[RISCV] Emit VP strided loads/stores in RISCVGatherScatterLowering

RISCVGatherScatterLowering is the main user of riscv_masked_strided_{load,store}, which we can remove if we replace them with their VP equivalents.

Submitting early as a draft to show the regressions in the test diff that llvm#97800 and llvm#97798 (or the CGP version) are needed to fix.

llvm/lib/Target/RISCV/RISCVGatherScatterLowering.cpp

@@ -515,17 +515,23 @@ bool RISCVGatherScatterLowering::tryCreateStridedLoadStore(IntrinsicInst *II,
 
   Builder.SetInsertPoint(II);
 
+  Value *EVL = Builder.CreateElementCount(
+      IntegerType::get(Ctx, 32), cast<VectorType>(DataType)->getElementCount());
+
   CallInst *Call;
-  if (II->getIntrinsicID() == Intrinsic::masked_gather)
+  if (II->getIntrinsicID() == Intrinsic::masked_gather) {
     Call = Builder.CreateIntrinsic(
-        Intrinsic::riscv_masked_strided_load,
+        Intrinsic::experimental_vp_strided_load,
         {DataType, BasePtr->getType(), Stride->getType()},
-        {II->getArgOperand(3), BasePtr, Stride, II->getArgOperand(2)});
-  else
+        {BasePtr, Stride, II->getArgOperand(2), EVL});
+    Call = Builder.CreateIntrinsic(
+        Intrinsic::vp_select, {DataType},
+        {II->getOperand(2), Call, II->getArgOperand(3), EVL});
+  } else
     Call = Builder.CreateIntrinsic(
-        Intrinsic::riscv_masked_strided_store,
+        Intrinsic::experimental_vp_strided_store,
         {DataType, BasePtr->getType(), Stride->getType()},
-        {II->getArgOperand(0), BasePtr, Stride, II->getArgOperand(3)});
+        {II->getArgOperand(0), BasePtr, Stride, II->getArgOperand(3), EVL});
 
   Call->takeName(II);
   II->replaceAllUsesWith(Call);

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll

@@ -144,9 +144,9 @@ define void @gather_zero_stride(ptr noalias nocapture %A, ptr noalias nocapture
 ; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-NEXT:  .LBB3_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    lbu a3, 0(a1)
-; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vadd.vx v8, v8, a3
+; CHECK-NEXT:    vlse8.v v8, (a1), zero
+; CHECK-NEXT:    vle8.v v9, (a0)
+; CHECK-NEXT:    vadd.vv v8, v9, v8
 ; CHECK-NEXT:    vse8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 32
 ; CHECK-NEXT:    addi a1, a1, 160
@@ -182,9 +182,9 @@ define void @gather_zero_stride_i32(ptr noalias nocapture %A, ptr noalias nocapt
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
 ; CHECK-NEXT:  .LBB4_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    lw a3, 0(a1)
-; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vadd.vx v8, v8, a3
+; CHECK-NEXT:    vlse32.v v8, (a1), zero
+; CHECK-NEXT:    vle32.v v9, (a0)
+; CHECK-NEXT:    vadd.vv v8, v9, v8
 ; CHECK-NEXT:    vse32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 8
 ; CHECK-NEXT:    addi a1, a1, 160
@@ -214,57 +214,22 @@ for.cond.cleanup:                                 ; preds = %vector.body
 }
 
 define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
-; V-LABEL: gather_zero_stride_unfold:
-; V:       # %bb.0: # %entry
-; V-NEXT:    addi a2, a0, 1024
-; V-NEXT:    li a3, 32
-; V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
-; V-NEXT:  .LBB5_1: # %vector.body
-; V-NEXT:    # =>This Inner Loop Header: Depth=1
-; V-NEXT:    vlse8.v v8, (a1), zero
-; V-NEXT:    vle8.v v9, (a0)
-; V-NEXT:    vdivu.vv v8, v8, v9
-; V-NEXT:    vse8.v v8, (a0)
-; V-NEXT:    addi a0, a0, 32
-; V-NEXT:    addi a1, a1, 160
-; V-NEXT:    bne a0, a2, .LBB5_1
-; V-NEXT:  # %bb.2: # %for.cond.cleanup
-; V-NEXT:    ret
-;
-; ZVE32F-LABEL: gather_zero_stride_unfold:
-; ZVE32F:       # %bb.0: # %entry
-; ZVE32F-NEXT:    addi a2, a0, 1024
-; ZVE32F-NEXT:    li a3, 32
-; ZVE32F-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
-; ZVE32F-NEXT:  .LBB5_1: # %vector.body
-; ZVE32F-NEXT:    # =>This Inner Loop Header: Depth=1
-; ZVE32F-NEXT:    vlse8.v v8, (a1), zero
-; ZVE32F-NEXT:    vle8.v v9, (a0)
-; ZVE32F-NEXT:    vdivu.vv v8, v8, v9
-; ZVE32F-NEXT:    vse8.v v8, (a0)
-; ZVE32F-NEXT:    addi a0, a0, 32
-; ZVE32F-NEXT:    addi a1, a1, 160
-; ZVE32F-NEXT:    bne a0, a2, .LBB5_1
-; ZVE32F-NEXT:  # %bb.2: # %for.cond.cleanup
-; ZVE32F-NEXT:    ret
-;
-; NOT-OPTIMIZED-LABEL: gather_zero_stride_unfold:
-; NOT-OPTIMIZED:       # %bb.0: # %entry
-; NOT-OPTIMIZED-NEXT:    addi a2, a0, 1024
-; NOT-OPTIMIZED-NEXT:    li a3, 32
-; NOT-OPTIMIZED-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
-; NOT-OPTIMIZED-NEXT:  .LBB5_1: # %vector.body
-; NOT-OPTIMIZED-NEXT:    # =>This Inner Loop Header: Depth=1
-; NOT-OPTIMIZED-NEXT:    lbu a3, 0(a1)
-; NOT-OPTIMIZED-NEXT:    vle8.v v8, (a0)
-; NOT-OPTIMIZED-NEXT:    vmv.v.x v9, a3
-; NOT-OPTIMIZED-NEXT:    vdivu.vv v8, v9, v8
-; NOT-OPTIMIZED-NEXT:    vse8.v v8, (a0)
-; NOT-OPTIMIZED-NEXT:    addi a0, a0, 32
-; NOT-OPTIMIZED-NEXT:    addi a1, a1, 160
-; NOT-OPTIMIZED-NEXT:    bne a0, a2, .LBB5_1
-; NOT-OPTIMIZED-NEXT:  # %bb.2: # %for.cond.cleanup
-; NOT-OPTIMIZED-NEXT:    ret
+; CHECK-LABEL: gather_zero_stride_unfold:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    addi a2, a0, 1024
+; CHECK-NEXT:    li a3, 32
+; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
+; CHECK-NEXT:  .LBB5_1: # %vector.body
+; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT:    vlse8.v v8, (a1), zero
+; CHECK-NEXT:    vle8.v v9, (a0)
+; CHECK-NEXT:    vdivu.vv v8, v8, v9
+; CHECK-NEXT:    vse8.v v8, (a0)
+; CHECK-NEXT:    addi a0, a0, 32
+; CHECK-NEXT:    addi a1, a1, 160
+; CHECK-NEXT:    bne a0, a2, .LBB5_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
 entry:
   br label %vector.body
 
@@ -962,9 +927,9 @@ define void @gather_zero_stride_fp(ptr noalias nocapture %A, ptr noalias nocaptu
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
 ; CHECK-NEXT:  .LBB16_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    flw fa5, 0(a1)
-; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vfadd.vf v8, v8, fa5
+; CHECK-NEXT:    vlse32.v v8, (a1), zero
+; CHECK-NEXT:    vle32.v v9, (a0)
+; CHECK-NEXT:    vfadd.vv v8, v9, v8
 ; CHECK-NEXT:    vse32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    addi a1, a1, 640
@@ -992,3 +957,5 @@ vector.body:                                      ; preds = %vector.body, %entry
 for.cond.cleanup:                                 ; preds = %vector.body
   ret void
 }
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; NOT-OPTIMIZED: {{.*}}