uni(S32, {{Sgpr32}, {Sgpr32, Sgpr32, Sgpr32}, Uni_BFE})
+ .Div(S32, {{Vgpr32}, {Vgpr32, Vgpr32, Vgpr32}})
+ .Uni(S64, {{Sgpr64}, {Sgpr64, Sgpr32, Sgpr32}, Uni_BFE})
+ .Div(S64, {{Vgpr64}, {Vgpr64, Vgpr32, Vgpr32}, Div_BFE});
+
// Note: we only write S1 rules for G_IMPLICIT_DEF, G_CONSTANT, G_FCONSTANT
// and G_FREEZE here, rest is trivially regbankselected earlier
addRulesForGOpcs({G_IMPLICIT_DEF}).Any({{UniS1}, {{Sgpr32Trunc}, {}}});
@@ -628,6 +636,9 @@ RegBankLegalizeRules::RegBankLegalizeRules(const GCNSubtarget &_ST,
.Div(S32, {{}, {Vgpr32, None, Vgpr32, Vgpr32}});
addRulesForIOpcs({amdgcn_readfirstlane})
- .Any({{UniS32, _, DivS32}, {{}, {Sgpr32, None, Vgpr32}}});
+ .Any({{UniS32, _, DivS32}, {{}, {Sgpr32, None, Vgpr32}}})
+ // this should not exist in the first place, it is from call lowering
+ // readfirstlaning just in case register is not in sgpr.
+ .Any({{UniS32, _, UniS32}, {{}, {Sgpr32, None, Vgpr32}}});
} // end initialize rules
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.h b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.h
index 6bde7f2cd676d..5edc88aaff73e 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegBankLegalizeRules.h
@@ -168,6 +168,8 @@ enum LoweringMethodID {
DoNotLower,
VccExtToSel,
UniExtToSel,
+ Uni_BFE,
+ Div_BFE,
VgprToVccCopy,
SplitTo32,
Ext32To64,
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-sbfx.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-sbfx.mir
index 97c006a1a7216..572f1ea2516f1 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-sbfx.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-sbfx.mir
@@ -1,6 +1,5 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass=regbankselect -regbankselect-fast -verify-machineinstrs -o - %s | FileCheck %s
-# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass=regbankselect -regbankselect-greedy -verify-machineinstrs -o - %s | FileCheck %s
+# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass="amdgpu-regbankselect,amdgpu-regbanklegalize" -verify-machineinstrs -o - %s | FileCheck %s
...
@@ -96,12 +95,11 @@ body: |
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr3
; CHECK-NEXT: [[ASHR:%[0-9]+]]:vgpr(s64) = G_ASHR [[COPY]], [[COPY1]](s32)
- ; CHECK-NEXT: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[ASHR]](s64)
- ; CHECK-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 64
+ ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 64
; CHECK-NEXT: [[SUB:%[0-9]+]]:vgpr(s32) = G_SUB [[C]], [[COPY2]]
; CHECK-NEXT: [[SHL:%[0-9]+]]:vgpr(s64) = G_SHL [[ASHR]], [[SUB]](s32)
; CHECK-NEXT: [[ASHR1:%[0-9]+]]:vgpr(s64) = G_ASHR [[SHL]], [[SUB]](s32)
- ; CHECK-NEXT: $vgpr0_vgpr1 = COPY %3:vgpr(s64)
+ ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[ASHR1]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
%1:_(s32) = COPY $vgpr2
%2:_(s32) = COPY $vgpr3
@@ -124,12 +122,11 @@ body: |
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
; CHECK-NEXT: [[ASHR:%[0-9]+]]:vgpr(s64) = G_ASHR [[COPY]], [[COPY1]](s32)
- ; CHECK-NEXT: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[ASHR]](s64)
- ; CHECK-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 64
+ ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 64
; CHECK-NEXT: [[SUB:%[0-9]+]]:vgpr(s32) = G_SUB [[C]], [[COPY2]]
; CHECK-NEXT: [[SHL:%[0-9]+]]:vgpr(s64) = G_SHL [[ASHR]], [[SUB]](s32)
; CHECK-NEXT: [[ASHR1:%[0-9]+]]:vgpr(s64) = G_ASHR [[SHL]], [[SUB]](s32)
- ; CHECK-NEXT: $vgpr0_vgpr1 = COPY %3:vgpr(s64)
+ ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[ASHR1]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
%1:_(s32) = COPY $vgpr0
%2:_(s32) = COPY $vgpr1
@@ -216,12 +213,11 @@ body: |
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr(s64) = COPY [[COPY]](s64)
; CHECK-NEXT: [[ASHR:%[0-9]+]]:vgpr(s64) = G_ASHR [[COPY3]], [[COPY1]](s32)
- ; CHECK-NEXT: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[ASHR]](s64)
- ; CHECK-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 64
+ ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 64
; CHECK-NEXT: [[SUB:%[0-9]+]]:vgpr(s32) = G_SUB [[C]], [[COPY2]]
; CHECK-NEXT: [[SHL:%[0-9]+]]:vgpr(s64) = G_SHL [[ASHR]], [[SUB]](s32)
; CHECK-NEXT: [[ASHR1:%[0-9]+]]:vgpr(s64) = G_ASHR [[SHL]], [[SUB]](s32)
- ; CHECK-NEXT: $vgpr0_vgpr1 = COPY %3:vgpr(s64)
+ ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[ASHR1]](s64)
%0:_(s64) = COPY $sgpr0_sgpr1
%1:_(s32) = COPY $vgpr0
%2:_(s32) = COPY $vgpr1
@@ -266,16 +262,19 @@ body: |
; CHECK-LABEL: name: test_sbfx_s32_sss
; CHECK: liveins: $sgpr0, $sgpr1, $sgpr3
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32(s32) = COPY $sgpr0
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[COPY1]], [[C]]
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C1]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_I32_:%[0-9]+]]:sreg_32(s32) = S_BFE_I32 [[COPY]](s32), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0 = COPY [[S_BFE_I32_]](s32)
+ ; CHECK-NEXT: [[OR:%[0-9]+]]:sgpr(s32) = G_OR [[AND]], [[SHL]]
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32(s32) = COPY [[COPY]](s32)
+ ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32(s32) = COPY [[OR]](s32)
+ ; CHECK-NEXT: [[S_BFE_I32_:%[0-9]+]]:sreg_32(s32) = S_BFE_I32 [[COPY3]](s32), [[COPY4]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY [[S_BFE_I32_]](s32)
+ ; CHECK-NEXT: $sgpr0 = COPY [[COPY5]](s32)
%0:_(s32) = COPY $sgpr0
%1:_(s32) = COPY $sgpr1
%2:_(s32) = COPY $sgpr2
@@ -294,16 +293,18 @@ body: |
; CHECK-LABEL: name: test_sbfx_s32_sii
; CHECK: liveins: $sgpr0
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32(s32) = COPY $sgpr0
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 10
; CHECK-NEXT: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
- ; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[C]], [[C2]]
; CHECK-NEXT: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
- ; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[C1]], [[C3]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_I32_:%[0-9]+]]:sreg_32(s32) = S_BFE_I32 [[COPY]](s32), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0 = COPY [[S_BFE_I32_]](s32)
+ ; CHECK-NEXT: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655360
+ ; CHECK-NEXT: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655361
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32(s32) = COPY [[COPY]](s32)
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32(s32) = COPY [[C5]](s32)
+ ; CHECK-NEXT: [[S_BFE_I32_:%[0-9]+]]:sreg_32(s32) = S_BFE_I32 [[COPY1]](s32), [[COPY2]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY [[S_BFE_I32_]](s32)
+ ; CHECK-NEXT: $sgpr0 = COPY [[COPY3]](s32)
%0:_(s32) = COPY $sgpr0
%1:_(s32) = G_CONSTANT i32 1
%2:_(s32) = G_CONSTANT i32 10
@@ -324,16 +325,19 @@ body: |
; CHECK-LABEL: name: test_sbfx_s64_sss
; CHECK: liveins: $sgpr0_sgpr1, $sgpr0, $sgpr1
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_64(s64) = COPY $sgpr0_sgpr1
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr1
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[COPY1]], [[C]]
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C1]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_I64_:%[0-9]+]]:sreg_64(s64) = S_BFE_I64 [[COPY]](s64), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0_sgpr1 = COPY [[S_BFE_I64_]](s64)
+ ; CHECK-NEXT: [[OR:%[0-9]+]]:sgpr(s32) = G_OR [[AND]], [[SHL]]
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_64(s64) = COPY [[COPY]](s64)
+ ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32(s32) = COPY [[OR]](s32)
+ ; CHECK-NEXT: [[S_BFE_I64_:%[0-9]+]]:sreg_64(s64) = S_BFE_I64 [[COPY3]](s64), [[COPY4]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s64) = COPY [[S_BFE_I64_]](s64)
+ ; CHECK-NEXT: $sgpr0_sgpr1 = COPY [[COPY5]](s64)
%0:_(s64) = COPY $sgpr0_sgpr1
%1:_(s32) = COPY $sgpr0
%2:_(s32) = COPY $sgpr1
@@ -352,15 +356,17 @@ body: |
; CHECK-LABEL: name: test_sbfx_s64_sii
; CHECK: liveins: $sgpr0_sgpr1
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_64(s64) = COPY $sgpr0_sgpr1
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 10
; CHECK-NEXT: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
- ; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[C]], [[C2]]
; CHECK-NEXT: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
- ; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[C1]], [[C3]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_I64_:%[0-9]+]]:sreg_64(s64) = S_BFE_I64 [[COPY]](s64), [[OR]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655360
+ ; CHECK-NEXT: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655361
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_64(s64) = COPY [[COPY]](s64)
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32(s32) = COPY [[C5]](s32)
+ ; CHECK-NEXT: [[S_BFE_I64_:%[0-9]+]]:sreg_64(s64) = S_BFE_I64 [[COPY1]](s64), [[COPY2]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s64) = COPY [[S_BFE_I64_]](s64)
%0:_(s64) = COPY $sgpr0_sgpr1
%1:_(s32) = G_CONSTANT i32 1
%2:_(s32) = G_CONSTANT i32 10
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-ubfx.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-ubfx.mir
index 20d280680a09d..267960ad74eff 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-ubfx.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-ubfx.mir
@@ -1,6 +1,5 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass=regbankselect -regbankselect-fast -verify-machineinstrs -o - %s | FileCheck %s
-# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass=regbankselect -regbankselect-greedy -verify-machineinstrs -o - %s | FileCheck %s
+# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass="amdgpu-regbankselect,amdgpu-regbanklegalize" -verify-machineinstrs -o - %s | FileCheck %s
...
@@ -96,12 +95,11 @@ body: |
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr3
; CHECK-NEXT: [[LSHR:%[0-9]+]]:vgpr(s64) = G_LSHR [[COPY]], [[COPY1]](s32)
- ; CHECK-NEXT: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[LSHR]](s64)
- ; CHECK-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 64
+ ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 64
; CHECK-NEXT: [[SUB:%[0-9]+]]:vgpr(s32) = G_SUB [[C]], [[COPY2]]
; CHECK-NEXT: [[SHL:%[0-9]+]]:vgpr(s64) = G_SHL [[LSHR]], [[SUB]](s32)
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:vgpr(s64) = G_LSHR [[SHL]], [[SUB]](s32)
- ; CHECK-NEXT: $vgpr0_vgpr1 = COPY %3:vgpr(s64)
+ ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[LSHR1]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
%1:_(s32) = COPY $vgpr2
%2:_(s32) = COPY $vgpr3
@@ -124,12 +122,11 @@ body: |
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
; CHECK-NEXT: [[LSHR:%[0-9]+]]:vgpr(s64) = G_LSHR [[COPY]], [[COPY1]](s32)
- ; CHECK-NEXT: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[LSHR]](s64)
- ; CHECK-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 64
+ ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 64
; CHECK-NEXT: [[SUB:%[0-9]+]]:vgpr(s32) = G_SUB [[C]], [[COPY2]]
; CHECK-NEXT: [[SHL:%[0-9]+]]:vgpr(s64) = G_SHL [[LSHR]], [[SUB]](s32)
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:vgpr(s64) = G_LSHR [[SHL]], [[SUB]](s32)
- ; CHECK-NEXT: $vgpr0_vgpr1 = COPY %3:vgpr(s64)
+ ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[LSHR1]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
%1:_(s32) = COPY $vgpr0
%2:_(s32) = COPY $vgpr1
@@ -214,12 +211,11 @@ body: |
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr(s64) = COPY [[COPY]](s64)
; CHECK-NEXT: [[LSHR:%[0-9]+]]:vgpr(s64) = G_LSHR [[COPY3]], [[COPY1]](s32)
- ; CHECK-NEXT: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[LSHR]](s64)
- ; CHECK-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 64
+ ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 64
; CHECK-NEXT: [[SUB:%[0-9]+]]:vgpr(s32) = G_SUB [[C]], [[COPY2]]
; CHECK-NEXT: [[SHL:%[0-9]+]]:vgpr(s64) = G_SHL [[LSHR]], [[SUB]](s32)
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:vgpr(s64) = G_LSHR [[SHL]], [[SUB]](s32)
- ; CHECK-NEXT: $vgpr0_vgpr1 = COPY %3:vgpr(s64)
+ ; CHECK-NEXT: $vgpr0_vgpr1 = COPY [[LSHR1]](s64)
%0:_(s64) = COPY $sgpr0_sgpr1
%1:_(s32) = COPY $vgpr0
%2:_(s32) = COPY $vgpr1
@@ -264,16 +260,19 @@ body: |
; CHECK-LABEL: name: test_ubfx_s32_sss
; CHECK: liveins: $sgpr0, $sgpr1, $sgpr2
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32(s32) = COPY $sgpr0
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[COPY1]], [[C]]
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C1]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_U32_:%[0-9]+]]:sreg_32(s32) = S_BFE_U32 [[COPY]](s32), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0 = COPY [[S_BFE_U32_]](s32)
+ ; CHECK-NEXT: [[OR:%[0-9]+]]:sgpr(s32) = G_OR [[AND]], [[SHL]]
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32(s32) = COPY [[COPY]](s32)
+ ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32(s32) = COPY [[OR]](s32)
+ ; CHECK-NEXT: [[S_BFE_U32_:%[0-9]+]]:sreg_32(s32) = S_BFE_U32 [[COPY3]](s32), [[COPY4]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY [[S_BFE_U32_]](s32)
+ ; CHECK-NEXT: $sgpr0 = COPY [[COPY5]](s32)
%0:_(s32) = COPY $sgpr0
%1:_(s32) = COPY $sgpr1
%2:_(s32) = COPY $sgpr2
@@ -292,16 +291,18 @@ body: |
; CHECK-LABEL: name: test_ubfx_s32_sii
; CHECK: liveins: $sgpr0, $sgpr1, $sgpr2
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32(s32) = COPY $sgpr0
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 10
; CHECK-NEXT: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
- ; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[C]], [[C2]]
; CHECK-NEXT: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
- ; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[C1]], [[C3]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_U32_:%[0-9]+]]:sreg_32(s32) = S_BFE_U32 [[COPY]](s32), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0 = COPY [[S_BFE_U32_]](s32)
+ ; CHECK-NEXT: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655360
+ ; CHECK-NEXT: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655361
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32(s32) = COPY [[COPY]](s32)
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32(s32) = COPY [[C5]](s32)
+ ; CHECK-NEXT: [[S_BFE_U32_:%[0-9]+]]:sreg_32(s32) = S_BFE_U32 [[COPY1]](s32), [[COPY2]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY [[S_BFE_U32_]](s32)
+ ; CHECK-NEXT: $sgpr0 = COPY [[COPY3]](s32)
%0:_(s32) = COPY $sgpr0
%1:_(s32) = G_CONSTANT i32 1
%2:_(s32) = G_CONSTANT i32 10
@@ -322,16 +323,19 @@ body: |
; CHECK-LABEL: name: test_ubfx_s64_sss
; CHECK: liveins: $sgpr0_sgpr1, $sgpr2, $sgpr3
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_64(s64) = COPY $sgpr0_sgpr1
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[COPY1]], [[C]]
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C1]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_U64_:%[0-9]+]]:sreg_64(s64) = S_BFE_U64 [[COPY]](s64), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0_sgpr1 = COPY [[S_BFE_U64_]](s64)
+ ; CHECK-NEXT: [[OR:%[0-9]+]]:sgpr(s32) = G_OR [[AND]], [[SHL]]
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_64(s64) = COPY [[COPY]](s64)
+ ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32(s32) = COPY [[OR]](s32)
+ ; CHECK-NEXT: [[S_BFE_U64_:%[0-9]+]]:sreg_64(s64) = S_BFE_U64 [[COPY3]](s64), [[COPY4]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s64) = COPY [[S_BFE_U64_]](s64)
+ ; CHECK-NEXT: $sgpr0_sgpr1 = COPY [[COPY5]](s64)
%0:_(s64) = COPY $sgpr0_sgpr1
%1:_(s32) = COPY $sgpr2
%2:_(s32) = COPY $sgpr3
@@ -350,16 +354,18 @@ body: |
; CHECK-LABEL: name: test_ubfx_s64_sii
; CHECK: liveins: $sgpr0_sgpr1
; CHECK-NEXT: {{ $}}
- ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_64(s64) = COPY $sgpr0_sgpr1
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
; CHECK-NEXT: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 10
; CHECK-NEXT: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 63
- ; CHECK-NEXT: [[AND:%[0-9]+]]:sgpr(s32) = G_AND [[C]], [[C2]]
; CHECK-NEXT: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 16
- ; CHECK-NEXT: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[C1]], [[C3]](s32)
- ; CHECK-NEXT: [[OR:%[0-9]+]]:sreg_32(s32) = G_OR [[AND]], [[SHL]]
- ; CHECK-NEXT: [[S_BFE_U64_:%[0-9]+]]:sreg_64(s64) = S_BFE_U64 [[COPY]](s64), [[OR]](s32), implicit-def $scc
- ; CHECK-NEXT: $sgpr0_sgpr1 = COPY [[S_BFE_U64_]](s64)
+ ; CHECK-NEXT: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655360
+ ; CHECK-NEXT: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 655361
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_64(s64) = COPY [[COPY]](s64)
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32(s32) = COPY [[C5]](s32)
+ ; CHECK-NEXT: [[S_BFE_U64_:%[0-9]+]]:sreg_64(s64) = S_BFE_U64 [[COPY1]](s64), [[COPY2]](s32), implicit-def $scc
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s64) = COPY [[S_BFE_U64_]](s64)
+ ; CHECK-NEXT: $sgpr0_sgpr1 = COPY [[COPY3]](s64)
%0:_(s64) = COPY $sgpr0_sgpr1
%1:_(s32) = G_CONSTANT i32 1
%2:_(s32) = G_CONSTANT i32 10
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/sbfx.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/sbfx.ll
index b1a36c7eaeedc..4f25f16069276 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/sbfx.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/sbfx.ll
@@ -1,7 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=fiji -o - < %s | FileCheck --check-prefixes=GCN %s
-; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -o - < %s | FileCheck --check-prefixes=GCN %s
-; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=gfx1010 -o - < %s | FileCheck --check-prefixes=GCN %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mcpu=fiji -o - < %s | FileCheck --check-prefixes=GCN %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -o - < %s | FileCheck --check-prefixes=GCN %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mcpu=gfx1010 -o - < %s | FileCheck --check-prefixes=GCN %s
; Test vector signed bitfield extract.
define signext i8 @v_ashr_i8_i32(i32 %value) {
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/ubfx.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/ubfx.ll
index 9fb359b768a5e..1fa58d41300ea 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/ubfx.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/ubfx.ll
@@ -1,7 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=fiji -o - < %s | FileCheck --check-prefixes=GCN %s
-; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -o - < %s | FileCheck --check-prefixes=GCN %s
-; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=gfx1010 -o - < %s | FileCheck --check-prefixes=GCN %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mcpu=fiji -o - < %s | FileCheck --check-prefixes=GCN %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -o - < %s | FileCheck --check-prefixes=GCN %s
+; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mcpu=gfx1010 -o - < %s | FileCheck --check-prefixes=GCN %s
; Test vector bitfield extract.
define i32 @v_srl_mask_i32(i32 %value) {
More information about the llvm-commits
mailing list