| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py |
| ; RUN: llc -mtriple=arm64-apple-ios -mattr=+sve -o - %s | FileCheck %s |
| ; RUN: llc -mtriple=aarch64_be-unknown-linux -mattr=+sve -o - %s | FileCheck --check-prefix=CHECK-BE %s |
| ; RUN: llc -mtriple=arm64-apple-ios -mattr=+global-isel -mattr=+sve -o - %s | FileCheck %s |
| ; RUN: llc -mtriple=aarch64_be-unknown-linux -mattr=+global-isel -mattr=+sve -o - %s | FileCheck --check-prefix=CHECK-BE %s |
| |
| define void @zext_of_concat(ptr %a, ptr %b, ptr %c, ptr %d) nounwind { |
| ; CHECK-LABEL: zext_of_concat: |
| ; CHECK: ; %bb.0: |
| ; CHECK-NEXT: ldr d0, [x0] |
| ; CHECK-NEXT: ldr d1, [x1] |
| ; CHECK-NEXT: add.2s v0, v0, v1 |
| ; CHECK-NEXT: ldr q1, [x2] |
| ; CHECK-NEXT: ushll.2d v0, v0, #0 |
| ; CHECK-NEXT: add.4s v0, v0, v1 |
| ; CHECK-NEXT: str q0, [x2] |
| ; CHECK-NEXT: ret |
| ; |
| ; CHECK-BE-LABEL: zext_of_concat: |
| ; CHECK-BE: // %bb.0: |
| ; CHECK-BE-NEXT: ld1 { v0.2s }, [x0] |
| ; CHECK-BE-NEXT: ld1 { v1.2s }, [x1] |
| ; CHECK-BE-NEXT: add v0.2s, v0.2s, v1.2s |
| ; CHECK-BE-NEXT: movi v1.2d, #0000000000000000 |
| ; CHECK-BE-NEXT: zip1 v0.4s, v0.4s, v0.4s |
| ; CHECK-BE-NEXT: trn2 v0.4s, v0.4s, v1.4s |
| ; CHECK-BE-NEXT: ld1 { v1.4s }, [x2] |
| ; CHECK-BE-NEXT: add v0.4s, v0.4s, v1.4s |
| ; CHECK-BE-NEXT: st1 { v0.4s }, [x2] |
| ; CHECK-BE-NEXT: ret |
| %i0.a = load <2 x i32>, ptr %a |
| %i0.b = load <2 x i32>, ptr %b |
| %i0 = add <2 x i32> %i0.a, %i0.b |
| %i1 = shufflevector <2 x i32> %i0, <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef> |
| %i2 = shufflevector <4 x i32> %i1, <4 x i32> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 1, i32 7> |
| %i3 = load <4 x i32>, ptr %c |
| %i4 = add <4 x i32> %i2, %i3 |
| store <4 x i32> %i4, ptr %c |
| ret void |
| } |
| |
| define void @zext_of_concat_extrause(ptr %a, ptr %b, ptr %c, ptr %d, ptr %e) nounwind { |
| ; CHECK-LABEL: zext_of_concat_extrause: |
| ; CHECK: ; %bb.0: |
| ; CHECK-NEXT: ldr d0, [x1] |
| ; CHECK-NEXT: ldr d1, [x0] |
| ; CHECK-NEXT: add.2s v0, v1, v0 |
| ; CHECK-NEXT: movi.2d v1, #0000000000000000 |
| ; CHECK-NEXT: mov.d v0[1], v0[0] |
| ; CHECK-NEXT: zip1.4s v1, v0, v1 |
| ; CHECK-NEXT: str q0, [x4] |
| ; CHECK-NEXT: ldr q0, [x2] |
| ; CHECK-NEXT: add.4s v0, v1, v0 |
| ; CHECK-NEXT: str q0, [x2] |
| ; CHECK-NEXT: ret |
| ; |
| ; CHECK-BE-LABEL: zext_of_concat_extrause: |
| ; CHECK-BE: // %bb.0: |
| ; CHECK-BE-NEXT: ld1 { v0.2s }, [x1] |
| ; CHECK-BE-NEXT: ld1 { v1.2s }, [x0] |
| ; CHECK-BE-NEXT: movi v2.2d, #0000000000000000 |
| ; CHECK-BE-NEXT: add v0.2s, v1.2s, v0.2s |
| ; CHECK-BE-NEXT: mov v0.d[1], v0.d[0] |
| ; CHECK-BE-NEXT: zip1 v1.4s, v0.4s, v0.4s |
| ; CHECK-BE-NEXT: st1 { v0.4s }, [x4] |
| ; CHECK-BE-NEXT: trn2 v0.4s, v1.4s, v2.4s |
| ; CHECK-BE-NEXT: ld1 { v1.4s }, [x2] |
| ; CHECK-BE-NEXT: add v0.4s, v0.4s, v1.4s |
| ; CHECK-BE-NEXT: st1 { v0.4s }, [x2] |
| ; CHECK-BE-NEXT: ret |
| %i0.a = load <2 x i32>, ptr %a |
| %i0.b = load <2 x i32>, ptr %b |
| %i0 = add <2 x i32> %i0.a, %i0.b |
| %i1 = shufflevector <2 x i32> %i0, <2 x i32> %i0, <4 x i32> <i32 0, i32 1, i32 2, i32 3> |
| store <4 x i32> %i1, ptr %e |
| %i2 = shufflevector <4 x i32> %i1, <4 x i32> zeroinitializer, <4 x i32> <i32 0, i32 5, i32 1, i32 7> |
| %i3 = load <4 x i32>, ptr %c |
| %i4 = add <4 x i32> %i2, %i3 |
| store <4 x i32> %i4, ptr %c |
| ret void |
| } |
| |
| define void @aext_of_concat(ptr %a, ptr %b, ptr %c, ptr %d) nounwind { |
| ; CHECK-LABEL: aext_of_concat: |
| ; CHECK: ; %bb.0: |
| ; CHECK-NEXT: ldr d0, [x0] |
| ; CHECK-NEXT: ldr d1, [x1] |
| ; CHECK-NEXT: add.2s v0, v0, v1 |
| ; CHECK-NEXT: ldr q1, [x2] |
| ; CHECK-NEXT: ushll.2d v0, v0, #0 |
| ; CHECK-NEXT: add.4s v0, v0, v1 |
| ; CHECK-NEXT: str q0, [x2] |
| ; CHECK-NEXT: ret |
| ; |
| ; CHECK-BE-LABEL: aext_of_concat: |
| ; CHECK-BE: // %bb.0: |
| ; CHECK-BE-NEXT: ld1 { v0.2s }, [x0] |
| ; CHECK-BE-NEXT: ld1 { v1.2s }, [x1] |
| ; CHECK-BE-NEXT: add v0.2s, v0.2s, v1.2s |
| ; CHECK-BE-NEXT: ld1 { v1.4s }, [x2] |
| ; CHECK-BE-NEXT: zip1 v0.4s, v0.4s, v0.4s |
| ; CHECK-BE-NEXT: add v0.4s, v0.4s, v1.4s |
| ; CHECK-BE-NEXT: st1 { v0.4s }, [x2] |
| ; CHECK-BE-NEXT: ret |
| %i0.a = load <2 x i32>, ptr %a |
| %i0.b = load <2 x i32>, ptr %b |
| %i0 = add <2 x i32> %i0.a, %i0.b |
| %i1 = shufflevector <2 x i32> %i0, <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef> |
| %i2 = shufflevector <4 x i32> %i1, <4 x i32> poison, <4 x i32> <i32 0, i32 undef, i32 1, i32 undef> |
| %i3 = load <4 x i32>, ptr %c |
| %i4 = add <4 x i32> %i2, %i3 |
| store <4 x i32> %i4, ptr %c |
| ret void |
| } |
| |
| define void @aext_of_concat_extrause(ptr %a, ptr %b, ptr %c, ptr %d, ptr %e) nounwind { |
| ; CHECK-LABEL: aext_of_concat_extrause: |
| ; CHECK: ; %bb.0: |
| ; CHECK-NEXT: ldr d0, [x1] |
| ; CHECK-NEXT: ldr d1, [x0] |
| ; CHECK-NEXT: add.2s v0, v1, v0 |
| ; CHECK-NEXT: mov.16b v1, v0 |
| ; CHECK-NEXT: mov.d v1[1], v0[0] |
| ; CHECK-NEXT: zip1.4s v0, v0, v0 |
| ; CHECK-NEXT: str q1, [x4] |
| ; CHECK-NEXT: ldr q1, [x2] |
| ; CHECK-NEXT: add.4s v0, v0, v1 |
| ; CHECK-NEXT: str q0, [x2] |
| ; CHECK-NEXT: ret |
| ; |
| ; CHECK-BE-LABEL: aext_of_concat_extrause: |
| ; CHECK-BE: // %bb.0: |
| ; CHECK-BE-NEXT: ld1 { v0.2s }, [x1] |
| ; CHECK-BE-NEXT: ld1 { v1.2s }, [x0] |
| ; CHECK-BE-NEXT: add v0.2s, v1.2s, v0.2s |
| ; CHECK-BE-NEXT: mov v1.16b, v0.16b |
| ; CHECK-BE-NEXT: mov v1.d[1], v0.d[0] |
| ; CHECK-BE-NEXT: zip1 v0.4s, v0.4s, v0.4s |
| ; CHECK-BE-NEXT: st1 { v1.4s }, [x4] |
| ; CHECK-BE-NEXT: ld1 { v1.4s }, [x2] |
| ; CHECK-BE-NEXT: add v0.4s, v0.4s, v1.4s |
| ; CHECK-BE-NEXT: st1 { v0.4s }, [x2] |
| ; CHECK-BE-NEXT: ret |
| %i0.a = load <2 x i32>, ptr %a |
| %i0.b = load <2 x i32>, ptr %b |
| %i0 = add <2 x i32> %i0.a, %i0.b |
| %i1 = shufflevector <2 x i32> %i0, <2 x i32> %i0, <4 x i32> <i32 0, i32 1, i32 2, i32 3> |
| store <4 x i32> %i1, ptr %e |
| %i2 = shufflevector <4 x i32> %i1, <4 x i32> zeroinitializer, <4 x i32> <i32 0, i32 undef, i32 1, i32 undef> |
| %i3 = load <4 x i32>, ptr %c |
| %i4 = add <4 x i32> %i2, %i3 |
| store <4 x i32> %i4, ptr %c |
| ret void |
| } |