Skip to content

[RISCV] Use EXTLOAD instead of ZEXTLOAD when lowering riscv_masked_strided_load with zero stride. #97317

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
Jul 1, 2024
Merged

[RISCV] Use EXTLOAD instead of ZEXTLOAD when lowering riscv_masked_strided_load with zero stride. #97317

merged 1 commit into from
Jul 1, 2024

Conversation

topperc
Copy link
Collaborator

@topperc topperc commented Jul 1, 2024 •
edited
Loading

The splat we generate after it doesn't use the extended bits, so it shouldn't matter which extend type we use.

EXTLOAD is lowered as SEXTLOAD on every element type except i8.

@topperc
[RISCV] Use EXTLOAD instead of ZEXTLOAD when lowering riscv_masked_st…
91f8bee 
…rided_load with zero stride.

The upper bits aren't used, no need to force a particular extend type.

EXTLOAD is lowered as SEXTLOAD on every element type except i8.
@llvmbot
Copy link
Member

llvmbot commented Jul 1, 2024

@llvm/pr-subscribers-backend-risc-v

Author: Craig Topper (topperc)

Changes

The upper bits aren't used, no need to force a particular extend type.

EXTLOAD is lowered as SEXTLOAD on every element type except i8.

Full diff: https://github.com/llvm/llvm-project/pull/97317.diff

2 Files Affected:

  • (modified) llvm/lib/Target/RISCV/RISCVISelLowering.cpp (+1-1)
  • (modified) llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll (+79-41)
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 5e94fbec5a04a..ee45f730dc450 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -9326,7 +9326,7 @@ SDValue RISCVTargetLowering::LowerINTRINSIC_W_CHAIN(SDValue Op,
     MVT ScalarVT = ContainerVT.getVectorElementType();
     if (IsUnmasked && isNullConstant(Stride) && ContainerVT.isInteger()) {
       SDValue ScalarLoad =
-          DAG.getExtLoad(ISD::ZEXTLOAD, DL, XLenVT, Load->getChain(), Ptr,
+          DAG.getExtLoad(ISD::EXTLOAD, DL, XLenVT, Load->getChain(), Ptr,
                          ScalarVT, Load->getMemOperand());
       Chain = ScalarLoad.getValue(1);
       Result = lowerScalarSplat(SDValue(), ScalarLoad, VL, ContainerVT, DL, DAG,
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll
index 64ad86db04959..da8db345c8783 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll
@@ -175,13 +175,51 @@ for.cond.cleanup:                                 ; preds = %vector.body
   ret void
 }
 
+define void @gather_zero_stride_i32(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
+; CHECK-LABEL: gather_zero_stride_i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    addi a2, a0, 1024
+; 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:    vse32.v v8, (a0)
+; CHECK-NEXT:    addi a0, a0, 8
+; CHECK-NEXT:    addi a1, a1, 160
+; CHECK-NEXT:    bne a0, a2, .LBB4_1
+; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
+; CHECK-NEXT:    ret
+entry:
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %entry
+  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
+  %vec.ind = phi <8 x i64> [ zeroinitializer, %entry ], [ %vec.ind.next, %vector.body ]
+  %i = mul nuw nsw <8 x i64> %vec.ind, <i64 5, i64 5, i64 5, i64 5, i64 5, i64 5, i64 5, i64 5>
+  %i1 = getelementptr inbounds i8, ptr %B, <8 x i64> %i
+  %wide.masked.gather = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i1, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i32> undef)
+  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
+  %wide.load = load <8 x i32>, ptr %i2, align 4
+  %i4 = add <8 x i32> %wide.load, %wide.masked.gather
+  store <8 x i32> %i4, ptr %i2, align 4
+  %index.next = add nuw i64 %index, 8
+  %vec.ind.next = add <8 x i64> %vec.ind, <i64 32, i64 32, i64 32, i64 32, i64 32, i64 32, i64 32, i64 32>
+  %i6 = icmp eq i64 %index.next, 1024
+  br i1 %i6, label %for.cond.cleanup, label %vector.body
+
+for.cond.cleanup:                                 ; preds = %vector.body
+  ret void
+}
+
 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:  .LBB4_1: # %vector.body
+; 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)
@@ -189,7 +227,7 @@ define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias noc
 ; V-NEXT:    vse8.v v8, (a0)
 ; V-NEXT:    addi a0, a0, 32
 ; V-NEXT:    addi a1, a1, 160
-; V-NEXT:    bne a0, a2, .LBB4_1
+; V-NEXT:    bne a0, a2, .LBB5_1
 ; V-NEXT:  # %bb.2: # %for.cond.cleanup
 ; V-NEXT:    ret
 ;
@@ -198,7 +236,7 @@ define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias noc
 ; ZVE32F-NEXT:    addi a2, a0, 1024
 ; ZVE32F-NEXT:    li a3, 32
 ; ZVE32F-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
-; ZVE32F-NEXT:  .LBB4_1: # %vector.body
+; 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)
@@ -206,7 +244,7 @@ define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias noc
 ; ZVE32F-NEXT:    vse8.v v8, (a0)
 ; ZVE32F-NEXT:    addi a0, a0, 32
 ; ZVE32F-NEXT:    addi a1, a1, 160
-; ZVE32F-NEXT:    bne a0, a2, .LBB4_1
+; ZVE32F-NEXT:    bne a0, a2, .LBB5_1
 ; ZVE32F-NEXT:  # %bb.2: # %for.cond.cleanup
 ; ZVE32F-NEXT:    ret
 ;
@@ -215,7 +253,7 @@ define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias noc
 ; 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:  .LBB4_1: # %vector.body
+; 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)
@@ -224,7 +262,7 @@ define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias noc
 ; 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, .LBB4_1
+; NOT-OPTIMIZED-NEXT:    bne a0, a2, .LBB5_1
 ; NOT-OPTIMIZED-NEXT:  # %bb.2: # %for.cond.cleanup
 ; NOT-OPTIMIZED-NEXT:    ret
 entry:
@@ -260,7 +298,7 @@ define void @scatter(ptr noalias nocapture %A, ptr noalias nocapture readonly %B
 ; CHECK-NEXT:    li a4, 32
 ; CHECK-NEXT:    li a3, 5
 ; CHECK-NEXT:    vsetvli zero, a4, e8, m1, ta, ma
-; CHECK-NEXT:  .LBB5_1: # %vector.body
+; CHECK-NEXT:  .LBB6_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vle8.v v8, (a1)
 ; CHECK-NEXT:    vlse8.v v9, (a0), a3
@@ -268,7 +306,7 @@ define void @scatter(ptr noalias nocapture %A, ptr noalias nocapture readonly %B
 ; CHECK-NEXT:    vsse8.v v8, (a0), a3
 ; CHECK-NEXT:    addi a1, a1, 32
 ; CHECK-NEXT:    addi a0, a0, 160
-; CHECK-NEXT:    bne a1, a2, .LBB5_1
+; CHECK-NEXT:    bne a1, a2, .LBB6_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
@@ -303,7 +341,7 @@ define void @scatter_masked(ptr noalias nocapture %A, ptr noalias nocapture read
 ; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.s.x v0, a4
 ; CHECK-NEXT:    li a4, 5
-; CHECK-NEXT:  .LBB6_1: # %vector.body
+; CHECK-NEXT:  .LBB7_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, mu
 ; CHECK-NEXT:    vle8.v v9, (a1)
@@ -313,7 +351,7 @@ define void @scatter_masked(ptr noalias nocapture %A, ptr noalias nocapture read
 ; CHECK-NEXT:    vsse8.v v9, (a0), a4, v0.t
 ; CHECK-NEXT:    addi a1, a1, 32
 ; CHECK-NEXT:    addi a0, a0, 160
-; CHECK-NEXT:    bne a1, a2, .LBB6_1
+; CHECK-NEXT:    bne a1, a2, .LBB7_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
@@ -349,7 +387,7 @@ define void @gather_pow2(ptr noalias nocapture %A, ptr noalias nocapture readonl
 ; CHECK-NEXT:    add a2, a0, a2
 ; CHECK-NEXT:    li a3, 16
 ; CHECK-NEXT:    li a4, 32
-; CHECK-NEXT:  .LBB7_1: # %vector.body
+; CHECK-NEXT:  .LBB8_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
 ; CHECK-NEXT:    vlse32.v v8, (a1), a3
@@ -361,7 +399,7 @@ define void @gather_pow2(ptr noalias nocapture %A, ptr noalias nocapture readonl
 ; CHECK-NEXT:    vse8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 32
 ; CHECK-NEXT:    addi a1, a1, 128
-; CHECK-NEXT:    bne a0, a2, .LBB7_1
+; CHECK-NEXT:    bne a0, a2, .LBB8_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
@@ -397,7 +435,7 @@ define void @scatter_pow2(ptr noalias nocapture %A, ptr noalias nocapture readon
 ; CHECK-NEXT:    add a2, a1, a2
 ; CHECK-NEXT:    li a3, 32
 ; CHECK-NEXT:    li a4, 16
-; CHECK-NEXT:  .LBB8_1: # %vector.body
+; CHECK-NEXT:  .LBB9_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a1)
@@ -407,7 +445,7 @@ define void @scatter_pow2(ptr noalias nocapture %A, ptr noalias nocapture readon
 ; CHECK-NEXT:    vsse32.v v8, (a0), a4
 ; CHECK-NEXT:    addi a1, a1, 32
 ; CHECK-NEXT:    addi a0, a0, 128
-; CHECK-NEXT:    bne a1, a2, .LBB8_1
+; CHECK-NEXT:    bne a1, a2, .LBB9_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
@@ -448,7 +486,7 @@ define void @struct_gather(ptr noalias nocapture %A, ptr noalias nocapture reado
 ; CHECK-NEXT:    add a2, a0, a2
 ; CHECK-NEXT:    li a3, 16
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
-; CHECK-NEXT:  .LBB9_1: # %vector.body
+; CHECK-NEXT:  .LBB10_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    addi a4, a0, 32
 ; CHECK-NEXT:    addi a5, a1, -128
@@ -462,7 +500,7 @@ define void @struct_gather(ptr noalias nocapture %A, ptr noalias nocapture reado
 ; CHECK-NEXT:    vse32.v v9, (a4)
 ; CHECK-NEXT:    addi a0, a0, 64
 ; CHECK-NEXT:    addi a1, a1, 256
-; CHECK-NEXT:    bne a0, a2, .LBB9_1
+; CHECK-NEXT:    bne a0, a2, .LBB10_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
@@ -508,7 +546,7 @@ define void @gather_unroll(ptr noalias nocapture %A, ptr noalias nocapture reado
 ; CHECK-NEXT:    li a3, 64
 ; CHECK-NEXT:    li a4, 16
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
-; CHECK-NEXT:  .LBB10_1: # %vector.body
+; CHECK-NEXT:  .LBB11_1: # %vector.body
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vlse32.v v8, (a1), a3
 ; CHECK-NEXT:    vlse32.v v9, (a0), a4
@@ -535,7 +573,7 @@ define void @gather_unroll(ptr noalias nocapture %A, ptr noalias nocapture reado
 ; CHECK-NEXT:    addi a2, a2, -8
 ; CHECK-NEXT:    addi a1, a1, 512
 ; CHECK-NEXT:    addi a0, a0, 128
-; CHECK-NEXT:    bnez a2, .LBB10_1
+; CHECK-NEXT:    bnez a2, .LBB11_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:
@@ -597,7 +635,7 @@ define void @gather_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptur
 ; V-NEXT:    add a2, a0, a2
 ; V-NEXT:    li a3, 40
 ; V-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; V-NEXT:  .LBB11_1: # %bb2
+; V-NEXT:  .LBB12_1: # %bb2
 ; V-NEXT:    # =>This Inner Loop Header: Depth=1
 ; V-NEXT:    addi a4, a1, 80
 ; V-NEXT:    vlse64.v v8, (a1), a3
@@ -607,7 +645,7 @@ define void @gather_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptur
 ; V-NEXT:    vse64.v v9, (a4)
 ; V-NEXT:    addi a0, a0, 32
 ; V-NEXT:    addi a1, a1, 160
-; V-NEXT:    bne a0, a2, .LBB11_1
+; V-NEXT:    bne a0, a2, .LBB12_1
 ; V-NEXT:  # %bb.2: # %bb18
 ; V-NEXT:    ret
 ;
@@ -618,7 +656,7 @@ define void @gather_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptur
 ; ZVE32F-NEXT:    add a3, a0, a3
 ; ZVE32F-NEXT:    li a4, 1
 ; ZVE32F-NEXT:    li a5, 40
-; ZVE32F-NEXT:  .LBB11_1: # %bb2
+; ZVE32F-NEXT:  .LBB12_1: # %bb2
 ; ZVE32F-NEXT:    # =>This Inner Loop Header: Depth=1
 ; ZVE32F-NEXT:    mul a6, a4, a5
 ; ZVE32F-NEXT:    add a6, a1, a6
@@ -635,7 +673,7 @@ define void @gather_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptur
 ; ZVE32F-NEXT:    addi a2, a2, 4
 ; ZVE32F-NEXT:    addi a0, a0, 32
 ; ZVE32F-NEXT:    addi a4, a4, 4
-; ZVE32F-NEXT:    bne a0, a3, .LBB11_1
+; ZVE32F-NEXT:    bne a0, a3, .LBB12_1
 ; ZVE32F-NEXT:  # %bb.2: # %bb18
 ; ZVE32F-NEXT:    ret
 bb:
@@ -674,7 +712,7 @@ define void @scatter_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptu
 ; V-NEXT:    add a2, a1, a2
 ; V-NEXT:    li a3, 40
 ; V-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
-; V-NEXT:  .LBB12_1: # %bb2
+; V-NEXT:  .LBB13_1: # %bb2
 ; V-NEXT:    # =>This Inner Loop Header: Depth=1
 ; V-NEXT:    addi a4, a1, 16
 ; V-NEXT:    vle64.v v8, (a1)
@@ -684,7 +722,7 @@ define void @scatter_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptu
 ; V-NEXT:    vsse64.v v9, (a4), a3
 ; V-NEXT:    addi a1, a1, 32
 ; V-NEXT:    addi a0, a0, 160
-; V-NEXT:    bne a1, a2, .LBB12_1
+; V-NEXT:    bne a1, a2, .LBB13_1
 ; V-NEXT:  # %bb.2: # %bb18
 ; V-NEXT:    ret
 ;
@@ -695,7 +733,7 @@ define void @scatter_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptu
 ; ZVE32F-NEXT:    add a3, a1, a3
 ; ZVE32F-NEXT:    li a4, 1
 ; ZVE32F-NEXT:    li a5, 40
-; ZVE32F-NEXT:  .LBB12_1: # %bb2
+; ZVE32F-NEXT:  .LBB13_1: # %bb2
 ; ZVE32F-NEXT:    # =>This Inner Loop Header: Depth=1
 ; ZVE32F-NEXT:    ld a6, 8(a1)
 ; ZVE32F-NEXT:    ld a7, 0(a1)
@@ -712,7 +750,7 @@ define void @scatter_of_pointers(ptr noalias nocapture %arg, ptr noalias nocaptu
 ; ZVE32F-NEXT:    addi a2, a2, 4
 ; ZVE32F-NEXT:    addi a1, a1, 32
 ; ZVE32F-NEXT:    addi a4, a4, 4
-; ZVE32F-NEXT:    bne a1, a3, .LBB12_1
+; ZVE32F-NEXT:    bne a1, a3, .LBB13_1
 ; ZVE32F-NEXT:  # %bb.2: # %bb18
 ; ZVE32F-NEXT:    ret
 bb:
@@ -747,13 +785,13 @@ define void @strided_load_startval_add_with_splat(ptr noalias nocapture %arg, pt
 ; CHECK-LABEL: strided_load_startval_add_with_splat:
 ; CHECK:       # %bb.0: # %bb
 ; CHECK-NEXT:    li a3, 1024
-; CHECK-NEXT:    beq a2, a3, .LBB13_7
+; CHECK-NEXT:    beq a2, a3, .LBB14_7
 ; CHECK-NEXT:  # %bb.1: # %bb3
 ; CHECK-NEXT:    li a3, 1023
 ; CHECK-NEXT:    subw a5, a3, a2
 ; CHECK-NEXT:    li a6, 31
 ; CHECK-NEXT:    mv a4, a2
-; CHECK-NEXT:    bltu a5, a6, .LBB13_5
+; CHECK-NEXT:    bltu a5, a6, .LBB14_5
 ; CHECK-NEXT:  # %bb.2: # %bb9
 ; CHECK-NEXT:    slli a5, a5, 32
 ; CHECK-NEXT:    srli a5, a5, 32
@@ -768,7 +806,7 @@ define void @strided_load_startval_add_with_splat(ptr noalias nocapture %arg, pt
 ; CHECK-NEXT:    li t2, 32
 ; CHECK-NEXT:    li t1, 5
 ; CHECK-NEXT:    vsetvli zero, t2, e8, m1, ta, ma
-; CHECK-NEXT:  .LBB13_3: # %bb15
+; CHECK-NEXT:  .LBB14_3: # %bb15
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vlse8.v v8, (a2), t1
 ; CHECK-NEXT:    vle8.v v9, (a7)
@@ -776,10 +814,10 @@ define void @strided_load_startval_add_with_splat(ptr noalias nocapture %arg, pt
 ; CHECK-NEXT:    vse8.v v8, (a7)
 ; CHECK-NEXT:    addi a7, a7, 32
 ; CHECK-NEXT:    addi a2, a2, 160
-; CHECK-NEXT:    bne a7, t0, .LBB13_3
+; CHECK-NEXT:    bne a7, t0, .LBB14_3
 ; CHECK-NEXT:  # %bb.4: # %bb30
-; CHECK-NEXT:    beq a5, a6, .LBB13_7
-; CHECK-NEXT:  .LBB13_5: # %bb32
+; CHECK-NEXT:    beq a5, a6, .LBB14_7
+; CHECK-NEXT:  .LBB14_5: # %bb32
 ; CHECK-NEXT:    add a2, a0, a4
 ; CHECK-NEXT:    slli a5, a4, 2
 ; CHECK-NEXT:    add a1, a1, a4
@@ -790,7 +828,7 @@ define void @strided_load_startval_add_with_splat(ptr noalias nocapture %arg, pt
 ; CHECK-NEXT:    add a0, a4, a0
 ; CHECK-NEXT:    add a0, a0, a3
 ; CHECK-NEXT:    addi a0, a0, 1
-; CHECK-NEXT:  .LBB13_6: # %bb35
+; CHECK-NEXT:  .LBB14_6: # %bb35
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    lbu a3, 0(a1)
 ; CHECK-NEXT:    lbu a4, 0(a2)
@@ -798,8 +836,8 @@ define void @strided_load_startval_add_with_splat(ptr noalias nocapture %arg, pt
 ; CHECK-NEXT:    sb a3, 0(a2)
 ; CHECK-NEXT:    addi a2, a2, 1
 ; CHECK-NEXT:    addi a1, a1, 5
-; CHECK-NEXT:    bne a2, a0, .LBB13_6
-; CHECK-NEXT:  .LBB13_7: # %bb34
+; CHECK-NEXT:    bne a2, a0, .LBB14_6
+; CHECK-NEXT:  .LBB14_7: # %bb34
 ; CHECK-NEXT:    ret
 bb:
   %i = icmp eq i32 %arg2, 1024
@@ -870,7 +908,7 @@ define void @gather_no_scalar_remainder(ptr noalias nocapture noundef %arg, ptr
 ; CHECK-LABEL: gather_no_scalar_remainder:
 ; CHECK:       # %bb.0: # %bb
 ; CHECK-NEXT:    slli a2, a2, 4
-; CHECK-NEXT:    beqz a2, .LBB14_3
+; CHECK-NEXT:    beqz a2, .LBB15_3
 ; CHECK-NEXT:  # %bb.1: # %bb2
 ; CHECK-NEXT:    addi a2, a2, -16
 ; CHECK-NEXT:    andi a2, a2, -16
@@ -878,7 +916,7 @@ define void @gather_no_scalar_remainder(ptr noalias nocapture noundef %arg, ptr
 ; CHECK-NEXT:    addi a2, a2, 16
 ; CHECK-NEXT:    li a3, 5
 ; CHECK-NEXT:    vsetivli zero, 16, e8, mf2, ta, ma
-; CHECK-NEXT:  .LBB14_2: # %bb4
+; CHECK-NEXT:  .LBB15_2: # %bb4
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    vlse8.v v8, (a1), a3
 ; CHECK-NEXT:    vle8.v v9, (a0)
@@ -886,8 +924,8 @@ define void @gather_no_scalar_remainder(ptr noalias nocapture noundef %arg, ptr
 ; CHECK-NEXT:    vse8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 16
 ; CHECK-NEXT:    addi a1, a1, 80
-; CHECK-NEXT:    bne a0, a2, .LBB14_2
-; CHECK-NEXT:  .LBB14_3: # %bb16
+; CHECK-NEXT:    bne a0, a2, .LBB15_2
+; CHECK-NEXT:  .LBB15_3: # %bb16
 ; CHECK-NEXT:    ret
 bb:
   %i = shl i64 %arg2, 4
@@ -922,7 +960,7 @@ define void @gather_zero_stride_fp(ptr noalias nocapture %A, ptr noalias nocaptu
 ; CHECK-NEXT:    lui a2, 1
 ; CHECK-NEXT:    add a2, a0, a2
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
-; CHECK-NEXT:  .LBB15_1: # %vector.body
+; 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)
@@ -930,7 +968,7 @@ define void @gather_zero_stride_fp(ptr noalias nocapture %A, ptr noalias nocaptu
 ; CHECK-NEXT:    vse32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    addi a1, a1, 640
-; CHECK-NEXT:    bne a0, a2, .LBB15_1
+; CHECK-NEXT:    bne a0, a2, .LBB16_1
 ; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
 ; CHECK-NEXT:    ret
 entry:

@preames
Copy link
Collaborator

preames commented Jul 1, 2024

To make sure I understand the reasoning here, let me restate. This is correct because the splat operation works on the element type (not the potentially wider register type), and thus the high bits of the loaded value are fully don't care?

Assuming that's a correct restatement, LGTM.

@topperc
Copy link
Collaborator Author

topperc commented Jul 1, 2024

To make sure I understand the reasoning here, let me restate. This is correct because the splat operation works on the element type (not the potentially wider register type), and thus the high bits of the loaded value are fully don't care?

Assuming that's a correct restatement, LGTM.

That's correct.

@topperc topperc merged commit 7e6e498 into llvm:main Jul 1, 2024
7 of 9 checks passed
@topperc topperc deleted the pr/zextload branch July 1, 2024 20:11
lravenclaw pushed a commit to lravenclaw/llvm-project that referenced this pull request Jul 3, 2024
@topperc @lravenclaw
[RISCV] Use EXTLOAD instead of ZEXTLOAD when lowering riscv_masked_st…
5cd2e15 
…rided_load with zero stride. (llvm#97317)

The splat we generate after the load doesn't use the extended bits, so it
shouldn't matter which extend type we use.

EXTLOAD is lowered as SEXTLOAD on every element type except i8.
kbluck pushed a commit to kbluck/llvm-project that referenced this pull request Jul 6, 2024
@topperc @kbluck
[RISCV] Use EXTLOAD instead of ZEXTLOAD when lowering riscv_masked_st…
996a323 
…rided_load with zero stride. (llvm#97317)

The splat we generate after the load doesn't use the extended bits, so it
shouldn't matter which extend type we use.

EXTLOAD is lowered as SEXTLOAD on every element type except i8.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@preames preames Awaiting requested review from preames

@lukel97 lukel97 Awaiting requested review from lukel97

@yetingk yetingk Awaiting requested review from yetingk

@wangpc-pp wangpc-pp Awaiting requested review from wangpc-pp

Assignees
No one assigned
Labels
backend:RISC-V
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@topperc @llvmbot @preames