; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -passes=instcombine < %s | FileCheck %s
target triple = "aarch64-unknown-linux-gnu"
define dso_local <vscale x 4 x float> @dupq_f32_ab_pattern(float %x, float %y) {
; CHECK-LABEL: @dupq_f32_ab_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x float> poison, float [[X:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x float> [[TMP1]], float [[Y:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.vector.insert.nxv4f32.v4f32(<vscale x 4 x float> poison, <4 x float> [[TMP2]], i64 0)
; CHECK-NEXT: [[TMP4:%.*]] = bitcast <vscale x 4 x float> [[TMP3]] to <vscale x 2 x i64>
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <vscale x 2 x i64> [[TMP4]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = bitcast <vscale x 2 x i64> [[TMP5]] to <vscale x 4 x float>
; CHECK-NEXT: ret <vscale x 4 x float> [[TMP6]]
;
%1 = insertelement <4 x float> poison, float %x, i64 0
%2 = insertelement <4 x float> %1, float %y, i64 1
%3 = insertelement <4 x float> %2, float %x, i64 2
%4 = insertelement <4 x float> %3, float %y, i64 3
%5 = tail call <vscale x 4 x float> @llvm.vector.insert.nxv4f32.v4f32(<vscale x 4 x float> poison, <4 x float> %4, i64 0)
%6 = tail call <vscale x 4 x float> @llvm.aarch64.sve.dupq.lane.nxv4f32(<vscale x 4 x float> %5, i64 0)
ret <vscale x 4 x float> %6
}
define dso_local <vscale x 8 x half> @dupq_f16_a_pattern(half %a) {
; CHECK-LABEL: @dupq_f16_a_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <8 x half> [[TMP1]], <8 x half> poison, <8 x i32> zeroinitializer
; CHECK-NEXT: [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP2]], i64 0)
; CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> [[TMP3]], i64 0)
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP4]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %a, i64 1
%3 = insertelement <8 x half> %2, half %a, i64 2
%4 = insertelement <8 x half> %3, half %a, i64 3
%5 = insertelement <8 x half> %4, half %a, i64 4
%6 = insertelement <8 x half> %5, half %a, i64 5
%7 = insertelement <8 x half> %6, half %a, i64 6
%8 = insertelement <8 x half> %7, half %a, i64 7
%9 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %8, i64 0)
%10 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %9, i64 0)
ret <vscale x 8 x half> %10
}
define dso_local <vscale x 8 x half> @dupq_f16_ab_pattern(half %a, half %b) {
; CHECK-LABEL: @dupq_f16_ab_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP2]], i64 0)
; CHECK-NEXT: [[TMP4:%.*]] = bitcast <vscale x 8 x half> [[TMP3]] to <vscale x 4 x i32>
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <vscale x 4 x i32> [[TMP4]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = bitcast <vscale x 4 x i32> [[TMP5]] to <vscale x 8 x half>
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP6]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = insertelement <8 x half> %2, half %a, i64 2
%4 = insertelement <8 x half> %3, half %b, i64 3
%5 = insertelement <8 x half> %4, half %a, i64 4
%6 = insertelement <8 x half> %5, half %b, i64 5
%7 = insertelement <8 x half> %6, half %a, i64 6
%8 = insertelement <8 x half> %7, half %b, i64 7
%9 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %8, i64 0)
%10 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %9, i64 0)
ret <vscale x 8 x half> %10
}
define dso_local <vscale x 8 x half> @dupq_f16_abcd_pattern(half %a, half %b, half %c, half %d) {
; CHECK-LABEL: @dupq_f16_abcd_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half [[C:%.*]], i64 2
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half [[D:%.*]], i64 3
; CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP4]], i64 0)
; CHECK-NEXT: [[TMP6:%.*]] = bitcast <vscale x 8 x half> [[TMP5]] to <vscale x 2 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP8:%.*]] = bitcast <vscale x 2 x i64> [[TMP7]] to <vscale x 8 x half>
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP8]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = insertelement <8 x half> %2, half %c, i64 2
%4 = insertelement <8 x half> %3, half %d, i64 3
%5 = insertelement <8 x half> %4, half %a, i64 4
%6 = insertelement <8 x half> %5, half %b, i64 5
%7 = insertelement <8 x half> %6, half %c, i64 6
%8 = insertelement <8 x half> %7, half %d, i64 7
%9 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %8, i64 0)
%10 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %9, i64 0)
ret <vscale x 8 x half> %10
}
define dso_local <vscale x 8 x half> @dupq_f16_abcnull_pattern(half %a, half %b, half %c, half %d) {
; CHECK-LABEL: @dupq_f16_abcnull_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half [[C:%.*]], i64 2
; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP3]], i64 0)
; CHECK-NEXT: [[TMP5:%.*]] = bitcast <vscale x 8 x half> [[TMP4]] to <vscale x 2 x i64>
; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <vscale x 2 x i64> [[TMP5]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP7:%.*]] = bitcast <vscale x 2 x i64> [[TMP6]] to <vscale x 8 x half>
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP7]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = insertelement <8 x half> %2, half %c, i64 2
%4 = insertelement <8 x half> %3, half %a, i64 4
%5 = insertelement <8 x half> %4, half %b, i64 5
%6 = insertelement <8 x half> %5, half %c, i64 6
%7 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %6, i64 0)
%8 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %7, i64 0)
ret <vscale x 8 x half> %8
}
define dso_local <vscale x 8 x half> @dupq_f16_abnull_pattern(half %a, half %b) {
; CHECK-LABEL: @dupq_f16_abnull_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP2]], i64 0)
; CHECK-NEXT: [[TMP4:%.*]] = bitcast <vscale x 8 x half> [[TMP3]] to <vscale x 4 x i32>
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <vscale x 4 x i32> [[TMP4]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = bitcast <vscale x 4 x i32> [[TMP5]] to <vscale x 8 x half>
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP6]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %2, i64 0)
%4 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %3, i64 0)
ret <vscale x 8 x half> %4
}
define dso_local <vscale x 8 x half> @neg_dupq_f16_non_poison_fixed(half %a, half %b, <8 x half> %v) {
; CHECK-LABEL: @neg_dupq_f16_non_poison_fixed(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> [[V:%.*]], half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP2]], i64 0)
; CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> [[TMP3]], i64 0)
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP4]]
;
%1 = insertelement <8 x half> %v, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = insertelement <8 x half> %2, half %a, i64 0
%4 = insertelement <8 x half> %3, half %b, i64 1
%5 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %4, i64 0)
%6 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %5, i64 0)
ret <vscale x 8 x half> %6
}
define dso_local <vscale x 8 x half> @neg_dupq_f16_into_non_poison_scalable(half %a, half %b, <vscale x 8 x half> %v) {
; CHECK-LABEL: @neg_dupq_f16_into_non_poison_scalable(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> [[V:%.*]], <8 x half> [[TMP2]], i64 0)
; CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> [[TMP3]], i64 0)
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP4]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = insertelement <8 x half> %2, half %a, i64 0
%4 = insertelement <8 x half> %3, half %b, i64 1
%5 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> %v, <8 x half> %4, i64 0)
%6 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %5, i64 0)
ret <vscale x 8 x half> %6
}
; Insert %c to override the last element in the insertelement chain, which will fail to combine
define dso_local <vscale x 8 x half> @neg_dupq_f16_abcd_pattern_double_insert(half %a, half %b, half %c, half %d) {
; CHECK-LABEL: @neg_dupq_f16_abcd_pattern_double_insert(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half [[C:%.*]], i64 2
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half [[D:%.*]], i64 3
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half [[A]], i64 4
; CHECK-NEXT: [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half [[B]], i64 5
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half [[C]], i64 6
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <8 x half> [[TMP7]], half [[C]], i64 7
; CHECK-NEXT: [[TMP9:%.*]] = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP8]], i64 0)
; CHECK-NEXT: [[TMP10:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> [[TMP9]], i64 0)
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP10]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %b, i64 1
%3 = insertelement <8 x half> %2, half %c, i64 2
%4 = insertelement <8 x half> %3, half %d, i64 3
%5 = insertelement <8 x half> %4, half %a, i64 4
%6 = insertelement <8 x half> %5, half %b, i64 5
%7 = insertelement <8 x half> %6, half %c, i64 6
%8 = insertelement <8 x half> %7, half %d, i64 7
%9 = insertelement <8 x half> %8, half %c, i64 7
%10 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %9, i64 0)
%11 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %10, i64 0)
ret <vscale x 8 x half> %11
}
define dso_local <vscale x 8 x half> @dupq_f16_abcd_pattern_reverted_insert(half %a, half %b, half %c, half %d) {
; CHECK-LABEL: @dupq_f16_abcd_pattern_reverted_insert(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[B:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half [[C:%.*]], i64 2
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half [[D:%.*]], i64 3
; CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP4]], i64 0)
; CHECK-NEXT: [[TMP6:%.*]] = bitcast <vscale x 8 x half> [[TMP5]] to <vscale x 2 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP8:%.*]] = bitcast <vscale x 2 x i64> [[TMP7]] to <vscale x 8 x half>
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP8]]
;
%1 = insertelement <8 x half> poison, half %d, i64 7
%2 = insertelement <8 x half> %1, half %c, i64 6
%3 = insertelement <8 x half> %2, half %b, i64 5
%4 = insertelement <8 x half> %3, half %a, i64 4
%5 = insertelement <8 x half> %4, half %d, i64 3
%6 = insertelement <8 x half> %5, half %c, i64 2
%7 = insertelement <8 x half> %6, half %b, i64 1
%8 = insertelement <8 x half> %7, half %a, i64 0
%9 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %8, i64 0)
%10 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %9, i64 0)
ret <vscale x 8 x half> %10
}
define dso_local <vscale x 8 x half> @dupq_f16_ab_no_front_pattern(half %a, half %b) {
; CHECK-LABEL: @dupq_f16_ab_no_front_pattern(
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <8 x half> poison, half [[A:%.*]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x half> [[TMP1]], half [[A]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x half> [[TMP2]], half [[A]], i64 2
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <8 x half> [[TMP3]], half [[B:%.*]], i64 3
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <8 x half> [[TMP4]], half [[A]], i64 4
; CHECK-NEXT: [[TMP6:%.*]] = insertelement <8 x half> [[TMP5]], half [[B]], i64 5
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <8 x half> [[TMP6]], half [[A]], i64 6
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <8 x half> [[TMP7]], half [[B]], i64 7
; CHECK-NEXT: [[TMP9:%.*]] = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> [[TMP8]], i64 0)
; CHECK-NEXT: [[TMP10:%.*]] = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> [[TMP9]], i64 0)
; CHECK-NEXT: ret <vscale x 8 x half> [[TMP10]]
;
%1 = insertelement <8 x half> poison, half %a, i64 0
%2 = insertelement <8 x half> %1, half %a, i64 1
%3 = insertelement <8 x half> %2, half %a, i64 2
%4 = insertelement <8 x half> %3, half %b, i64 3
%5 = insertelement <8 x half> %4, half %a, i64 4
%6 = insertelement <8 x half> %5, half %b, i64 5
%7 = insertelement <8 x half> %6, half %a, i64 6
%8 = insertelement <8 x half> %7, half %b, i64 7
%9 = tail call <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half> poison, <8 x half> %8, i64 0)
%10 = tail call <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half> %9, i64 0)
ret <vscale x 8 x half> %10
}
declare <vscale x 8 x half> @llvm.vector.insert.nxv8f16.v8f16(<vscale x 8 x half>, <8 x half>, i64)
declare <vscale x 8 x half> @llvm.aarch64.sve.dupq.lane.nxv8f16(<vscale x 8 x half>, i64)
declare <vscale x 4 x float> @llvm.vector.insert.nxv4f32.v4f32(<vscale x 4 x float>, <4 x float>, i64)
declare <vscale x 4 x float> @llvm.vector.insert.nxv2f32.v2f32(<vscale x 4 x float>, <2 x float>, i64)
declare <vscale x 4 x float> @llvm.aarch64.sve.dupq.lane.nxv4f32(<vscale x 4 x float>, i64)
declare <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32>, <4 x i32>, i64)
declare <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32>, i64)
declare <vscale x 8 x i16> @llvm.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16>, <8 x i16>, i64)
declare <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16>, i64)
attributes #0 = { "target-features"="+sve" }