llvmorg-github-actions[bot] wrote:
<!--LLVM PR SUMMARY COMMENT--> @llvm/pr-subscribers-llvm-transforms Author: Alexey Bataev (alexey-bataev) <details> <summary>Changes</summary> --- Full diff: https://github.com/llvm/llvm-project/pull/235285.diff 1 Files Affected: - (added) llvm/test/Transforms/SLPVectorizer/X86/bit-test-narrowed-samesign.ll (+178) ``````````diff diff --git a/llvm/test/Transforms/SLPVectorizer/X86/bit-test-narrowed-samesign.ll b/llvm/test/Transforms/SLPVectorizer/X86/bit-test-narrowed-samesign.ll new file mode 100644 index 0000000000000..55beaeb6ee0c2 --- /dev/null +++ b/llvm/test/Transforms/SLPVectorizer/X86/bit-test-narrowed-samesign.ll @@ -0,0 +1,178 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6 +; RUN: opt -S --passes=slp-vectorizer -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s + +define void @samesign_ne_sext_mask_sign_bit(ptr noalias %pa, ptr noalias %dst) { +; CHECK-LABEL: define void @samesign_ne_sext_mask_sign_bit( +; CHECK-SAME: ptr noalias [[PA:%.*]], ptr noalias [[DST:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[PA]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = and <4 x i8> [[TMP0]], <i8 -128, i8 -128, i8 -128, i8 1> +; CHECK-NEXT: [[TMP2:%.*]] = icmp samesign ne <4 x i8> [[TMP1]], zeroinitializer +; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[TMP2]], <4 x i32> splat (i32 3), <4 x i32> splat (i32 5) +; CHECK-NEXT: store <4 x i32> [[TMP3]], ptr [[DST]], align 4 +; CHECK-NEXT: ret void +; +entry: + %pa1 = getelementptr i8, ptr %pa, i64 1 + %pa2 = getelementptr i8, ptr %pa, i64 2 + %pa3 = getelementptr i8, ptr %pa, i64 3 + %a0 = load i8, ptr %pa, align 1 + %a1 = load i8, ptr %pa1, align 1 + %a2 = load i8, ptr %pa2, align 1 + %a3 = load i8, ptr %pa3, align 1 + %A0 = sext i8 %a0 to i32 + %A1 = sext i8 %a1 to i32 + %A2 = sext i8 %a2 to i32 + %A3 = sext i8 %a3 to i32 + %m0 = and i32 %A0, 128 + %m1 = and i32 %A1, 128 + %m2 = and i32 %A2, 128 + %c0 = icmp samesign ne i32 %m0, 0 + %c1 = icmp samesign ne i32 %m1, 0 + %c2 = icmp samesign ne i32 %m2, 0 + %c3 = trunc i32 %A3 to i1 + %s0 = select i1 %c0, i32 3, i32 5 + %s1 = select i1 %c1, i32 3, i32 5 + %s2 = select i1 %c2, i32 3, i32 5 + %s3 = select i1 %c3, i32 3, i32 5 + %pd1 = getelementptr i32, ptr %dst, i64 1 + %pd2 = getelementptr i32, ptr %dst, i64 2 + %pd3 = getelementptr i32, ptr %dst, i64 3 + store i32 %s0, ptr %dst, align 4 + store i32 %s1, ptr %pd1, align 4 + store i32 %s2, ptr %pd2, align 4 + store i32 %s3, ptr %pd3, align 4 + ret void +} + +define void @samesign_eq_sext_mask_sign_bit(ptr noalias %pa, ptr noalias %dst) { +; CHECK-LABEL: define void @samesign_eq_sext_mask_sign_bit( +; CHECK-SAME: ptr noalias [[PA:%.*]], ptr noalias [[DST:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[PA]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = and <4 x i8> [[TMP0]], <i8 -128, i8 -128, i8 -128, i8 1> +; CHECK-NEXT: [[TMP2:%.*]] = icmp samesign eq <4 x i8> [[TMP1]], <i8 0, i8 0, i8 0, i8 1> +; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[TMP2]], <4 x i32> splat (i32 3), <4 x i32> splat (i32 5) +; CHECK-NEXT: store <4 x i32> [[TMP3]], ptr [[DST]], align 4 +; CHECK-NEXT: ret void +; +entry: + %pa1 = getelementptr i8, ptr %pa, i64 1 + %pa2 = getelementptr i8, ptr %pa, i64 2 + %pa3 = getelementptr i8, ptr %pa, i64 3 + %a0 = load i8, ptr %pa, align 1 + %a1 = load i8, ptr %pa1, align 1 + %a2 = load i8, ptr %pa2, align 1 + %a3 = load i8, ptr %pa3, align 1 + %A0 = sext i8 %a0 to i32 + %A1 = sext i8 %a1 to i32 + %A2 = sext i8 %a2 to i32 + %A3 = sext i8 %a3 to i32 + %m0 = and i32 %A0, 128 + %m1 = and i32 %A1, 128 + %m2 = and i32 %A2, 128 + %c0 = icmp samesign eq i32 %m0, 0 + %c1 = icmp samesign eq i32 %m1, 0 + %c2 = icmp samesign eq i32 %m2, 0 + %c3 = trunc i32 %A3 to i1 + %s0 = select i1 %c0, i32 3, i32 5 + %s1 = select i1 %c1, i32 3, i32 5 + %s2 = select i1 %c2, i32 3, i32 5 + %s3 = select i1 %c3, i32 3, i32 5 + %pd1 = getelementptr i32, ptr %dst, i64 1 + %pd2 = getelementptr i32, ptr %dst, i64 2 + %pd3 = getelementptr i32, ptr %dst, i64 3 + store i32 %s0, ptr %dst, align 4 + store i32 %s1, ptr %pd1, align 4 + store i32 %s2, ptr %pd2, align 4 + store i32 %s3, ptr %pd3, align 4 + ret void +} + +define void @samesign_ne_sext_mask_non_sign_bit(ptr noalias %pa, ptr noalias %dst) { +; CHECK-LABEL: define void @samesign_ne_sext_mask_non_sign_bit( +; CHECK-SAME: ptr noalias [[PA:%.*]], ptr noalias [[DST:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[PA]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = and <4 x i8> [[TMP0]], <i8 64, i8 64, i8 64, i8 1> +; CHECK-NEXT: [[TMP2:%.*]] = icmp samesign ne <4 x i8> [[TMP1]], zeroinitializer +; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[TMP2]], <4 x i32> splat (i32 3), <4 x i32> splat (i32 5) +; CHECK-NEXT: store <4 x i32> [[TMP3]], ptr [[DST]], align 4 +; CHECK-NEXT: ret void +; +entry: + %pa1 = getelementptr i8, ptr %pa, i64 1 + %pa2 = getelementptr i8, ptr %pa, i64 2 + %pa3 = getelementptr i8, ptr %pa, i64 3 + %a0 = load i8, ptr %pa, align 1 + %a1 = load i8, ptr %pa1, align 1 + %a2 = load i8, ptr %pa2, align 1 + %a3 = load i8, ptr %pa3, align 1 + %A0 = sext i8 %a0 to i32 + %A1 = sext i8 %a1 to i32 + %A2 = sext i8 %a2 to i32 + %A3 = sext i8 %a3 to i32 + %m0 = and i32 %A0, 64 + %m1 = and i32 %A1, 64 + %m2 = and i32 %A2, 64 + %c0 = icmp samesign ne i32 %m0, 0 + %c1 = icmp samesign ne i32 %m1, 0 + %c2 = icmp samesign ne i32 %m2, 0 + %c3 = trunc i32 %A3 to i1 + %s0 = select i1 %c0, i32 3, i32 5 + %s1 = select i1 %c1, i32 3, i32 5 + %s2 = select i1 %c2, i32 3, i32 5 + %s3 = select i1 %c3, i32 3, i32 5 + %pd1 = getelementptr i32, ptr %dst, i64 1 + %pd2 = getelementptr i32, ptr %dst, i64 2 + %pd3 = getelementptr i32, ptr %dst, i64 3 + store i32 %s0, ptr %dst, align 4 + store i32 %s1, ptr %pd1, align 4 + store i32 %s2, ptr %pd2, align 4 + store i32 %s3, ptr %pd3, align 4 + ret void +} + +define void @samesign_ne_zext_mask_sign_bit(ptr noalias %pa, ptr noalias %dst) { +; CHECK-LABEL: define void @samesign_ne_zext_mask_sign_bit( +; CHECK-SAME: ptr noalias [[PA:%.*]], ptr noalias [[DST:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[PA]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = and <4 x i8> [[TMP0]], <i8 -128, i8 -128, i8 -128, i8 1> +; CHECK-NEXT: [[TMP2:%.*]] = icmp ne <4 x i8> [[TMP1]], zeroinitializer +; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[TMP2]], <4 x i32> splat (i32 3), <4 x i32> splat (i32 5) +; CHECK-NEXT: store <4 x i32> [[TMP3]], ptr [[DST]], align 4 +; CHECK-NEXT: ret void +; +entry: + %pa1 = getelementptr i8, ptr %pa, i64 1 + %pa2 = getelementptr i8, ptr %pa, i64 2 + %pa3 = getelementptr i8, ptr %pa, i64 3 + %a0 = load i8, ptr %pa, align 1 + %a1 = load i8, ptr %pa1, align 1 + %a2 = load i8, ptr %pa2, align 1 + %a3 = load i8, ptr %pa3, align 1 + %A0 = zext i8 %a0 to i32 + %A1 = zext i8 %a1 to i32 + %A2 = zext i8 %a2 to i32 + %A3 = zext i8 %a3 to i32 + %m0 = and i32 %A0, 128 + %m1 = and i32 %A1, 128 + %m2 = and i32 %A2, 128 + %c0 = icmp samesign ne i32 %m0, 0 + %c1 = icmp samesign ne i32 %m1, 0 + %c2 = icmp samesign ne i32 %m2, 0 + %c3 = trunc i32 %A3 to i1 + %s0 = select i1 %c0, i32 3, i32 5 + %s1 = select i1 %c1, i32 3, i32 5 + %s2 = select i1 %c2, i32 3, i32 5 + %s3 = select i1 %c3, i32 3, i32 5 + %pd1 = getelementptr i32, ptr %dst, i64 1 + %pd2 = getelementptr i32, ptr %dst, i64 2 + %pd3 = getelementptr i32, ptr %dst, i64 3 + store i32 %s0, ptr %dst, align 4 + store i32 %s1, ptr %pd1, align 4 + store i32 %s2, ptr %pd2, align 4 + store i32 %s3, ptr %pd3, align 4 + ret void +} `````````` </details> https://github.com/llvm/llvm-project/pull/235285 _______________________________________________ llvm-branch-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
