blob: bf113a7cabc15a985d3e9c3a43430f7c0b07a703 [file] [log] [blame]
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
;
; Unpredicated dup instruction (which is an alias for mov):
; * register + register,
; * register + immediate
;
define <vscale x 16 x i8> @dup_i8(i8 %b) {
; CHECK-LABEL: dup_i8:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.b, w0
; CHECK-NEXT: ret
%out = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 %b)
ret <vscale x 16 x i8> %out
}
define <vscale x 16 x i8> @dup_imm_i8() {
; CHECK-LABEL: dup_imm_i8:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.b, #16 // =0x10
; CHECK-NEXT: ret
%out = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 16)
ret <vscale x 16 x i8> %out
}
define <vscale x 8 x i16> @dup_i16(i16 %b) {
; CHECK-LABEL: dup_i16:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.h, w0
; CHECK-NEXT: ret
%out = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %b)
ret <vscale x 8 x i16> %out
}
define <vscale x 8 x i16> @dup_imm_i16(i16 %b) {
; CHECK-LABEL: dup_imm_i16:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.h, #16 // =0x10
; CHECK-NEXT: ret
%out = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 16)
ret <vscale x 8 x i16> %out
}
define <vscale x 4 x i32> @dup_i32(i32 %b) {
; CHECK-LABEL: dup_i32:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.s, w0
; CHECK-NEXT: ret
%out = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 %b)
ret <vscale x 4 x i32> %out
}
define <vscale x 4 x i32> @dup_imm_i32(i32 %b) {
; CHECK-LABEL: dup_imm_i32:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.s, #16 // =0x10
; CHECK-NEXT: ret
%out = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 16)
ret <vscale x 4 x i32> %out
}
define <vscale x 2 x i64> @dup_i64(i64 %b) {
; CHECK-LABEL: dup_i64:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.d, x0
; CHECK-NEXT: ret
%out = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 %b)
ret <vscale x 2 x i64> %out
}
define <vscale x 2 x i64> @dup_imm_i64(i64 %b) {
; CHECK-LABEL: dup_imm_i64:
; CHECK: // %bb.0:
; CHECK-NEXT: mov z0.d, #16 // =0x10
; CHECK-NEXT: ret
%out = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 16)
ret <vscale x 2 x i64> %out
}
define <vscale x 8 x half> @dup_f16(half %b) {
; CHECK-LABEL: dup_f16:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $h0 killed $h0 def $z0
; CHECK-NEXT: mov z0.h, h0
; CHECK-NEXT: ret
%out = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half %b)
ret <vscale x 8 x half> %out
}
define <vscale x 8 x bfloat> @dup_bf16(bfloat %b) #0 {
; CHECK-LABEL: dup_bf16:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $h0 killed $h0 def $z0
; CHECK-NEXT: mov z0.h, h0
; CHECK-NEXT: ret
%out = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat %b)
ret <vscale x 8 x bfloat> %out
}
define <vscale x 8 x half> @dup_imm_f16(half %b) {
; CHECK-LABEL: dup_imm_f16:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov z0.h, #16.00000000
; CHECK-NEXT: ret
%out = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 16.)
ret <vscale x 8 x half> %out
}
define <vscale x 4 x float> @dup_f32(float %b) {
; CHECK-LABEL: dup_f32:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $s0 killed $s0 def $z0
; CHECK-NEXT: mov z0.s, s0
; CHECK-NEXT: ret
%out = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float %b)
ret <vscale x 4 x float> %out
}
define <vscale x 4 x float> @dup_imm_f32(float %b) {
; CHECK-LABEL: dup_imm_f32:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov z0.s, #16.00000000
; CHECK-NEXT: ret
%out = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 16.)
ret <vscale x 4 x float> %out
}
define <vscale x 2 x double> @dup_f64(double %b) {
; CHECK-LABEL: dup_f64:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT: mov z0.d, d0
; CHECK-NEXT: ret
%out = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double %b)
ret <vscale x 2 x double> %out
}
define <vscale x 2 x double> @dup_imm_f64(double %b) {
; CHECK-LABEL: dup_imm_f64:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov z0.d, #16.00000000
; CHECK-NEXT: ret
%out = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 16.)
ret <vscale x 2 x double> %out
}
define <vscale x 2 x float> @dup_fmov_imm_f32_2() {
; CHECK-LABEL: dup_fmov_imm_f32_2:
; CHECK: // %bb.0:
; CHECK-NEXT: mov w8, #1109917696
; CHECK-NEXT: mov z0.s, w8
; CHECK-NEXT: ret
%out = tail call <vscale x 2 x float> @llvm.aarch64.sve.dup.x.nxv2f32(float 4.200000e+01)
ret <vscale x 2 x float> %out
}
define <vscale x 4 x float> @dup_fmov_imm_f32_4() {
; CHECK-LABEL: dup_fmov_imm_f32_4:
; CHECK: // %bb.0:
; CHECK-NEXT: mov w8, #1109917696
; CHECK-NEXT: mov z0.s, w8
; CHECK-NEXT: ret
%out = tail call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 4.200000e+01)
ret <vscale x 4 x float> %out
}
define <vscale x 2 x double> @dup_fmov_imm_f64_2() {
; CHECK-LABEL: dup_fmov_imm_f64_2:
; CHECK: // %bb.0:
; CHECK-NEXT: mov x8, #4631107791820423168
; CHECK-NEXT: mov z0.d, x8
; CHECK-NEXT: ret
%out = tail call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 4.200000e+01)
ret <vscale x 2 x double> %out
}
declare <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8( i8)
declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
declare <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32)
declare <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64)
declare <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half)
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat)
declare <vscale x 2 x float> @llvm.aarch64.sve.dup.x.nxv2f32(float)
declare <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float)
declare <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double)
; +bf16 is required for the bfloat version.
attributes #0 = { "target-features"="+sve,+bf16" }