Skip to content

[PowerPC] Lower ucmp using subtractions #146446

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Oct 11, 2025
Merged

[PowerPC] Lower ucmp using subtractions #146446

merged 1 commit into from
Oct 11, 2025

Conversation

@AZero13
Copy link
Contributor

@AZero13 AZero13 commented Jul 1, 2025

Source: Hacker's delight, page 21.

Using the carry, we can use contractions to use the ucmp.

@llvmbot
Copy link
Member

llvmbot commented Jul 1, 2025

@llvm/pr-subscribers-backend-powerpc

Author: AZero13 (AZero13)

Changes

Source: Hacker's delight, page 21.

Using the carry, we can use contractions to use the ucmp.

Full diff: https://github.com/llvm/llvm-project/pull/146446.diff

6 Files Affected:

  • (modified) llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp (+52)
  • (modified) llvm/lib/Target/PowerPC/GISel/PPCLegalizerInfo.cpp (+5)
  • (modified) llvm/lib/Target/PowerPC/PPCISelLowering.cpp (+39)
  • (modified) llvm/lib/Target/PowerPC/PPCISelLowering.h (+1)
  • (modified) llvm/test/CodeGen/PowerPC/memcmp.ll (+9-11)
  • (modified) llvm/test/CodeGen/PowerPC/ucmp.ll (+28-46)
diff --git a/llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp b/llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp
index 3283a5bb69404..8f2e570299b3b 100644
--- a/llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp
+++ b/llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp
@@ -65,6 +65,9 @@ class PPCInstructionSelector : public InstructionSelector {
   bool selectI64Imm(MachineInstr &I, MachineBasicBlock &MBB,
                     MachineRegisterInfo &MRI) const;
 
+  bool selectUCMP(MachineInstr &I, MachineBasicBlock &MBB,
+                  MachineRegisterInfo &MRI) const;
+
   const PPCTargetMachine &TM;
   const PPCSubtarget &STI;
   const PPCInstrInfo &TII;
@@ -705,6 +708,53 @@ bool PPCInstructionSelector::selectConstantPool(
   return constrainSelectedInstRegOperands(*MI, TII, TRI, RBI);
 }
 
+bool PPCInstructionSelector::selectUCMP(MachineInstr &I, MachineBasicBlock &MBB,
+                                        MachineRegisterInfo &MRI) const {
+  const DebugLoc &DbgLoc = I.getDebugLoc();
+  Register DstReg = I.getOperand(0).getReg();
+  Register LHS = I.getOperand(1).getReg();
+  Register RHS = I.getOperand(2).getReg();
+
+  LLT Ty = MRI.getType(LHS);
+  bool Is64Bit = Ty.getSizeInBits() == 64;
+
+  // Select appropriate opcodes based on operand size
+  unsigned SubfOp = Is64Bit ? PPC::SUBF8 : PPC::SUBF;
+  unsigned SubfcOp = Is64Bit ? PPC::SUBFC8 : PPC::SUBFC;
+  unsigned SubfeOp = Is64Bit ? PPC::SUBFE8 : PPC::SUBFE;
+
+  const TargetRegisterClass *RC =
+      Is64Bit ? &PPC::G8RCRegClass : &PPC::GPRCRegClass;
+
+  // diff = LHS - RHS (subf RHS, LHS -> LHS - RHS)
+  Register DiffReg = MRI.createVirtualRegister(RC);
+  auto Diff =
+      BuildMI(MBB, I, DbgLoc, TII.get(SubfOp), DiffReg).addReg(RHS).addReg(LHS);
+
+  // t1 = RHS - LHS, set carry (subfc LHS, RHS -> RHS - LHS)
+  Register T1Reg = MRI.createVirtualRegister(RC);
+  auto T1 =
+      BuildMI(MBB, I, DbgLoc, TII.get(SubfcOp), T1Reg).addReg(LHS).addReg(RHS);
+
+  // t2 = LHS - RHS + carry (subfe RHS, LHS -> LHS - RHS + CA)
+  Register T2Reg = MRI.createVirtualRegister(RC);
+  auto T2 =
+      BuildMI(MBB, I, DbgLoc, TII.get(SubfeOp), T2Reg).addReg(RHS).addReg(LHS);
+
+  // result = diff - t2 + carry (subfe T2Reg, DiffReg -> diff - t2 + CA)
+  auto Result = BuildMI(MBB, I, DbgLoc, TII.get(SubfeOp), DstReg)
+                    .addReg(T2Reg)
+                    .addReg(DiffReg);
+
+  I.eraseFromParent();
+
+  // Constrain registers
+  return constrainSelectedInstRegOperands(*Diff, TII, TRI, RBI) &&
+         constrainSelectedInstRegOperands(*T1, TII, TRI, RBI) &&
+         constrainSelectedInstRegOperands(*T2, TII, TRI, RBI) &&
+         constrainSelectedInstRegOperands(*Result, TII, TRI, RBI);
+}
+
 bool PPCInstructionSelector::select(MachineInstr &I) {
   auto &MBB = *I.getParent();
   auto &MF = *MBB.getParent();
@@ -775,6 +825,8 @@ bool PPCInstructionSelector::select(MachineInstr &I) {
     return selectI64Imm(I, MBB, MRI);
   case TargetOpcode::G_CONSTANT_POOL:
     return selectConstantPool(I, MBB, MRI);
+  case TargetOpcode::G_UCMP:
+    return selectUCMP(I, MBB, MRI);
   }
   return false;
 }
diff --git a/llvm/lib/Target/PowerPC/GISel/PPCLegalizerInfo.cpp b/llvm/lib/Target/PowerPC/GISel/PPCLegalizerInfo.cpp
index afc8f6bbde1b7..9f48333b5be8d 100644
--- a/llvm/lib/Target/PowerPC/GISel/PPCLegalizerInfo.cpp
+++ b/llvm/lib/Target/PowerPC/GISel/PPCLegalizerInfo.cpp
@@ -72,6 +72,11 @@ PPCLegalizerInfo::PPCLegalizerInfo(const PPCSubtarget &ST) {
   getActionDefinitionsBuilder(G_FCMP).legalForCartesianProduct({S1},
                                                                {S32, S64});
 
+  // Add unsigned 3-way comparison support
+  getActionDefinitionsBuilder(G_UCMP)
+      .legalFor({S32, S64})
+      .clampScalar(0, S32, S64);
+
   getActionDefinitionsBuilder({G_FPTOSI, G_FPTOUI})
       .legalForCartesianProduct({S64}, {S32, S64});
 
diff --git a/llvm/lib/Target/PowerPC/PPCISelLowering.cpp b/llvm/lib/Target/PowerPC/PPCISelLowering.cpp
index 5a4a63469ad6e..b70ef182a2a72 100644
--- a/llvm/lib/Target/PowerPC/PPCISelLowering.cpp
+++ b/llvm/lib/Target/PowerPC/PPCISelLowering.cpp
@@ -1409,6 +1409,11 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
 
   setStackPointerRegisterToSaveRestore(isPPC64 ? PPC::X1 : PPC::R1);
 
+  // Custom handling for PowerPC ucmp instruction
+  setOperationAction(ISD::UCMP, MVT::i32, Custom);
+  if (Subtarget.isPPC64())
+    setOperationAction(ISD::UCMP, MVT::i64, Custom);
+
   // We have target-specific dag combine patterns for the following nodes:
   setTargetDAGCombine({ISD::AND, ISD::ADD, ISD::SHL, ISD::SRA, ISD::SRL,
                        ISD::MUL, ISD::FMA, ISD::SINT_TO_FP, ISD::BUILD_VECTOR});
@@ -12470,6 +12475,38 @@ SDValue PPCTargetLowering::LowerSSUBO(SDValue Op, SelectionDAG &DAG) const {
   return DAG.getMergeValues({Sub, OverflowTrunc}, dl);
 }
 
+// Lower unsigned 3-way compare producing -1/0/1.
+SDValue PPCTargetLowering::LowerUCMP(SDValue Op, SelectionDAG &DAG) const {
+  SDLoc DL(Op);
+  SDValue A = Op.getOperand(0);
+  SDValue B = Op.getOperand(1);
+  EVT OpVT = A.getValueType();   // operand type (i32 or i64)
+  EVT ResVT = Op.getValueType(); // result type (should be i32)
+
+  // First compute diff = A - B (will become subf).
+  SDValue Diff = DAG.getNode(ISD::SUB, DL, OpVT, A, B);
+
+  // Generate B - A using SUBC to capture carry.
+  SDVTList VTs = DAG.getVTList(OpVT, MVT::i32);
+  SDValue SubC = DAG.getNode(PPCISD::SUBC, DL, VTs, B, A);
+  SDValue CA0 = SubC.getValue(1);
+
+  // t2 = A - B + CA0 using SUBE.
+  SDValue SubE1 = DAG.getNode(PPCISD::SUBE, DL, VTs, A, B, CA0);
+  SDValue CA1 = SubE1.getValue(1);
+
+  // res = diff - t2 + CA1 using SUBE (produces desired -1/0/1).
+  SDValue ResPair = DAG.getNode(PPCISD::SUBE, DL, VTs, Diff, SubE1, CA1);
+
+  // Extract the first result and truncate to result type if needed
+  SDValue Result = ResPair.getValue(0);
+  if (OpVT != ResVT) {
+    Result = DAG.getNode(ISD::TRUNCATE, DL, ResVT, Result);
+  }
+
+  return Result;
+}
+
 /// LowerOperation - Provide custom lowering hooks for some operations.
 ///
 SDValue PPCTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
@@ -12574,6 +12611,8 @@ SDValue PPCTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::UADDO_CARRY:
   case ISD::USUBO_CARRY:
     return LowerADDSUBO_CARRY(Op, DAG);
+  case ISD::UCMP:
+    return LowerUCMP(Op, DAG);
   }
 }
 
diff --git a/llvm/lib/Target/PowerPC/PPCISelLowering.h b/llvm/lib/Target/PowerPC/PPCISelLowering.h
index 4c88bd372b106..7e8dd166ad724 100644
--- a/llvm/lib/Target/PowerPC/PPCISelLowering.h
+++ b/llvm/lib/Target/PowerPC/PPCISelLowering.h
@@ -1329,6 +1329,7 @@ namespace llvm {
     SDValue LowerIS_FPCLASS(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerADDSUBO_CARRY(SDValue Op, SelectionDAG &DAG) const;
     SDValue LowerADDSUBO(SDValue Op, SelectionDAG &DAG) const;
+    SDValue LowerUCMP(SDValue Op, SelectionDAG &DAG) const;
     SDValue lowerToLibCall(const char *LibCallName, SDValue Op,
                            SelectionDAG &DAG) const;
     SDValue lowerLibCallBasedOnType(const char *LibCallFloatName,
diff --git a/llvm/test/CodeGen/PowerPC/memcmp.ll b/llvm/test/CodeGen/PowerPC/memcmp.ll
index 39f9269997315..4998d87cf397b 100644
--- a/llvm/test/CodeGen/PowerPC/memcmp.ll
+++ b/llvm/test/CodeGen/PowerPC/memcmp.ll
@@ -6,12 +6,10 @@ define signext i32 @memcmp8(ptr nocapture readonly %buffer1, ptr nocapture reado
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    ldbrx 3, 0, 3
 ; CHECK-NEXT:    ldbrx 4, 0, 4
-; CHECK-NEXT:    cmpld 3, 4
-; CHECK-NEXT:    subc 3, 4, 3
-; CHECK-NEXT:    subfe 3, 4, 4
-; CHECK-NEXT:    li 4, -1
-; CHECK-NEXT:    neg 3, 3
-; CHECK-NEXT:    isellt 3, 4, 3
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    extsw 3, 3
 ; CHECK-NEXT:    blr
   %call = tail call signext i32 @memcmp(ptr %buffer1, ptr %buffer2, i64 8)
@@ -23,11 +21,11 @@ define signext i32 @memcmp4(ptr nocapture readonly %buffer1, ptr nocapture reado
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lwbrx 3, 0, 3
 ; CHECK-NEXT:    lwbrx 4, 0, 4
-; CHECK-NEXT:    cmplw 3, 4
-; CHECK-NEXT:    sub 5, 4, 3
-; CHECK-NEXT:    li 3, -1
-; CHECK-NEXT:    rldicl 5, 5, 1, 63
-; CHECK-NEXT:    isellt 3, 3, 5
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
+; CHECK-NEXT:    extsw 3, 3
 ; CHECK-NEXT:    blr
   %call = tail call signext i32 @memcmp(ptr %buffer1, ptr %buffer2, i64 4)
   ret i32 %call
diff --git a/llvm/test/CodeGen/PowerPC/ucmp.ll b/llvm/test/CodeGen/PowerPC/ucmp.ll
index d2dff6e7e05c8..4d393dd00e3db 100644
--- a/llvm/test/CodeGen/PowerPC/ucmp.ll
+++ b/llvm/test/CodeGen/PowerPC/ucmp.ll
@@ -4,12 +4,10 @@
 define i8 @ucmp_8_8(i8 zeroext %x, i8 zeroext %y) nounwind {
 ; CHECK-LABEL: ucmp_8_8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    cmplw 3, 4
-; CHECK-NEXT:    sub 5, 4, 3
-; CHECK-NEXT:    li 3, -1
-; CHECK-NEXT:    rldicl 5, 5, 1, 63
-; CHECK-NEXT:    rldic 3, 3, 0, 32
-; CHECK-NEXT:    isellt 3, 3, 5
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i8 @llvm.ucmp(i8 %x, i8 %y)
   ret i8 %1
@@ -18,12 +16,10 @@ define i8 @ucmp_8_8(i8 zeroext %x, i8 zeroext %y) nounwind {
 define i8 @ucmp_8_16(i16 zeroext %x, i16 zeroext %y) nounwind {
 ; CHECK-LABEL: ucmp_8_16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    cmplw 3, 4
-; CHECK-NEXT:    sub 5, 4, 3
-; CHECK-NEXT:    li 3, -1
-; CHECK-NEXT:    rldicl 5, 5, 1, 63
-; CHECK-NEXT:    rldic 3, 3, 0, 32
-; CHECK-NEXT:    isellt 3, 3, 5
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i8 @llvm.ucmp(i16 %x, i16 %y)
   ret i8 %1
@@ -32,14 +28,10 @@ define i8 @ucmp_8_16(i16 zeroext %x, i16 zeroext %y) nounwind {
 define i8 @ucmp_8_32(i32 %x, i32 %y) nounwind {
 ; CHECK-LABEL: ucmp_8_32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    clrldi 5, 4, 32
-; CHECK-NEXT:    clrldi 6, 3, 32
-; CHECK-NEXT:    sub 5, 5, 6
-; CHECK-NEXT:    cmplw 3, 4
-; CHECK-NEXT:    li 3, -1
-; CHECK-NEXT:    rldic 3, 3, 0, 32
-; CHECK-NEXT:    rldicl 5, 5, 1, 63
-; CHECK-NEXT:    isellt 3, 3, 5
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i8 @llvm.ucmp(i32 %x, i32 %y)
   ret i8 %1
@@ -48,12 +40,10 @@ define i8 @ucmp_8_32(i32 %x, i32 %y) nounwind {
 define i8 @ucmp_8_64(i64 %x, i64 %y) nounwind {
 ; CHECK-LABEL: ucmp_8_64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    cmpld 3, 4
-; CHECK-NEXT:    subc 3, 4, 3
-; CHECK-NEXT:    subfe 3, 4, 4
-; CHECK-NEXT:    li 4, -1
-; CHECK-NEXT:    neg 3, 3
-; CHECK-NEXT:    isellt 3, 4, 3
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i8 @llvm.ucmp(i64 %x, i64 %y)
   ret i8 %1
@@ -82,14 +72,10 @@ define i8 @ucmp_8_128(i128 %x, i128 %y) nounwind {
 define i32 @ucmp_32_32(i32 %x, i32 %y) nounwind {
 ; CHECK-LABEL: ucmp_32_32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    clrldi 5, 4, 32
-; CHECK-NEXT:    clrldi 6, 3, 32
-; CHECK-NEXT:    sub 5, 5, 6
-; CHECK-NEXT:    cmplw 3, 4
-; CHECK-NEXT:    li 3, -1
-; CHECK-NEXT:    rldic 3, 3, 0, 32
-; CHECK-NEXT:    rldicl 5, 5, 1, 63
-; CHECK-NEXT:    isellt 3, 3, 5
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i32 @llvm.ucmp(i32 %x, i32 %y)
   ret i32 %1
@@ -98,12 +84,10 @@ define i32 @ucmp_32_32(i32 %x, i32 %y) nounwind {
 define i32 @ucmp_32_64(i64 %x, i64 %y) nounwind {
 ; CHECK-LABEL: ucmp_32_64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    cmpld 3, 4
-; CHECK-NEXT:    subc 3, 4, 3
-; CHECK-NEXT:    subfe 3, 4, 4
-; CHECK-NEXT:    li 4, -1
-; CHECK-NEXT:    neg 3, 3
-; CHECK-NEXT:    isellt 3, 4, 3
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i32 @llvm.ucmp(i64 %x, i64 %y)
   ret i32 %1
@@ -112,12 +96,10 @@ define i32 @ucmp_32_64(i64 %x, i64 %y) nounwind {
 define i64 @ucmp_64_64(i64 %x, i64 %y) nounwind {
 ; CHECK-LABEL: ucmp_64_64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    subc 5, 4, 3
-; CHECK-NEXT:    cmpld 3, 4
-; CHECK-NEXT:    li 3, -1
-; CHECK-NEXT:    subfe 5, 4, 4
-; CHECK-NEXT:    neg 5, 5
-; CHECK-NEXT:    isellt 3, 3, 5
+; CHECK-NEXT:    subc 6, 4, 3
+; CHECK-NEXT:    sub 5, 3, 4
+; CHECK-NEXT:    subfe 3, 4, 3
+; CHECK-NEXT:    subfe 3, 3, 5
 ; CHECK-NEXT:    blr
   %1 = call i64 @llvm.ucmp(i64 %x, i64 %y)
   ret i64 %1

@AZero13 AZero13 force-pushed the powerpc branch 4 times, most recently from 63c8377 to a5edbe9 Compare July 16, 2025 17:12
@AZero13
Copy link
Contributor Author

AZero13 commented Sep 22, 2025

Note: I do not have merge permissions, so you can please merge if you approve this.

@AZero13
Copy link
Contributor Author

AZero13 commented Sep 24, 2025

@lei137

arsenm
arsenm reviewed Sep 25, 2025
View reviewed changes
@AZero13 AZero13 requested a review from arsenm September 25, 2025 13:56
@AZero13
[PowerPC] Lower ucmp using subtractions
8918751 
Source: Hacker's delight, page 21.

Using the carry, we can use contractions to use the ucmp.
@arsenm arsenm merged commit 07eeb5f into llvm:main Oct 11, 2025
9 checks passed
@llvm-ci
Copy link
Collaborator

llvm-ci commented Oct 11, 2025

LLVM Buildbot has detected a new failure on builder clang-aarch64-quick running on linaro-clang-aarch64-quick while building llvm at step 5 "ninja check 1".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/65/builds/23871

Here is the relevant piece of the build log for the reference 
Step 5 (ninja check 1) failure: stage 1 checked (failure)
******************** TEST 'lit :: timeout-hang.py' FAILED ********************
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 18
not env -u FILECHECK_OPTS "/usr/bin/python3.10" /home/tcwg-buildbot/worker/clang-aarch64-quick/llvm/llvm/utils/lit/lit.py -j1 --order=lexical Inputs/timeout-hang/run-nonexistent.txt  --timeout=15 --param external=0 | "/usr/bin/python3.10" /home/tcwg-buildbot/worker/clang-aarch64-quick/stage1/utils/lit/tests/timeout-hang.py 5
# executed command: not env -u FILECHECK_OPTS /usr/bin/python3.10 /home/tcwg-buildbot/worker/clang-aarch64-quick/llvm/llvm/utils/lit/lit.py -j1 --order=lexical Inputs/timeout-hang/run-nonexistent.txt --timeout=15 --param external=0
# .---command stderr------------
# | lit.py: /home/tcwg-buildbot/worker/clang-aarch64-quick/llvm/llvm/utils/lit/lit/main.py:74: note: The test suite configuration requested an individual test timeout of 0 seconds but a timeout of 15 seconds was requested on the command line. Forcing timeout to be 15 seconds.
# `-----------------------------
# executed command: /usr/bin/python3.10 /home/tcwg-buildbot/worker/clang-aarch64-quick/stage1/utils/lit/tests/timeout-hang.py 5
# .---command stdout------------
# | Testing took 6.56s, which is beyond the grace period of 5.0s
# `-----------------------------
# error: command failed with exit status: 1

--

********************

@AZero13 AZero13 deleted the powerpc branch October 11, 2025 03:56
DharuniRAcharya pushed a commit to DharuniRAcharya/llvm-project that referenced this pull request Oct 13, 2025
@AZero13 @DharuniRAcharya
[PowerPC] Lower ucmp using subtractions (llvm#146446)
5846118 
Source: Hacker's delight, page 21.

Using the carry, we can use contractions to use the ucmp.
akadutta pushed a commit to akadutta/llvm-project that referenced this pull request Oct 14, 2025
@AZero13 @akadutta
[PowerPC] Lower ucmp using subtractions (llvm#146446)
0b51fb4 
Source: Hacker's delight, page 21.

Using the carry, we can use contractions to use the ucmp.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@arsenm arsenm arsenm approved these changes

@amy-kwan amy-kwan Awaiting requested review from amy-kwan

@diggerlin diggerlin Awaiting requested review from diggerlin

@lei137 lei137 Awaiting requested review from lei137

Assignees

No one assigned

Labels

backend:PowerPC

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@AZero13 @llvmbot @llvm-ci @arsenm