Google Git
Sign in
llvm / llvm-project / 18308e171b5b1dd99627a4d88c7d6c5ff21b8c96 / . / llvm / test / CodeGen / AArch64 / fptosi-sat-vector.ll
blob: 964c806ad092b0d7db258fa89cbcf4daa5b878c3 [file] [log] [blame]
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=aarch64 | FileCheck %s --check-prefixes=CHECK,CHECK-CVT
; RUN: llc < %s -mtriple=aarch64 -mattr=+fullfp16 | FileCheck %s --check-prefixes=CHECK,CHECK-FP16
;
; Float to signed 32-bit -- Vector size variation
;
declare <1 x i32> @llvm.fptosi.sat.v1f32.v1i32 (<1 x float>)
declare <2 x i32> @llvm.fptosi.sat.v2f32.v2i32 (<2 x float>)
declare <3 x i32> @llvm.fptosi.sat.v3f32.v3i32 (<3 x float>)
declare <4 x i32> @llvm.fptosi.sat.v4f32.v4i32 (<4 x float>)
declare <5 x i32> @llvm.fptosi.sat.v5f32.v5i32 (<5 x float>)
declare <6 x i32> @llvm.fptosi.sat.v6f32.v6i32 (<6 x float>)
declare <7 x i32> @llvm.fptosi.sat.v7f32.v7i32 (<7 x float>)
declare <8 x i32> @llvm.fptosi.sat.v8f32.v8i32 (<8 x float>)
define <1 x i32> @test_signed_v1f32_v1i32(<1 x float> %f) {
; CHECK-LABEL: test_signed_v1f32_v1i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: ret
%x = call <1 x i32> @llvm.fptosi.sat.v1f32.v1i32(<1 x float> %f)
ret <1 x i32> %x
}
define <2 x i32> @test_signed_v2f32_v2i32(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptosi.sat.v2f32.v2i32(<2 x float> %f)
ret <2 x i32> %x
}
define <3 x i32> @test_signed_v3f32_v3i32(<3 x float> %f) {
; CHECK-LABEL: test_signed_v3f32_v3i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <3 x i32> @llvm.fptosi.sat.v3f32.v3i32(<3 x float> %f)
ret <3 x i32> %x
}
define <4 x i32> @test_signed_v4f32_v4i32(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptosi.sat.v4f32.v4i32(<4 x float> %f)
ret <4 x i32> %x
}
define <5 x i32> @test_signed_v5f32_v5i32(<5 x float> %f) {
; CHECK-LABEL: test_signed_v5f32_v5i32:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-NEXT: // kill: def $s1 killed $s1 def $q1
; CHECK-NEXT: // kill: def $s2 killed $s2 def $q2
; CHECK-NEXT: // kill: def $s3 killed $s3 def $q3
; CHECK-NEXT: // kill: def $s4 killed $s4 def $q4
; CHECK-NEXT: mov v0.s[1], v1.s[0]
; CHECK-NEXT: fcvtzs v4.4s, v4.4s
; CHECK-NEXT: mov v0.s[2], v2.s[0]
; CHECK-NEXT: fmov w4, s4
; CHECK-NEXT: mov v0.s[3], v3.s[0]
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: mov w1, v0.s[1]
; CHECK-NEXT: mov w2, v0.s[2]
; CHECK-NEXT: mov w3, v0.s[3]
; CHECK-NEXT: fmov w0, s0
; CHECK-NEXT: ret
%x = call <5 x i32> @llvm.fptosi.sat.v5f32.v5i32(<5 x float> %f)
ret <5 x i32> %x
}
define <6 x i32> @test_signed_v6f32_v6i32(<6 x float> %f) {
; CHECK-LABEL: test_signed_v6f32_v6i32:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-NEXT: // kill: def $s1 killed $s1 def $q1
; CHECK-NEXT: // kill: def $s2 killed $s2 def $q2
; CHECK-NEXT: // kill: def $s4 killed $s4 def $q4
; CHECK-NEXT: // kill: def $s5 killed $s5 def $q5
; CHECK-NEXT: // kill: def $s3 killed $s3 def $q3
; CHECK-NEXT: mov v0.s[1], v1.s[0]
; CHECK-NEXT: mov v4.s[1], v5.s[0]
; CHECK-NEXT: mov v0.s[2], v2.s[0]
; CHECK-NEXT: fcvtzs v1.4s, v4.4s
; CHECK-NEXT: mov v0.s[3], v3.s[0]
; CHECK-NEXT: mov w5, v1.s[1]
; CHECK-NEXT: fmov w4, s1
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: mov w1, v0.s[1]
; CHECK-NEXT: mov w2, v0.s[2]
; CHECK-NEXT: mov w3, v0.s[3]
; CHECK-NEXT: fmov w0, s0
; CHECK-NEXT: ret
%x = call <6 x i32> @llvm.fptosi.sat.v6f32.v6i32(<6 x float> %f)
ret <6 x i32> %x
}
define <7 x i32> @test_signed_v7f32_v7i32(<7 x float> %f) {
; CHECK-LABEL: test_signed_v7f32_v7i32:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-NEXT: // kill: def $s1 killed $s1 def $q1
; CHECK-NEXT: // kill: def $s4 killed $s4 def $q4
; CHECK-NEXT: // kill: def $s5 killed $s5 def $q5
; CHECK-NEXT: // kill: def $s2 killed $s2 def $q2
; CHECK-NEXT: // kill: def $s6 killed $s6 def $q6
; CHECK-NEXT: // kill: def $s3 killed $s3 def $q3
; CHECK-NEXT: mov v0.s[1], v1.s[0]
; CHECK-NEXT: mov v4.s[1], v5.s[0]
; CHECK-NEXT: mov v0.s[2], v2.s[0]
; CHECK-NEXT: mov v4.s[2], v6.s[0]
; CHECK-NEXT: mov v0.s[3], v3.s[0]
; CHECK-NEXT: fcvtzs v1.4s, v4.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: mov w5, v1.s[1]
; CHECK-NEXT: mov w6, v1.s[2]
; CHECK-NEXT: fmov w4, s1
; CHECK-NEXT: mov w1, v0.s[1]
; CHECK-NEXT: mov w2, v0.s[2]
; CHECK-NEXT: mov w3, v0.s[3]
; CHECK-NEXT: fmov w0, s0
; CHECK-NEXT: ret
%x = call <7 x i32> @llvm.fptosi.sat.v7f32.v7i32(<7 x float> %f)
ret <7 x i32> %x
}
define <8 x i32> @test_signed_v8f32_v8i32(<8 x float> %f) {
; CHECK-LABEL: test_signed_v8f32_v8i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: fcvtzs v1.4s, v1.4s
; CHECK-NEXT: ret
%x = call <8 x i32> @llvm.fptosi.sat.v8f32.v8i32(<8 x float> %f)
ret <8 x i32> %x
}
;
; Double to signed 32-bit -- Vector size variation
;
declare <1 x i32> @llvm.fptosi.sat.v1f64.v1i32 (<1 x double>)
declare <2 x i32> @llvm.fptosi.sat.v2f64.v2i32 (<2 x double>)
declare <3 x i32> @llvm.fptosi.sat.v3f64.v3i32 (<3 x double>)
declare <4 x i32> @llvm.fptosi.sat.v4f64.v4i32 (<4 x double>)
declare <5 x i32> @llvm.fptosi.sat.v5f64.v5i32 (<5 x double>)
declare <6 x i32> @llvm.fptosi.sat.v6f64.v6i32 (<6 x double>)
define <1 x i32> @test_signed_v1f64_v1i32(<1 x double> %f) {
; CHECK-LABEL: test_signed_v1f64_v1i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs w8, d0
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: ret
%x = call <1 x i32> @llvm.fptosi.sat.v1f64.v1i32(<1 x double> %f)
ret <1 x i32> %x
}
define <2 x i32> @test_signed_v2f64_v2i32(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i32:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: fcvtzs w8, d0
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: fcvtzs w8, d1
; CHECK-NEXT: mov v0.s[1], w8
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptosi.sat.v2f64.v2i32(<2 x double> %f)
ret <2 x i32> %x
}
define <3 x i32> @test_signed_v3f64_v3i32(<3 x double> %f) {
; CHECK-LABEL: test_signed_v3f64_v3i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs w8, d0
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: fcvtzs w8, d1
; CHECK-NEXT: mov v0.s[1], w8
; CHECK-NEXT: fcvtzs w8, d2
; CHECK-NEXT: mov v0.s[2], w8
; CHECK-NEXT: fcvtzs w8, d0
; CHECK-NEXT: mov v0.s[3], w8
; CHECK-NEXT: ret
%x = call <3 x i32> @llvm.fptosi.sat.v3f64.v3i32(<3 x double> %f)
ret <3 x i32> %x
}
define <4 x i32> @test_signed_v4f64_v4i32(<4 x double> %f) {
; CHECK-LABEL: test_signed_v4f64_v4i32:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d2, v0.d[1]
; CHECK-NEXT: fcvtzs w8, d0
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: fcvtzs w8, d2
; CHECK-NEXT: mov v0.s[1], w8
; CHECK-NEXT: fcvtzs w8, d1
; CHECK-NEXT: mov d1, v1.d[1]
; CHECK-NEXT: mov v0.s[2], w8
; CHECK-NEXT: fcvtzs w8, d1
; CHECK-NEXT: mov v0.s[3], w8
; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptosi.sat.v4f64.v4i32(<4 x double> %f)
ret <4 x i32> %x
}
define <5 x i32> @test_signed_v5f64_v5i32(<5 x double> %f) {
; CHECK-LABEL: test_signed_v5f64_v5i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs w0, d0
; CHECK-NEXT: fcvtzs w1, d1
; CHECK-NEXT: fcvtzs w2, d2
; CHECK-NEXT: fcvtzs w3, d3
; CHECK-NEXT: fcvtzs w4, d4
; CHECK-NEXT: ret
%x = call <5 x i32> @llvm.fptosi.sat.v5f64.v5i32(<5 x double> %f)
ret <5 x i32> %x
}
define <6 x i32> @test_signed_v6f64_v6i32(<6 x double> %f) {
; CHECK-LABEL: test_signed_v6f64_v6i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs w0, d0
; CHECK-NEXT: fcvtzs w1, d1
; CHECK-NEXT: fcvtzs w2, d2
; CHECK-NEXT: fcvtzs w3, d3
; CHECK-NEXT: fcvtzs w4, d4
; CHECK-NEXT: fcvtzs w5, d5
; CHECK-NEXT: ret
%x = call <6 x i32> @llvm.fptosi.sat.v6f64.v6i32(<6 x double> %f)
ret <6 x i32> %x
}
;
; FP128 to signed 32-bit -- Vector size variation
;
declare <1 x i32> @llvm.fptosi.sat.v1f128.v1i32 (<1 x fp128>)
declare <2 x i32> @llvm.fptosi.sat.v2f128.v2i32 (<2 x fp128>)
declare <3 x i32> @llvm.fptosi.sat.v3f128.v3i32 (<3 x fp128>)
declare <4 x i32> @llvm.fptosi.sat.v4f128.v4i32 (<4 x fp128>)
define <1 x i32> @test_signed_v1f128_v1i32(<1 x fp128> %f) {
; CHECK-LABEL: test_signed_v1f128_v1i32:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #32
; CHECK-NEXT: stp x30, x19, [sp, #16] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 32
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w30, -16
; CHECK-NEXT: adrp x8, .LCPI14_0
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI14_0]
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: mov w8, #-2147483648
; CHECK-NEXT: csel w19, w8, w0, lt
; CHECK-NEXT: adrp x8, .LCPI14_1
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI14_1]
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov w8, #2147483647
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w8, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: ldp x30, x19, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: add sp, sp, #32
; CHECK-NEXT: ret
%x = call <1 x i32> @llvm.fptosi.sat.v1f128.v1i32(<1 x fp128> %f)
ret <1 x i32> %x
}
define <2 x i32> @test_signed_v2f128_v2i32(<2 x fp128> %f) {
; CHECK-LABEL: test_signed_v2f128_v2i32:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #112
; CHECK-NEXT: str x30, [sp, #64] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 112
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -48
; CHECK-NEXT: mov v2.16b, v1.16b
; CHECK-NEXT: adrp x8, .LCPI15_0
; CHECK-NEXT: stp q1, q0, [sp, #32] // 32-byte Folded Spill
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI15_0]
; CHECK-NEXT: mov v0.16b, v2.16b
; CHECK-NEXT: str q1, [sp, #16] // 16-byte Folded Spill
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: adrp x8, .LCPI15_1
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: mov w20, #-2147483648
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI15_1]
; CHECK-NEXT: str q1, [sp] // 16-byte Folded Spill
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: mov w21, #2147483647
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csel w22, wzr, w19, ne
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: ldr q1, [sp] // 16-byte Folded Reload
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr x30, [sp, #64] // 8-byte Folded Reload
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: ldp x20, x19, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w22
; CHECK-NEXT: ldp x22, x21, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: add sp, sp, #112
; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptosi.sat.v2f128.v2i32(<2 x fp128> %f)
ret <2 x i32> %x
}
define <3 x i32> @test_signed_v3f128_v3i32(<3 x fp128> %f) {
; CHECK-LABEL: test_signed_v3f128_v3i32:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #128
; CHECK-NEXT: str x30, [sp, #80] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #112] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 128
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -48
; CHECK-NEXT: stp q0, q2, [sp, #48] // 32-byte Folded Spill
; CHECK-NEXT: adrp x8, .LCPI16_0
; CHECK-NEXT: mov v2.16b, v1.16b
; CHECK-NEXT: str q1, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI16_0]
; CHECK-NEXT: mov v0.16b, v2.16b
; CHECK-NEXT: str q1, [sp, #16] // 16-byte Folded Spill
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: adrp x8, .LCPI16_1
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: mov w20, #-2147483648
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI16_1]
; CHECK-NEXT: str q1, [sp] // 16-byte Folded Spill
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: mov w21, #2147483647
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csel w22, wzr, w19, ne
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: ldr q1, [sp] // 16-byte Folded Reload
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w22
; CHECK-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: ldr q1, [sp] // 16-byte Folded Reload
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: ldr x30, [sp, #80] // 8-byte Folded Reload
; CHECK-NEXT: ldp x20, x19, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: mov v0.s[2], w8
; CHECK-NEXT: add sp, sp, #128
; CHECK-NEXT: ret
%x = call <3 x i32> @llvm.fptosi.sat.v3f128.v3i32(<3 x fp128> %f)
ret <3 x i32> %x
}
define <4 x i32> @test_signed_v4f128_v4i32(<4 x fp128> %f) {
; CHECK-LABEL: test_signed_v4f128_v4i32:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #144
; CHECK-NEXT: str x30, [sp, #96] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #112] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #128] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 144
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -48
; CHECK-NEXT: stp q2, q3, [sp, #64] // 32-byte Folded Spill
; CHECK-NEXT: adrp x8, .LCPI17_0
; CHECK-NEXT: mov v2.16b, v1.16b
; CHECK-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: str q1, [sp] // 16-byte Folded Spill
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI17_0]
; CHECK-NEXT: mov v0.16b, v2.16b
; CHECK-NEXT: str q1, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: adrp x8, .LCPI17_1
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: mov w20, #-2147483648
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: ldr q1, [x8, :lo12:.LCPI17_1]
; CHECK-NEXT: str q1, [sp, #16] // 16-byte Folded Spill
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov w21, #2147483647
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: ldp q1, q0, [sp, #32] // 32-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w22, wzr, w19, ne
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q1, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w22
; CHECK-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: ldp q1, q0, [sp, #32] // 32-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: mov v0.s[2], w8
; CHECK-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: ldr q0, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: bl __getf2
; CHECK-NEXT: ldr q0, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: mov w19, w0
; CHECK-NEXT: bl __fixtfsi
; CHECK-NEXT: ldr q0, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csel w19, w20, w0, lt
; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: ldr q0, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csel w19, w21, w19, gt
; CHECK-NEXT: mov v1.16b, v0.16b
; CHECK-NEXT: bl __unordtf2
; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel w8, wzr, w19, ne
; CHECK-NEXT: ldr x30, [sp, #96] // 8-byte Folded Reload
; CHECK-NEXT: ldp x20, x19, [sp, #128] // 16-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: mov v0.s[3], w8
; CHECK-NEXT: add sp, sp, #144
; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptosi.sat.v4f128.v4i32(<4 x fp128> %f)
ret <4 x i32> %x
}
;
; FP16 to signed 32-bit -- Vector size variation
;
declare <1 x i32> @llvm.fptosi.sat.v1f16.v1i32 (<1 x half>)
declare <2 x i32> @llvm.fptosi.sat.v2f16.v2i32 (<2 x half>)
declare <3 x i32> @llvm.fptosi.sat.v3f16.v3i32 (<3 x half>)
declare <4 x i32> @llvm.fptosi.sat.v4f16.v4i32 (<4 x half>)
declare <5 x i32> @llvm.fptosi.sat.v5f16.v5i32 (<5 x half>)
declare <6 x i32> @llvm.fptosi.sat.v6f16.v6i32 (<6 x half>)
declare <7 x i32> @llvm.fptosi.sat.v7f16.v7i32 (<7 x half>)
declare <8 x i32> @llvm.fptosi.sat.v8f16.v8i32 (<8 x half>)
define <1 x i32> @test_signed_v1f16_v1i32(<1 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v1f16_v1i32:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvt s0, h0
; CHECK-CVT-NEXT: fcvtzs w8, s0
; CHECK-CVT-NEXT: fmov s0, w8
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v1f16_v1i32:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: fcvtzs w8, h0
; CHECK-FP16-NEXT: fmov s0, w8
; CHECK-FP16-NEXT: ret
%x = call <1 x i32> @llvm.fptosi.sat.v1f16.v1i32(<1 x half> %f)
ret <1 x i32> %x
}
define <2 x i32> @test_signed_v2f16_v2i32(<2 x half> %f) {
; CHECK-LABEL: test_signed_v2f16_v2i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptosi.sat.v2f16.v2i32(<2 x half> %f)
ret <2 x i32> %x
}
define <3 x i32> @test_signed_v3f16_v3i32(<3 x half> %f) {
; CHECK-LABEL: test_signed_v3f16_v3i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <3 x i32> @llvm.fptosi.sat.v3f16.v3i32(<3 x half> %f)
ret <3 x i32> %x
}
define <4 x i32> @test_signed_v4f16_v4i32(<4 x half> %f) {
; CHECK-LABEL: test_signed_v4f16_v4i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptosi.sat.v4f16.v4i32(<4 x half> %f)
ret <4 x i32> %x
}
define <5 x i32> @test_signed_v5f16_v5i32(<5 x half> %f) {
; CHECK-LABEL: test_signed_v5f16_v5i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v1.4s, v0.4h
; CHECK-NEXT: fcvtl2 v0.4s, v0.8h
; CHECK-NEXT: fcvtzs v1.4s, v1.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: mov w1, v1.s[1]
; CHECK-NEXT: mov w2, v1.s[2]
; CHECK-NEXT: mov w3, v1.s[3]
; CHECK-NEXT: fmov w0, s1
; CHECK-NEXT: fmov w4, s0
; CHECK-NEXT: ret
%x = call <5 x i32> @llvm.fptosi.sat.v5f16.v5i32(<5 x half> %f)
ret <5 x i32> %x
}
define <6 x i32> @test_signed_v6f16_v6i32(<6 x half> %f) {
; CHECK-LABEL: test_signed_v6f16_v6i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v1.4s, v0.4h
; CHECK-NEXT: fcvtl2 v0.4s, v0.8h
; CHECK-NEXT: fcvtzs v1.4s, v1.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: mov w1, v1.s[1]
; CHECK-NEXT: mov w2, v1.s[2]
; CHECK-NEXT: mov w3, v1.s[3]
; CHECK-NEXT: mov w5, v0.s[1]
; CHECK-NEXT: fmov w0, s1
; CHECK-NEXT: fmov w4, s0
; CHECK-NEXT: ret
%x = call <6 x i32> @llvm.fptosi.sat.v6f16.v6i32(<6 x half> %f)
ret <6 x i32> %x
}
define <7 x i32> @test_signed_v7f16_v7i32(<7 x half> %f) {
; CHECK-LABEL: test_signed_v7f16_v7i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v1.4s, v0.4h
; CHECK-NEXT: fcvtl2 v0.4s, v0.8h
; CHECK-NEXT: fcvtzs v1.4s, v1.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: mov w1, v1.s[1]
; CHECK-NEXT: mov w2, v1.s[2]
; CHECK-NEXT: mov w3, v1.s[3]
; CHECK-NEXT: mov w5, v0.s[1]
; CHECK-NEXT: mov w6, v0.s[2]
; CHECK-NEXT: fmov w0, s1
; CHECK-NEXT: fmov w4, s0
; CHECK-NEXT: ret
%x = call <7 x i32> @llvm.fptosi.sat.v7f16.v7i32(<7 x half> %f)
ret <7 x i32> %x
}
define <8 x i32> @test_signed_v8f16_v8i32(<8 x half> %f) {
; CHECK-LABEL: test_signed_v8f16_v8i32:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl2 v1.4s, v0.8h
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: fcvtzs v1.4s, v1.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <8 x i32> @llvm.fptosi.sat.v8f16.v8i32(<8 x half> %f)
ret <8 x i32> %x
}
;
; 2-Vector float to signed integer -- result size variation
;
declare <2 x i1> @llvm.fptosi.sat.v2f32.v2i1 (<2 x float>)
declare <2 x i8> @llvm.fptosi.sat.v2f32.v2i8 (<2 x float>)
declare <2 x i13> @llvm.fptosi.sat.v2f32.v2i13 (<2 x float>)
declare <2 x i16> @llvm.fptosi.sat.v2f32.v2i16 (<2 x float>)
declare <2 x i19> @llvm.fptosi.sat.v2f32.v2i19 (<2 x float>)
declare <2 x i50> @llvm.fptosi.sat.v2f32.v2i50 (<2 x float>)
declare <2 x i64> @llvm.fptosi.sat.v2f32.v2i64 (<2 x float>)
declare <2 x i100> @llvm.fptosi.sat.v2f32.v2i100(<2 x float>)
declare <2 x i128> @llvm.fptosi.sat.v2f32.v2i128(<2 x float>)
define <2 x i1> @test_signed_v2f32_v2i1(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i1:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.2d, #0000000000000000
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: smin v0.2s, v0.2s, v1.2s
; CHECK-NEXT: movi v1.2d, #0xffffffffffffffff
; CHECK-NEXT: smax v0.2s, v0.2s, v1.2s
; CHECK-NEXT: ret
%x = call <2 x i1> @llvm.fptosi.sat.v2f32.v2i1(<2 x float> %f)
ret <2 x i1> %x
}
define <2 x i8> @test_signed_v2f32_v2i8(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i8:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.2s, #127
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: smin v0.2s, v0.2s, v1.2s
; CHECK-NEXT: mvni v1.2s, #127
; CHECK-NEXT: smax v0.2s, v0.2s, v1.2s
; CHECK-NEXT: ret
%x = call <2 x i8> @llvm.fptosi.sat.v2f32.v2i8(<2 x float> %f)
ret <2 x i8> %x
}
define <2 x i13> @test_signed_v2f32_v2i13(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i13:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.2s, #15, msl #8
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: smin v0.2s, v0.2s, v1.2s
; CHECK-NEXT: mvni v1.2s, #15, msl #8
; CHECK-NEXT: smax v0.2s, v0.2s, v1.2s
; CHECK-NEXT: ret
%x = call <2 x i13> @llvm.fptosi.sat.v2f32.v2i13(<2 x float> %f)
ret <2 x i13> %x
}
define <2 x i16> @test_signed_v2f32_v2i16(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i16:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.2s, #127, msl #8
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: smin v0.2s, v0.2s, v1.2s
; CHECK-NEXT: mvni v1.2s, #127, msl #8
; CHECK-NEXT: smax v0.2s, v0.2s, v1.2s
; CHECK-NEXT: ret
%x = call <2 x i16> @llvm.fptosi.sat.v2f32.v2i16(<2 x float> %f)
ret <2 x i16> %x
}
define <2 x i19> @test_signed_v2f32_v2i19(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i19:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.2s, #3, msl #16
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: smin v0.2s, v0.2s, v1.2s
; CHECK-NEXT: mvni v1.2s, #3, msl #16
; CHECK-NEXT: smax v0.2s, v0.2s, v1.2s
; CHECK-NEXT: ret
%x = call <2 x i19> @llvm.fptosi.sat.v2f32.v2i19(<2 x float> %f)
ret <2 x i19> %x
}
define <2 x i32> @test_signed_v2f32_v2i32_duplicate(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i32_duplicate:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.2s, v0.2s
; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptosi.sat.v2f32.v2i32(<2 x float> %f)
ret <2 x i32> %x
}
define <2 x i50> @test_signed_v2f32_v2i50(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i50:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: mov s1, v0.s[1]
; CHECK-NEXT: mov x8, #562949953421311
; CHECK-NEXT: fcvtzs x10, s0
; CHECK-NEXT: mov x11, #-562949953421312
; CHECK-NEXT: fcvtzs x9, s1
; CHECK-NEXT: cmp x9, x8
; CHECK-NEXT: csel x9, x9, x8, lt
; CHECK-NEXT: cmp x9, x11
; CHECK-NEXT: csel x9, x9, x11, gt
; CHECK-NEXT: cmp x10, x8
; CHECK-NEXT: csel x8, x10, x8, lt
; CHECK-NEXT: cmp x8, x11
; CHECK-NEXT: csel x8, x8, x11, gt
; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x9
; CHECK-NEXT: ret
%x = call <2 x i50> @llvm.fptosi.sat.v2f32.v2i50(<2 x float> %f)
ret <2 x i50> %x
}
define <2 x i64> @test_signed_v2f32_v2i64(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i64:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: mov s1, v0.s[1]
; CHECK-NEXT: fcvtzs x8, s0
; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: fcvtzs x8, s1
; CHECK-NEXT: mov v0.d[1], x8
; CHECK-NEXT: ret
%x = call <2 x i64> @llvm.fptosi.sat.v2f32.v2i64(<2 x float> %f)
ret <2 x i64> %x
}
define <2 x i100> @test_signed_v2f32_v2i100(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i100:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #80
; CHECK-NEXT: str d10, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #24] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #40] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 80
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -40
; CHECK-NEXT: .cfi_offset b8, -48
; CHECK-NEXT: .cfi_offset b9, -56
; CHECK-NEXT: .cfi_offset b10, -64
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: mov s8, v0.s[1]
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-251658240
; CHECK-NEXT: mov x21, #-34359738368
; CHECK-NEXT: mov x22, #34359738367
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: mov w8, #1895825407
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload
; CHECK-NEXT: ldp x20, x19, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: ldp d9, d8, [sp, #24] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, x21, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s0, s10
; CHECK-NEXT: ldr d10, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x22, x8, gt
; CHECK-NEXT: fcmp s0, s0
; CHECK-NEXT: ldp x22, x21, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #80
; CHECK-NEXT: ret
%x = call <2 x i100> @llvm.fptosi.sat.v2f32.v2i100(<2 x float> %f)
ret <2 x i100> %x
}
define <2 x i128> @test_signed_v2f32_v2i128(<2 x float> %f) {
; CHECK-LABEL: test_signed_v2f32_v2i128:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #80
; CHECK-NEXT: str d10, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #24] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #40] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 80
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -40
; CHECK-NEXT: .cfi_offset b8, -48
; CHECK-NEXT: .cfi_offset b9, -56
; CHECK-NEXT: .cfi_offset b10, -64
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: mov s8, v0.s[1]
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-16777216
; CHECK-NEXT: mov x21, #-9223372036854775808
; CHECK-NEXT: mov x22, #9223372036854775807
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: mov w8, #2130706431
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload
; CHECK-NEXT: ldp x20, x19, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: ldp d9, d8, [sp, #24] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, x21, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s0, s10
; CHECK-NEXT: ldr d10, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x22, x8, gt
; CHECK-NEXT: fcmp s0, s0
; CHECK-NEXT: ldp x22, x21, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #80
; CHECK-NEXT: ret
%x = call <2 x i128> @llvm.fptosi.sat.v2f32.v2i128(<2 x float> %f)
ret <2 x i128> %x
}
;
; 4-Vector float to signed integer -- result size variation
;
declare <4 x i1> @llvm.fptosi.sat.v4f32.v4i1 (<4 x float>)
declare <4 x i8> @llvm.fptosi.sat.v4f32.v4i8 (<4 x float>)
declare <4 x i13> @llvm.fptosi.sat.v4f32.v4i13 (<4 x float>)
declare <4 x i16> @llvm.fptosi.sat.v4f32.v4i16 (<4 x float>)
declare <4 x i19> @llvm.fptosi.sat.v4f32.v4i19 (<4 x float>)
declare <4 x i50> @llvm.fptosi.sat.v4f32.v4i50 (<4 x float>)
declare <4 x i64> @llvm.fptosi.sat.v4f32.v4i64 (<4 x float>)
declare <4 x i100> @llvm.fptosi.sat.v4f32.v4i100(<4 x float>)
declare <4 x i128> @llvm.fptosi.sat.v4f32.v4i128(<4 x float>)
define <4 x i1> @test_signed_v4f32_v4i1(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i1:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.2d, #0000000000000000
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT: movi v1.2d, #0xffffffffffffffff
; CHECK-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-NEXT: xtn v0.4h, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i1> @llvm.fptosi.sat.v4f32.v4i1(<4 x float> %f)
ret <4 x i1> %x
}
define <4 x i8> @test_signed_v4f32_v4i8(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i8:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.4s, #127
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT: mvni v1.4s, #127
; CHECK-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-NEXT: xtn v0.4h, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i8> @llvm.fptosi.sat.v4f32.v4i8(<4 x float> %f)
ret <4 x i8> %x
}
define <4 x i13> @test_signed_v4f32_v4i13(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i13:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.4s, #15, msl #8
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT: mvni v1.4s, #15, msl #8
; CHECK-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-NEXT: xtn v0.4h, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i13> @llvm.fptosi.sat.v4f32.v4i13(<4 x float> %f)
ret <4 x i13> %x
}
define <4 x i16> @test_signed_v4f32_v4i16(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i16:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: sqxtn v0.4h, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i16> @llvm.fptosi.sat.v4f32.v4i16(<4 x float> %f)
ret <4 x i16> %x
}
define <4 x i19> @test_signed_v4f32_v4i19(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i19:
; CHECK: // %bb.0:
; CHECK-NEXT: movi v1.4s, #3, msl #16
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT: mvni v1.4s, #3, msl #16
; CHECK-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-NEXT: ret
%x = call <4 x i19> @llvm.fptosi.sat.v4f32.v4i19(<4 x float> %f)
ret <4 x i19> %x
}
define <4 x i32> @test_signed_v4f32_v4i32_duplicate(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i32_duplicate:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptosi.sat.v4f32.v4i32(<4 x float> %f)
ret <4 x i32> %x
}
define <4 x i50> @test_signed_v4f32_v4i50(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i50:
; CHECK: // %bb.0:
; CHECK-NEXT: ext v1.16b, v0.16b, v0.16b, #8
; CHECK-NEXT: mov x9, #562949953421311
; CHECK-NEXT: mov x10, #-562949953421312
; CHECK-NEXT: fcvtzs x12, s0
; CHECK-NEXT: mov s2, v1.s[1]
; CHECK-NEXT: fcvtzs x8, s1
; CHECK-NEXT: mov s1, v0.s[1]
; CHECK-NEXT: cmp x8, x9
; CHECK-NEXT: fcvtzs x11, s2
; CHECK-NEXT: csel x8, x8, x9, lt
; CHECK-NEXT: cmp x8, x10
; CHECK-NEXT: csel x2, x8, x10, gt
; CHECK-NEXT: cmp x11, x9
; CHECK-NEXT: csel x8, x11, x9, lt
; CHECK-NEXT: fcvtzs x11, s1
; CHECK-NEXT: cmp x8, x10
; CHECK-NEXT: csel x3, x8, x10, gt
; CHECK-NEXT: cmp x12, x9
; CHECK-NEXT: csel x8, x12, x9, lt
; CHECK-NEXT: cmp x8, x10
; CHECK-NEXT: csel x0, x8, x10, gt
; CHECK-NEXT: cmp x11, x9
; CHECK-NEXT: csel x8, x11, x9, lt
; CHECK-NEXT: cmp x8, x10
; CHECK-NEXT: csel x1, x8, x10, gt
; CHECK-NEXT: ret
%x = call <4 x i50> @llvm.fptosi.sat.v4f32.v4i50(<4 x float> %f)
ret <4 x i50> %x
}
define <4 x i64> @test_signed_v4f32_v4i64(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i64:
; CHECK: // %bb.0:
; CHECK-NEXT: ext v1.16b, v0.16b, v0.16b, #8
; CHECK-NEXT: mov s3, v0.s[1]
; CHECK-NEXT: fcvtzs x9, s0
; CHECK-NEXT: mov s2, v1.s[1]
; CHECK-NEXT: fcvtzs x8, s1
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: fcvtzs x9, s3
; CHECK-NEXT: fmov d1, x8
; CHECK-NEXT: fcvtzs x8, s2
; CHECK-NEXT: mov v0.d[1], x9
; CHECK-NEXT: mov v1.d[1], x8
; CHECK-NEXT: ret
%x = call <4 x i64> @llvm.fptosi.sat.v4f32.v4i64(<4 x float> %f)
ret <4 x i64> %x
}
define <4 x i100> @test_signed_v4f32_v4i100(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i100:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #128
; CHECK-NEXT: str d10, [sp, #32] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #40] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #56] // 8-byte Folded Spill
; CHECK-NEXT: stp x26, x25, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: stp x24, x23, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #112] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 128
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w23, -40
; CHECK-NEXT: .cfi_offset w24, -48
; CHECK-NEXT: .cfi_offset w25, -56
; CHECK-NEXT: .cfi_offset w26, -64
; CHECK-NEXT: .cfi_offset w30, -72
; CHECK-NEXT: .cfi_offset b8, -80
; CHECK-NEXT: .cfi_offset b9, -88
; CHECK-NEXT: .cfi_offset b10, -96
; CHECK-NEXT: mov s8, v0.s[1]
; CHECK-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-251658240
; CHECK-NEXT: mov x25, #-34359738368
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: mov x26, #34359738367
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: mov w8, #1895825407
; CHECK-NEXT: ext v0.16b, v0.16b, v0.16b, #8
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: mov s8, v0.s[1]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s0, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s0, s0
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: csel x21, xzr, x8, vs
; CHECK-NEXT: csel x22, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: csel x23, xzr, x8, vs
; CHECK-NEXT: csel x24, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: mov x4, x21
; CHECK-NEXT: mov x5, x22
; CHECK-NEXT: mov x6, x23
; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: mov x7, x24
; CHECK-NEXT: ldp x20, x19, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s0, s10
; CHECK-NEXT: ldr x30, [sp, #56] // 8-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x26, x8, gt
; CHECK-NEXT: fcmp s0, s0
; CHECK-NEXT: ldr d10, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: ldp x24, x23, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: ldp x26, x25, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: ldp d9, d8, [sp, #40] // 16-byte Folded Reload
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #128
; CHECK-NEXT: ret
%x = call <4 x i100> @llvm.fptosi.sat.v4f32.v4i100(<4 x float> %f)
ret <4 x i100> %x
}
define <4 x i128> @test_signed_v4f32_v4i128(<4 x float> %f) {
; CHECK-LABEL: test_signed_v4f32_v4i128:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #128
; CHECK-NEXT: str d10, [sp, #32] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #40] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #56] // 8-byte Folded Spill
; CHECK-NEXT: stp x26, x25, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: stp x24, x23, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #112] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 128
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w23, -40
; CHECK-NEXT: .cfi_offset w24, -48
; CHECK-NEXT: .cfi_offset w25, -56
; CHECK-NEXT: .cfi_offset w26, -64
; CHECK-NEXT: .cfi_offset w30, -72
; CHECK-NEXT: .cfi_offset b8, -80
; CHECK-NEXT: .cfi_offset b9, -88
; CHECK-NEXT: .cfi_offset b10, -96
; CHECK-NEXT: mov s8, v0.s[1]
; CHECK-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-16777216
; CHECK-NEXT: mov x25, #-9223372036854775808
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: mov x26, #9223372036854775807
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: mov w8, #2130706431
; CHECK-NEXT: ext v0.16b, v0.16b, v0.16b, #8
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: mov s8, v0.s[1]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s0, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s0, s0
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: csel x21, xzr, x8, vs
; CHECK-NEXT: csel x22, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: csel x23, xzr, x8, vs
; CHECK-NEXT: csel x24, xzr, x9, vs
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: mov x4, x21
; CHECK-NEXT: mov x5, x22
; CHECK-NEXT: mov x6, x23
; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: mov x7, x24
; CHECK-NEXT: ldp x20, x19, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s0, s10
; CHECK-NEXT: ldr x30, [sp, #56] // 8-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x26, x8, gt
; CHECK-NEXT: fcmp s0, s0
; CHECK-NEXT: ldr d10, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: ldp x24, x23, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: ldp x26, x25, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: ldp d9, d8, [sp, #40] // 16-byte Folded Reload
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #128
; CHECK-NEXT: ret
%x = call <4 x i128> @llvm.fptosi.sat.v4f32.v4i128(<4 x float> %f)
ret <4 x i128> %x
}
;
; 2-Vector double to signed integer -- result size variation
;
declare <2 x i1> @llvm.fptosi.sat.v2f64.v2i1 (<2 x double>)
declare <2 x i8> @llvm.fptosi.sat.v2f64.v2i8 (<2 x double>)
declare <2 x i13> @llvm.fptosi.sat.v2f64.v2i13 (<2 x double>)
declare <2 x i16> @llvm.fptosi.sat.v2f64.v2i16 (<2 x double>)
declare <2 x i19> @llvm.fptosi.sat.v2f64.v2i19 (<2 x double>)
declare <2 x i50> @llvm.fptosi.sat.v2f64.v2i50 (<2 x double>)
declare <2 x i64> @llvm.fptosi.sat.v2f64.v2i64 (<2 x double>)
declare <2 x i100> @llvm.fptosi.sat.v2f64.v2i100(<2 x double>)
declare <2 x i128> @llvm.fptosi.sat.v2f64.v2i128(<2 x double>)
define <2 x i1> @test_signed_v2f64_v2i1(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i1:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: fcvtzs w9, d0
; CHECK-NEXT: fcvtzs w8, d1
; CHECK-NEXT: cmp w8, #0
; CHECK-NEXT: csel w8, w8, wzr, lt
; CHECK-NEXT: cmp w8, #0
; CHECK-NEXT: csinv w8, w8, wzr, ge
; CHECK-NEXT: cmp w9, #0
; CHECK-NEXT: csel w9, w9, wzr, lt
; CHECK-NEXT: cmp w9, #0
; CHECK-NEXT: csinv w9, w9, wzr, ge
; CHECK-NEXT: fmov s0, w9
; CHECK-NEXT: mov v0.s[1], w8
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i1> @llvm.fptosi.sat.v2f64.v2i1(<2 x double> %f)
ret <2 x i1> %x
}
define <2 x i8> @test_signed_v2f64_v2i8(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i8:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: mov w8, #127
; CHECK-NEXT: fcvtzs w10, d0
; CHECK-NEXT: mov w11, #-128
; CHECK-NEXT: fcvtzs w9, d1
; CHECK-NEXT: cmp w9, #127
; CHECK-NEXT: csel w9, w9, w8, lt
; CHECK-NEXT: cmn w9, #128
; CHECK-NEXT: csel w9, w9, w11, gt
; CHECK-NEXT: cmp w10, #127
; CHECK-NEXT: csel w8, w10, w8, lt
; CHECK-NEXT: cmn w8, #128
; CHECK-NEXT: csel w8, w8, w11, gt
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w9
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i8> @llvm.fptosi.sat.v2f64.v2i8(<2 x double> %f)
ret <2 x i8> %x
}
define <2 x i13> @test_signed_v2f64_v2i13(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i13:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: mov w8, #4095
; CHECK-NEXT: fcvtzs w10, d0
; CHECK-NEXT: mov w11, #-4096
; CHECK-NEXT: fcvtzs w9, d1
; CHECK-NEXT: cmp w9, #4095
; CHECK-NEXT: csel w9, w9, w8, lt
; CHECK-NEXT: cmn w9, #1, lsl #12 // =4096
; CHECK-NEXT: csel w9, w9, w11, gt
; CHECK-NEXT: cmp w10, #4095
; CHECK-NEXT: csel w8, w10, w8, lt
; CHECK-NEXT: cmn w8, #1, lsl #12 // =4096
; CHECK-NEXT: csel w8, w8, w11, gt
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w9
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i13> @llvm.fptosi.sat.v2f64.v2i13(<2 x double> %f)
ret <2 x i13> %x
}
define <2 x i16> @test_signed_v2f64_v2i16(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i16:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: mov w8, #32767
; CHECK-NEXT: fcvtzs w10, d0
; CHECK-NEXT: mov w11, #-32768
; CHECK-NEXT: fcvtzs w9, d1
; CHECK-NEXT: cmp w9, w8
; CHECK-NEXT: csel w9, w9, w8, lt
; CHECK-NEXT: cmn w9, #8, lsl #12 // =32768
; CHECK-NEXT: csel w9, w9, w11, gt
; CHECK-NEXT: cmp w10, w8
; CHECK-NEXT: csel w8, w10, w8, lt
; CHECK-NEXT: cmn w8, #8, lsl #12 // =32768
; CHECK-NEXT: csel w8, w8, w11, gt
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w9
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i16> @llvm.fptosi.sat.v2f64.v2i16(<2 x double> %f)
ret <2 x i16> %x
}
define <2 x i19> @test_signed_v2f64_v2i19(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i19:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: mov w8, #262143
; CHECK-NEXT: fcvtzs w10, d0
; CHECK-NEXT: mov w11, #-262144
; CHECK-NEXT: fcvtzs w9, d1
; CHECK-NEXT: cmp w9, w8
; CHECK-NEXT: csel w9, w9, w8, lt
; CHECK-NEXT: cmn w9, #64, lsl #12 // =262144
; CHECK-NEXT: csel w9, w9, w11, gt
; CHECK-NEXT: cmp w10, w8
; CHECK-NEXT: csel w8, w10, w8, lt
; CHECK-NEXT: cmn w8, #64, lsl #12 // =262144
; CHECK-NEXT: csel w8, w8, w11, gt
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w9
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i19> @llvm.fptosi.sat.v2f64.v2i19(<2 x double> %f)
ret <2 x i19> %x
}
define <2 x i32> @test_signed_v2f64_v2i32_duplicate(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i32_duplicate:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: fcvtzs w8, d0
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: fcvtzs w8, d1
; CHECK-NEXT: mov v0.s[1], w8
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptosi.sat.v2f64.v2i32(<2 x double> %f)
ret <2 x i32> %x
}
define <2 x i50> @test_signed_v2f64_v2i50(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i50:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: mov x8, #562949953421311
; CHECK-NEXT: fcvtzs x10, d0
; CHECK-NEXT: mov x11, #-562949953421312
; CHECK-NEXT: fcvtzs x9, d1
; CHECK-NEXT: cmp x9, x8
; CHECK-NEXT: csel x9, x9, x8, lt
; CHECK-NEXT: cmp x9, x11
; CHECK-NEXT: csel x9, x9, x11, gt
; CHECK-NEXT: cmp x10, x8
; CHECK-NEXT: csel x8, x10, x8, lt
; CHECK-NEXT: cmp x8, x11
; CHECK-NEXT: csel x8, x8, x11, gt
; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x9
; CHECK-NEXT: ret
%x = call <2 x i50> @llvm.fptosi.sat.v2f64.v2i50(<2 x double> %f)
ret <2 x i50> %x
}
define <2 x i64> @test_signed_v2f64_v2i64(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i64:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtzs v0.2d, v0.2d
; CHECK-NEXT: ret
%x = call <2 x i64> @llvm.fptosi.sat.v2f64.v2i64(<2 x double> %f)
ret <2 x i64> %x
}
define <2 x i100> @test_signed_v2f64_v2i100(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i100:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #80
; CHECK-NEXT: str d10, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #24] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #40] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 80
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -40
; CHECK-NEXT: .cfi_offset b8, -48
; CHECK-NEXT: .cfi_offset b9, -56
; CHECK-NEXT: .cfi_offset b10, -64
; CHECK-NEXT: mov d8, v0.d[1]
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: fmov d0, d8
; CHECK-NEXT: bl __fixdfti
; CHECK-NEXT: mov x8, #-4170333254945079296
; CHECK-NEXT: mov x21, #-34359738368
; CHECK-NEXT: mov x22, #34359738367
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: fmov d9, x8
; CHECK-NEXT: mov x8, #5053038781909696511
; CHECK-NEXT: fcmp d8, d9
; CHECK-NEXT: fmov d10, x8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp d8, d10
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp d8, d8
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: bl __fixdfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload
; CHECK-NEXT: ldp x20, x19, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fcmp d0, d9
; CHECK-NEXT: ldp d9, d8, [sp, #24] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, x21, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp d0, d10
; CHECK-NEXT: ldr d10, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x22, x8, gt
; CHECK-NEXT: fcmp d0, d0
; CHECK-NEXT: ldp x22, x21, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #80
; CHECK-NEXT: ret
%x = call <2 x i100> @llvm.fptosi.sat.v2f64.v2i100(<2 x double> %f)
ret <2 x i100> %x
}
define <2 x i128> @test_signed_v2f64_v2i128(<2 x double> %f) {
; CHECK-LABEL: test_signed_v2f64_v2i128:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #80
; CHECK-NEXT: str d10, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #24] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #40] // 8-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 80
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w30, -40
; CHECK-NEXT: .cfi_offset b8, -48
; CHECK-NEXT: .cfi_offset b9, -56
; CHECK-NEXT: .cfi_offset b10, -64
; CHECK-NEXT: mov d8, v0.d[1]
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: fmov d0, d8
; CHECK-NEXT: bl __fixdfti
; CHECK-NEXT: mov x8, #-4044232465378705408
; CHECK-NEXT: mov x21, #-9223372036854775808
; CHECK-NEXT: mov x22, #9223372036854775807
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: fmov d9, x8
; CHECK-NEXT: mov x8, #5179139571476070399
; CHECK-NEXT: fcmp d8, d9
; CHECK-NEXT: fmov d10, x8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp d8, d10
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp d8, d8
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: bl __fixdfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload
; CHECK-NEXT: ldp x20, x19, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fcmp d0, d9
; CHECK-NEXT: ldp d9, d8, [sp, #24] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, x21, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp d0, d10
; CHECK-NEXT: ldr d10, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x22, x8, gt
; CHECK-NEXT: fcmp d0, d0
; CHECK-NEXT: ldp x22, x21, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #80
; CHECK-NEXT: ret
%x = call <2 x i128> @llvm.fptosi.sat.v2f64.v2i128(<2 x double> %f)
ret <2 x i128> %x
}
;
; 4-Vector half to signed integer -- result size variation
;
declare <4 x i1> @llvm.fptosi.sat.v4f16.v4i1 (<4 x half>)
declare <4 x i8> @llvm.fptosi.sat.v4f16.v4i8 (<4 x half>)
declare <4 x i13> @llvm.fptosi.sat.v4f16.v4i13 (<4 x half>)
declare <4 x i16> @llvm.fptosi.sat.v4f16.v4i16 (<4 x half>)
declare <4 x i19> @llvm.fptosi.sat.v4f16.v4i19 (<4 x half>)
declare <4 x i50> @llvm.fptosi.sat.v4f16.v4i50 (<4 x half>)
declare <4 x i64> @llvm.fptosi.sat.v4f16.v4i64 (<4 x half>)
declare <4 x i100> @llvm.fptosi.sat.v4f16.v4i100(<4 x half>)
declare <4 x i128> @llvm.fptosi.sat.v4f16.v4i128(<4 x half>)
define <4 x i1> @test_signed_v4f16_v4i1(<4 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v4f16_v4i1:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: movi v1.2d, #0000000000000000
; CHECK-CVT-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-CVT-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-CVT-NEXT: movi v1.2d, #0xffffffffffffffff
; CHECK-CVT-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-CVT-NEXT: xtn v0.4h, v0.4s
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v4f16_v4i1:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: movi v1.2d, #0000000000000000
; CHECK-FP16-NEXT: fcvtzs v0.4h, v0.4h
; CHECK-FP16-NEXT: smin v0.4h, v0.4h, v1.4h
; CHECK-FP16-NEXT: movi v1.2d, #0xffffffffffffffff
; CHECK-FP16-NEXT: smax v0.4h, v0.4h, v1.4h
; CHECK-FP16-NEXT: ret
%x = call <4 x i1> @llvm.fptosi.sat.v4f16.v4i1(<4 x half> %f)
ret <4 x i1> %x
}
define <4 x i8> @test_signed_v4f16_v4i8(<4 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v4f16_v4i8:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: movi v1.4s, #127
; CHECK-CVT-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-CVT-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-CVT-NEXT: mvni v1.4s, #127
; CHECK-CVT-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-CVT-NEXT: xtn v0.4h, v0.4s
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v4f16_v4i8:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: movi v1.4h, #127
; CHECK-FP16-NEXT: fcvtzs v0.4h, v0.4h
; CHECK-FP16-NEXT: smin v0.4h, v0.4h, v1.4h
; CHECK-FP16-NEXT: mvni v1.4h, #127
; CHECK-FP16-NEXT: smax v0.4h, v0.4h, v1.4h
; CHECK-FP16-NEXT: ret
%x = call <4 x i8> @llvm.fptosi.sat.v4f16.v4i8(<4 x half> %f)
ret <4 x i8> %x
}
define <4 x i13> @test_signed_v4f16_v4i13(<4 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v4f16_v4i13:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: movi v1.4s, #15, msl #8
; CHECK-CVT-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-CVT-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-CVT-NEXT: mvni v1.4s, #15, msl #8
; CHECK-CVT-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-CVT-NEXT: xtn v0.4h, v0.4s
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v4f16_v4i13:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: mvni v1.4h, #240, lsl #8
; CHECK-FP16-NEXT: fcvtzs v0.4h, v0.4h
; CHECK-FP16-NEXT: smin v0.4h, v0.4h, v1.4h
; CHECK-FP16-NEXT: movi v1.4h, #240, lsl #8
; CHECK-FP16-NEXT: smax v0.4h, v0.4h, v1.4h
; CHECK-FP16-NEXT: ret
%x = call <4 x i13> @llvm.fptosi.sat.v4f16.v4i13(<4 x half> %f)
ret <4 x i13> %x
}
define <4 x i16> @test_signed_v4f16_v4i16(<4 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v4f16_v4i16:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-CVT-NEXT: sqxtn v0.4h, v0.4s
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v4f16_v4i16:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: fcvtzs v0.4h, v0.4h
; CHECK-FP16-NEXT: ret
%x = call <4 x i16> @llvm.fptosi.sat.v4f16.v4i16(<4 x half> %f)
ret <4 x i16> %x
}
define <4 x i19> @test_signed_v4f16_v4i19(<4 x half> %f) {
; CHECK-LABEL: test_signed_v4f16_v4i19:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: movi v1.4s, #3, msl #16
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT: mvni v1.4s, #3, msl #16
; CHECK-NEXT: smax v0.4s, v0.4s, v1.4s
; CHECK-NEXT: ret
%x = call <4 x i19> @llvm.fptosi.sat.v4f16.v4i19(<4 x half> %f)
ret <4 x i19> %x
}
define <4 x i32> @test_signed_v4f16_v4i32_duplicate(<4 x half> %f) {
; CHECK-LABEL: test_signed_v4f16_v4i32_duplicate:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptosi.sat.v4f16.v4i32(<4 x half> %f)
ret <4 x i32> %x
}
define <4 x i50> @test_signed_v4f16_v4i50(<4 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v4f16_v4i50:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-CVT-NEXT: mov h1, v0.h[1]
; CHECK-CVT-NEXT: fcvt s2, h0
; CHECK-CVT-NEXT: mov h3, v0.h[2]
; CHECK-CVT-NEXT: mov h0, v0.h[3]
; CHECK-CVT-NEXT: mov x8, #562949953421311
; CHECK-CVT-NEXT: mov x11, #-562949953421312
; CHECK-CVT-NEXT: fcvt s1, h1
; CHECK-CVT-NEXT: fcvtzs x9, s2
; CHECK-CVT-NEXT: fcvt s0, h0
; CHECK-CVT-NEXT: cmp x9, x8
; CHECK-CVT-NEXT: fcvtzs x10, s1
; CHECK-CVT-NEXT: fcvt s1, h3
; CHECK-CVT-NEXT: csel x9, x9, x8, lt
; CHECK-CVT-NEXT: cmp x9, x11
; CHECK-CVT-NEXT: csel x0, x9, x11, gt
; CHECK-CVT-NEXT: cmp x10, x8
; CHECK-CVT-NEXT: fcvtzs x9, s1
; CHECK-CVT-NEXT: csel x10, x10, x8, lt
; CHECK-CVT-NEXT: cmp x10, x11
; CHECK-CVT-NEXT: csel x1, x10, x11, gt
; CHECK-CVT-NEXT: fcvtzs x10, s0
; CHECK-CVT-NEXT: cmp x9, x8
; CHECK-CVT-NEXT: csel x9, x9, x8, lt
; CHECK-CVT-NEXT: cmp x9, x11
; CHECK-CVT-NEXT: csel x2, x9, x11, gt
; CHECK-CVT-NEXT: cmp x10, x8
; CHECK-CVT-NEXT: csel x8, x10, x8, lt
; CHECK-CVT-NEXT: cmp x8, x11
; CHECK-CVT-NEXT: csel x3, x8, x11, gt
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v4f16_v4i50:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-FP16-NEXT: mov h1, v0.h[1]
; CHECK-FP16-NEXT: fcvtzs x9, h0
; CHECK-FP16-NEXT: mov x8, #562949953421311
; CHECK-FP16-NEXT: mov x11, #-562949953421312
; CHECK-FP16-NEXT: cmp x9, x8
; CHECK-FP16-NEXT: fcvtzs x10, h1
; CHECK-FP16-NEXT: mov h1, v0.h[2]
; CHECK-FP16-NEXT: csel x9, x9, x8, lt
; CHECK-FP16-NEXT: mov h0, v0.h[3]
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: csel x0, x9, x11, gt
; CHECK-FP16-NEXT: cmp x10, x8
; CHECK-FP16-NEXT: fcvtzs x9, h1
; CHECK-FP16-NEXT: csel x10, x10, x8, lt
; CHECK-FP16-NEXT: cmp x10, x11
; CHECK-FP16-NEXT: csel x1, x10, x11, gt
; CHECK-FP16-NEXT: fcvtzs x10, h0
; CHECK-FP16-NEXT: cmp x9, x8
; CHECK-FP16-NEXT: csel x9, x9, x8, lt
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: csel x2, x9, x11, gt
; CHECK-FP16-NEXT: cmp x10, x8
; CHECK-FP16-NEXT: csel x8, x10, x8, lt
; CHECK-FP16-NEXT: cmp x8, x11
; CHECK-FP16-NEXT: csel x3, x8, x11, gt
; CHECK-FP16-NEXT: ret
%x = call <4 x i50> @llvm.fptosi.sat.v4f16.v4i50(<4 x half> %f)
ret <4 x i50> %x
}
define <4 x i64> @test_signed_v4f16_v4i64(<4 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v4f16_v4i64:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-CVT-NEXT: mov h1, v0.h[2]
; CHECK-CVT-NEXT: mov h2, v0.h[1]
; CHECK-CVT-NEXT: fcvt s3, h0
; CHECK-CVT-NEXT: mov h0, v0.h[3]
; CHECK-CVT-NEXT: fcvt s1, h1
; CHECK-CVT-NEXT: fcvt s2, h2
; CHECK-CVT-NEXT: fcvtzs x8, s3
; CHECK-CVT-NEXT: fcvt s3, h0
; CHECK-CVT-NEXT: fcvtzs x9, s1
; CHECK-CVT-NEXT: fmov d0, x8
; CHECK-CVT-NEXT: fcvtzs x8, s2
; CHECK-CVT-NEXT: fmov d1, x9
; CHECK-CVT-NEXT: fcvtzs x9, s3
; CHECK-CVT-NEXT: mov v0.d[1], x8
; CHECK-CVT-NEXT: mov v1.d[1], x9
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v4f16_v4i64:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-FP16-NEXT: mov h1, v0.h[2]
; CHECK-FP16-NEXT: mov h2, v0.h[1]
; CHECK-FP16-NEXT: mov h3, v0.h[3]
; CHECK-FP16-NEXT: fcvtzs x8, h0
; CHECK-FP16-NEXT: fcvtzs x9, h1
; CHECK-FP16-NEXT: fmov d0, x8
; CHECK-FP16-NEXT: fcvtzs x8, h2
; CHECK-FP16-NEXT: fmov d1, x9
; CHECK-FP16-NEXT: fcvtzs x9, h3
; CHECK-FP16-NEXT: mov v0.d[1], x8
; CHECK-FP16-NEXT: mov v1.d[1], x9
; CHECK-FP16-NEXT: ret
%x = call <4 x i64> @llvm.fptosi.sat.v4f16.v4i64(<4 x half> %f)
ret <4 x i64> %x
}
define <4 x i100> @test_signed_v4f16_v4i100(<4 x half> %f) {
; CHECK-LABEL: test_signed_v4f16_v4i100:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #112
; CHECK-NEXT: str d10, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #24] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #40] // 8-byte Folded Spill
; CHECK-NEXT: stp x26, x25, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: stp x24, x23, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 112
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w23, -40
; CHECK-NEXT: .cfi_offset w24, -48
; CHECK-NEXT: .cfi_offset w25, -56
; CHECK-NEXT: .cfi_offset w26, -64
; CHECK-NEXT: .cfi_offset w30, -72
; CHECK-NEXT: .cfi_offset b8, -80
; CHECK-NEXT: .cfi_offset b9, -88
; CHECK-NEXT: .cfi_offset b10, -96
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: mov h1, v0.h[1]
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: fcvt s8, h1
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-251658240
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov x25, #-34359738368
; CHECK-NEXT: mov x26, #34359738367
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: mov w8, #1895825407
; CHECK-NEXT: mov h0, v0.h[2]
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: mov h0, v0.h[3]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x21, xzr, x8, vs
; CHECK-NEXT: csel x22, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x23, xzr, x8, vs
; CHECK-NEXT: csel x24, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: mov x4, x21
; CHECK-NEXT: mov x5, x22
; CHECK-NEXT: mov x6, x23
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov x7, x24
; CHECK-NEXT: ldp x20, x19, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x26, x8, gt
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: ldp x24, x23, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: ldr d10, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: ldp x26, x25, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: ldp d9, d8, [sp, #24] // 16-byte Folded Reload
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #112
; CHECK-NEXT: ret
%x = call <4 x i100> @llvm.fptosi.sat.v4f16.v4i100(<4 x half> %f)
ret <4 x i100> %x
}
define <4 x i128> @test_signed_v4f16_v4i128(<4 x half> %f) {
; CHECK-LABEL: test_signed_v4f16_v4i128:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #112
; CHECK-NEXT: str d10, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #24] // 16-byte Folded Spill
; CHECK-NEXT: str x30, [sp, #40] // 8-byte Folded Spill
; CHECK-NEXT: stp x26, x25, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: stp x24, x23, [sp, #64] // 16-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 112
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w23, -40
; CHECK-NEXT: .cfi_offset w24, -48
; CHECK-NEXT: .cfi_offset w25, -56
; CHECK-NEXT: .cfi_offset w26, -64
; CHECK-NEXT: .cfi_offset w30, -72
; CHECK-NEXT: .cfi_offset b8, -80
; CHECK-NEXT: .cfi_offset b9, -88
; CHECK-NEXT: .cfi_offset b10, -96
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: mov h1, v0.h[1]
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: fcvt s8, h1
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-16777216
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: mov x25, #-9223372036854775808
; CHECK-NEXT: mov x26, #9223372036854775807
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: mov w8, #2130706431
; CHECK-NEXT: mov h0, v0.h[2]
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x19, xzr, x8, vs
; CHECK-NEXT: csel x20, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: mov h0, v0.h[3]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x21, xzr, x8, vs
; CHECK-NEXT: csel x22, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: csel x9, x26, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x23, xzr, x8, vs
; CHECK-NEXT: csel x24, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: mov x2, x19
; CHECK-NEXT: mov x3, x20
; CHECK-NEXT: mov x4, x21
; CHECK-NEXT: mov x5, x22
; CHECK-NEXT: mov x6, x23
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov x7, x24
; CHECK-NEXT: ldp x20, x19, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x26, x8, gt
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: ldr x30, [sp, #40] // 8-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x1, xzr, x8, vs
; CHECK-NEXT: ldp x24, x23, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x9
; CHECK-NEXT: ldr d10, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: ldp x26, x25, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: ldp d9, d8, [sp, #24] // 16-byte Folded Reload
; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: add sp, sp, #112
; CHECK-NEXT: ret
%x = call <4 x i128> @llvm.fptosi.sat.v4f16.v4i128(<4 x half> %f)
ret <4 x i128> %x
}
;
; 8-Vector half to signed integer -- result size variation
;
declare <8 x i1> @llvm.fptosi.sat.v8f16.v8i1 (<8 x half>)
declare <8 x i8> @llvm.fptosi.sat.v8f16.v8i8 (<8 x half>)
declare <8 x i13> @llvm.fptosi.sat.v8f16.v8i13 (<8 x half>)
declare <8 x i16> @llvm.fptosi.sat.v8f16.v8i16 (<8 x half>)
declare <8 x i19> @llvm.fptosi.sat.v8f16.v8i19 (<8 x half>)
declare <8 x i50> @llvm.fptosi.sat.v8f16.v8i50 (<8 x half>)
declare <8 x i64> @llvm.fptosi.sat.v8f16.v8i64 (<8 x half>)
declare <8 x i100> @llvm.fptosi.sat.v8f16.v8i100(<8 x half>)
declare <8 x i128> @llvm.fptosi.sat.v8f16.v8i128(<8 x half>)
define <8 x i1> @test_signed_v8f16_v8i1(<8 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v8f16_v8i1:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl2 v1.4s, v0.8h
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: mov s2, v1.s[1]
; CHECK-CVT-NEXT: fcvtzs w9, s1
; CHECK-CVT-NEXT: fcvtzs w13, s0
; CHECK-CVT-NEXT: fcvtzs w8, s2
; CHECK-CVT-NEXT: mov s2, v1.s[2]
; CHECK-CVT-NEXT: mov s1, v1.s[3]
; CHECK-CVT-NEXT: cmp w8, #0
; CHECK-CVT-NEXT: csel w8, w8, wzr, lt
; CHECK-CVT-NEXT: fcvtzs w10, s2
; CHECK-CVT-NEXT: cmp w8, #0
; CHECK-CVT-NEXT: mov s2, v0.s[1]
; CHECK-CVT-NEXT: csinv w8, w8, wzr, ge
; CHECK-CVT-NEXT: cmp w9, #0
; CHECK-CVT-NEXT: csel w9, w9, wzr, lt
; CHECK-CVT-NEXT: fcvtzs w11, s1
; CHECK-CVT-NEXT: cmp w9, #0
; CHECK-CVT-NEXT: mov s1, v0.s[2]
; CHECK-CVT-NEXT: csinv w9, w9, wzr, ge
; CHECK-CVT-NEXT: cmp w10, #0
; CHECK-CVT-NEXT: csel w10, w10, wzr, lt
; CHECK-CVT-NEXT: fcvtzs w12, s2
; CHECK-CVT-NEXT: cmp w10, #0
; CHECK-CVT-NEXT: mov s0, v0.s[3]
; CHECK-CVT-NEXT: csinv w10, w10, wzr, ge
; CHECK-CVT-NEXT: cmp w11, #0
; CHECK-CVT-NEXT: csel w11, w11, wzr, lt
; CHECK-CVT-NEXT: fcvtzs w14, s1
; CHECK-CVT-NEXT: cmp w11, #0
; CHECK-CVT-NEXT: fmov s2, w9
; CHECK-CVT-NEXT: csinv w11, w11, wzr, ge
; CHECK-CVT-NEXT: cmp w12, #0
; CHECK-CVT-NEXT: csel w12, w12, wzr, lt
; CHECK-CVT-NEXT: cmp w12, #0
; CHECK-CVT-NEXT: csinv w12, w12, wzr, ge
; CHECK-CVT-NEXT: cmp w13, #0
; CHECK-CVT-NEXT: csel w13, w13, wzr, lt
; CHECK-CVT-NEXT: cmp w13, #0
; CHECK-CVT-NEXT: csinv w13, w13, wzr, ge
; CHECK-CVT-NEXT: cmp w14, #0
; CHECK-CVT-NEXT: csel w9, w14, wzr, lt
; CHECK-CVT-NEXT: cmp w9, #0
; CHECK-CVT-NEXT: fmov s1, w13
; CHECK-CVT-NEXT: fcvtzs w13, s0
; CHECK-CVT-NEXT: csinv w9, w9, wzr, ge
; CHECK-CVT-NEXT: mov v2.s[1], w8
; CHECK-CVT-NEXT: mov v1.s[1], w12
; CHECK-CVT-NEXT: cmp w13, #0
; CHECK-CVT-NEXT: csel w8, w13, wzr, lt
; CHECK-CVT-NEXT: cmp w8, #0
; CHECK-CVT-NEXT: csinv w8, w8, wzr, ge
; CHECK-CVT-NEXT: mov v1.s[2], w9
; CHECK-CVT-NEXT: mov v2.s[2], w10
; CHECK-CVT-NEXT: mov v1.s[3], w8
; CHECK-CVT-NEXT: mov v2.s[3], w11
; CHECK-CVT-NEXT: xtn v0.4h, v1.4s
; CHECK-CVT-NEXT: xtn2 v0.8h, v2.4s
; CHECK-CVT-NEXT: xtn v0.8b, v0.8h
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v8f16_v8i1:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: movi v1.2d, #0000000000000000
; CHECK-FP16-NEXT: fcvtzs v0.8h, v0.8h
; CHECK-FP16-NEXT: smin v0.8h, v0.8h, v1.8h
; CHECK-FP16-NEXT: movi v1.2d, #0xffffffffffffffff
; CHECK-FP16-NEXT: smax v0.8h, v0.8h, v1.8h
; CHECK-FP16-NEXT: xtn v0.8b, v0.8h
; CHECK-FP16-NEXT: ret
%x = call <8 x i1> @llvm.fptosi.sat.v8f16.v8i1(<8 x half> %f)
ret <8 x i1> %x
}
define <8 x i8> @test_signed_v8f16_v8i8(<8 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v8f16_v8i8:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl2 v1.4s, v0.8h
; CHECK-CVT-NEXT: mov w8, #127
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: mov w10, #-128
; CHECK-CVT-NEXT: mov s2, v1.s[1]
; CHECK-CVT-NEXT: fcvtzs w11, s1
; CHECK-CVT-NEXT: fcvtzs w15, s0
; CHECK-CVT-NEXT: fcvtzs w9, s2
; CHECK-CVT-NEXT: mov s2, v1.s[2]
; CHECK-CVT-NEXT: mov s1, v1.s[3]
; CHECK-CVT-NEXT: cmp w9, #127
; CHECK-CVT-NEXT: csel w9, w9, w8, lt
; CHECK-CVT-NEXT: fcvtzs w12, s2
; CHECK-CVT-NEXT: cmn w9, #128
; CHECK-CVT-NEXT: mov s2, v0.s[1]
; CHECK-CVT-NEXT: csel w9, w9, w10, gt
; CHECK-CVT-NEXT: cmp w11, #127
; CHECK-CVT-NEXT: csel w11, w11, w8, lt
; CHECK-CVT-NEXT: fcvtzs w13, s1
; CHECK-CVT-NEXT: cmn w11, #128
; CHECK-CVT-NEXT: mov s1, v0.s[2]
; CHECK-CVT-NEXT: csel w11, w11, w10, gt
; CHECK-CVT-NEXT: cmp w12, #127
; CHECK-CVT-NEXT: csel w12, w12, w8, lt
; CHECK-CVT-NEXT: fcvtzs w14, s2
; CHECK-CVT-NEXT: cmn w12, #128
; CHECK-CVT-NEXT: mov s0, v0.s[3]
; CHECK-CVT-NEXT: csel w12, w12, w10, gt
; CHECK-CVT-NEXT: cmp w13, #127
; CHECK-CVT-NEXT: csel w13, w13, w8, lt
; CHECK-CVT-NEXT: fcvtzs w16, s1
; CHECK-CVT-NEXT: cmn w13, #128
; CHECK-CVT-NEXT: fmov s2, w11
; CHECK-CVT-NEXT: csel w13, w13, w10, gt
; CHECK-CVT-NEXT: cmp w14, #127
; CHECK-CVT-NEXT: csel w14, w14, w8, lt
; CHECK-CVT-NEXT: cmn w14, #128
; CHECK-CVT-NEXT: csel w14, w14, w10, gt
; CHECK-CVT-NEXT: cmp w15, #127
; CHECK-CVT-NEXT: csel w15, w15, w8, lt
; CHECK-CVT-NEXT: cmn w15, #128
; CHECK-CVT-NEXT: csel w15, w15, w10, gt
; CHECK-CVT-NEXT: cmp w16, #127
; CHECK-CVT-NEXT: csel w11, w16, w8, lt
; CHECK-CVT-NEXT: cmn w11, #128
; CHECK-CVT-NEXT: fmov s1, w15
; CHECK-CVT-NEXT: fcvtzs w15, s0
; CHECK-CVT-NEXT: csel w11, w11, w10, gt
; CHECK-CVT-NEXT: mov v2.s[1], w9
; CHECK-CVT-NEXT: mov v1.s[1], w14
; CHECK-CVT-NEXT: cmp w15, #127
; CHECK-CVT-NEXT: csel w8, w15, w8, lt
; CHECK-CVT-NEXT: cmn w8, #128
; CHECK-CVT-NEXT: csel w8, w8, w10, gt
; CHECK-CVT-NEXT: mov v1.s[2], w11
; CHECK-CVT-NEXT: mov v2.s[2], w12
; CHECK-CVT-NEXT: mov v1.s[3], w8
; CHECK-CVT-NEXT: mov v2.s[3], w13
; CHECK-CVT-NEXT: xtn v0.4h, v1.4s
; CHECK-CVT-NEXT: xtn2 v0.8h, v2.4s
; CHECK-CVT-NEXT: xtn v0.8b, v0.8h
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v8f16_v8i8:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: fcvtzs v0.8h, v0.8h
; CHECK-FP16-NEXT: sqxtn v0.8b, v0.8h
; CHECK-FP16-NEXT: ret
%x = call <8 x i8> @llvm.fptosi.sat.v8f16.v8i8(<8 x half> %f)
ret <8 x i8> %x
}
define <8 x i13> @test_signed_v8f16_v8i13(<8 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v8f16_v8i13:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl2 v1.4s, v0.8h
; CHECK-CVT-NEXT: mov w8, #4095
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: mov w10, #-4096
; CHECK-CVT-NEXT: mov s2, v1.s[1]
; CHECK-CVT-NEXT: fcvtzs w11, s1
; CHECK-CVT-NEXT: fcvtzs w15, s0
; CHECK-CVT-NEXT: fcvtzs w9, s2
; CHECK-CVT-NEXT: mov s2, v1.s[2]
; CHECK-CVT-NEXT: mov s1, v1.s[3]
; CHECK-CVT-NEXT: cmp w9, #4095
; CHECK-CVT-NEXT: csel w9, w9, w8, lt
; CHECK-CVT-NEXT: fcvtzs w12, s2
; CHECK-CVT-NEXT: cmn w9, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: mov s2, v0.s[1]
; CHECK-CVT-NEXT: csel w9, w9, w10, gt
; CHECK-CVT-NEXT: cmp w11, #4095
; CHECK-CVT-NEXT: csel w11, w11, w8, lt
; CHECK-CVT-NEXT: fcvtzs w13, s1
; CHECK-CVT-NEXT: cmn w11, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: mov s1, v0.s[2]
; CHECK-CVT-NEXT: csel w11, w11, w10, gt
; CHECK-CVT-NEXT: cmp w12, #4095
; CHECK-CVT-NEXT: csel w12, w12, w8, lt
; CHECK-CVT-NEXT: fcvtzs w14, s2
; CHECK-CVT-NEXT: cmn w12, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: mov s0, v0.s[3]
; CHECK-CVT-NEXT: csel w12, w12, w10, gt
; CHECK-CVT-NEXT: cmp w13, #4095
; CHECK-CVT-NEXT: csel w13, w13, w8, lt
; CHECK-CVT-NEXT: fcvtzs w16, s1
; CHECK-CVT-NEXT: cmn w13, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: fmov s2, w11
; CHECK-CVT-NEXT: csel w13, w13, w10, gt
; CHECK-CVT-NEXT: cmp w14, #4095
; CHECK-CVT-NEXT: csel w14, w14, w8, lt
; CHECK-CVT-NEXT: cmn w14, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: csel w14, w14, w10, gt
; CHECK-CVT-NEXT: cmp w15, #4095
; CHECK-CVT-NEXT: csel w15, w15, w8, lt
; CHECK-CVT-NEXT: cmn w15, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: csel w15, w15, w10, gt
; CHECK-CVT-NEXT: cmp w16, #4095
; CHECK-CVT-NEXT: csel w11, w16, w8, lt
; CHECK-CVT-NEXT: cmn w11, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: fmov s1, w15
; CHECK-CVT-NEXT: fcvtzs w15, s0
; CHECK-CVT-NEXT: csel w11, w11, w10, gt
; CHECK-CVT-NEXT: mov v2.s[1], w9
; CHECK-CVT-NEXT: mov v1.s[1], w14
; CHECK-CVT-NEXT: cmp w15, #4095
; CHECK-CVT-NEXT: csel w8, w15, w8, lt
; CHECK-CVT-NEXT: cmn w8, #1, lsl #12 // =4096
; CHECK-CVT-NEXT: csel w8, w8, w10, gt
; CHECK-CVT-NEXT: mov v1.s[2], w11
; CHECK-CVT-NEXT: mov v2.s[2], w12
; CHECK-CVT-NEXT: mov v1.s[3], w8
; CHECK-CVT-NEXT: mov v2.s[3], w13
; CHECK-CVT-NEXT: xtn v0.4h, v1.4s
; CHECK-CVT-NEXT: xtn2 v0.8h, v2.4s
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v8f16_v8i13:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: mvni v1.8h, #240, lsl #8
; CHECK-FP16-NEXT: fcvtzs v0.8h, v0.8h
; CHECK-FP16-NEXT: smin v0.8h, v0.8h, v1.8h
; CHECK-FP16-NEXT: movi v1.8h, #240, lsl #8
; CHECK-FP16-NEXT: smax v0.8h, v0.8h, v1.8h
; CHECK-FP16-NEXT: ret
%x = call <8 x i13> @llvm.fptosi.sat.v8f16.v8i13(<8 x half> %f)
ret <8 x i13> %x
}
define <8 x i16> @test_signed_v8f16_v8i16(<8 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v8f16_v8i16:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: fcvtl2 v1.4s, v0.8h
; CHECK-CVT-NEXT: mov w8, #32767
; CHECK-CVT-NEXT: fcvtl v0.4s, v0.4h
; CHECK-CVT-NEXT: mov w10, #-32768
; CHECK-CVT-NEXT: mov s2, v1.s[1]
; CHECK-CVT-NEXT: fcvtzs w11, s1
; CHECK-CVT-NEXT: fcvtzs w15, s0
; CHECK-CVT-NEXT: fcvtzs w9, s2
; CHECK-CVT-NEXT: mov s2, v1.s[2]
; CHECK-CVT-NEXT: mov s1, v1.s[3]
; CHECK-CVT-NEXT: cmp w9, w8
; CHECK-CVT-NEXT: csel w9, w9, w8, lt
; CHECK-CVT-NEXT: fcvtzs w12, s2
; CHECK-CVT-NEXT: cmn w9, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: mov s2, v0.s[1]
; CHECK-CVT-NEXT: csel w9, w9, w10, gt
; CHECK-CVT-NEXT: cmp w11, w8
; CHECK-CVT-NEXT: csel w11, w11, w8, lt
; CHECK-CVT-NEXT: fcvtzs w13, s1
; CHECK-CVT-NEXT: cmn w11, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: mov s1, v0.s[2]
; CHECK-CVT-NEXT: csel w11, w11, w10, gt
; CHECK-CVT-NEXT: cmp w12, w8
; CHECK-CVT-NEXT: csel w12, w12, w8, lt
; CHECK-CVT-NEXT: fcvtzs w14, s2
; CHECK-CVT-NEXT: cmn w12, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: mov s0, v0.s[3]
; CHECK-CVT-NEXT: csel w12, w12, w10, gt
; CHECK-CVT-NEXT: cmp w13, w8
; CHECK-CVT-NEXT: csel w13, w13, w8, lt
; CHECK-CVT-NEXT: fcvtzs w16, s1
; CHECK-CVT-NEXT: cmn w13, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: fmov s2, w11
; CHECK-CVT-NEXT: csel w13, w13, w10, gt
; CHECK-CVT-NEXT: cmp w14, w8
; CHECK-CVT-NEXT: csel w14, w14, w8, lt
; CHECK-CVT-NEXT: cmn w14, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: csel w14, w14, w10, gt
; CHECK-CVT-NEXT: cmp w15, w8
; CHECK-CVT-NEXT: csel w15, w15, w8, lt
; CHECK-CVT-NEXT: cmn w15, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: csel w15, w15, w10, gt
; CHECK-CVT-NEXT: cmp w16, w8
; CHECK-CVT-NEXT: csel w11, w16, w8, lt
; CHECK-CVT-NEXT: cmn w11, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: fmov s1, w15
; CHECK-CVT-NEXT: fcvtzs w15, s0
; CHECK-CVT-NEXT: csel w11, w11, w10, gt
; CHECK-CVT-NEXT: mov v2.s[1], w9
; CHECK-CVT-NEXT: mov v1.s[1], w14
; CHECK-CVT-NEXT: cmp w15, w8
; CHECK-CVT-NEXT: csel w8, w15, w8, lt
; CHECK-CVT-NEXT: cmn w8, #8, lsl #12 // =32768
; CHECK-CVT-NEXT: csel w8, w8, w10, gt
; CHECK-CVT-NEXT: mov v1.s[2], w11
; CHECK-CVT-NEXT: mov v2.s[2], w12
; CHECK-CVT-NEXT: mov v1.s[3], w8
; CHECK-CVT-NEXT: mov v2.s[3], w13
; CHECK-CVT-NEXT: xtn v0.4h, v1.4s
; CHECK-CVT-NEXT: xtn2 v0.8h, v2.4s
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v8f16_v8i16:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: fcvtzs v0.8h, v0.8h
; CHECK-FP16-NEXT: ret
%x = call <8 x i16> @llvm.fptosi.sat.v8f16.v8i16(<8 x half> %f)
ret <8 x i16> %x
}
define <8 x i19> @test_signed_v8f16_v8i19(<8 x half> %f) {
; CHECK-LABEL: test_signed_v8f16_v8i19:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl v2.4s, v0.4h
; CHECK-NEXT: fcvtl2 v0.4s, v0.8h
; CHECK-NEXT: movi v1.4s, #3, msl #16
; CHECK-NEXT: mvni v3.4s, #3, msl #16
; CHECK-NEXT: fcvtzs v2.4s, v2.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: smin v2.4s, v2.4s, v1.4s
; CHECK-NEXT: smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT: smax v1.4s, v2.4s, v3.4s
; CHECK-NEXT: smax v0.4s, v0.4s, v3.4s
; CHECK-NEXT: mov w1, v1.s[1]
; CHECK-NEXT: mov w2, v1.s[2]
; CHECK-NEXT: mov w5, v0.s[1]
; CHECK-NEXT: mov w3, v1.s[3]
; CHECK-NEXT: mov w6, v0.s[2]
; CHECK-NEXT: mov w7, v0.s[3]
; CHECK-NEXT: fmov w4, s0
; CHECK-NEXT: fmov w0, s1
; CHECK-NEXT: ret
%x = call <8 x i19> @llvm.fptosi.sat.v8f16.v8i19(<8 x half> %f)
ret <8 x i19> %x
}
define <8 x i32> @test_signed_v8f16_v8i32_duplicate(<8 x half> %f) {
; CHECK-LABEL: test_signed_v8f16_v8i32_duplicate:
; CHECK: // %bb.0:
; CHECK-NEXT: fcvtl2 v1.4s, v0.8h
; CHECK-NEXT: fcvtl v0.4s, v0.4h
; CHECK-NEXT: fcvtzs v1.4s, v1.4s
; CHECK-NEXT: fcvtzs v0.4s, v0.4s
; CHECK-NEXT: ret
%x = call <8 x i32> @llvm.fptosi.sat.v8f16.v8i32(<8 x half> %f)
ret <8 x i32> %x
}
define <8 x i50> @test_signed_v8f16_v8i50(<8 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v8f16_v8i50:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: ext v1.16b, v0.16b, v0.16b, #8
; CHECK-CVT-NEXT: mov x8, #562949953421311
; CHECK-CVT-NEXT: mov x12, #-562949953421312
; CHECK-CVT-NEXT: fcvt s5, h0
; CHECK-CVT-NEXT: mov h2, v1.h[1]
; CHECK-CVT-NEXT: fcvt s3, h1
; CHECK-CVT-NEXT: mov h4, v1.h[2]
; CHECK-CVT-NEXT: mov h1, v1.h[3]
; CHECK-CVT-NEXT: fcvtzs x10, s5
; CHECK-CVT-NEXT: fcvt s2, h2
; CHECK-CVT-NEXT: fcvtzs x9, s3
; CHECK-CVT-NEXT: fcvt s3, h4
; CHECK-CVT-NEXT: fcvt s1, h1
; CHECK-CVT-NEXT: cmp x9, x8
; CHECK-CVT-NEXT: fcvtzs x11, s2
; CHECK-CVT-NEXT: csel x9, x9, x8, lt
; CHECK-CVT-NEXT: cmp x9, x12
; CHECK-CVT-NEXT: fcvtzs x13, s3
; CHECK-CVT-NEXT: csel x4, x9, x12, gt
; CHECK-CVT-NEXT: mov h2, v0.h[1]
; CHECK-CVT-NEXT: cmp x11, x8
; CHECK-CVT-NEXT: fcvtzs x9, s1
; CHECK-CVT-NEXT: csel x11, x11, x8, lt
; CHECK-CVT-NEXT: mov h1, v0.h[2]
; CHECK-CVT-NEXT: cmp x11, x12
; CHECK-CVT-NEXT: mov h0, v0.h[3]
; CHECK-CVT-NEXT: csel x5, x11, x12, gt
; CHECK-CVT-NEXT: cmp x13, x8
; CHECK-CVT-NEXT: csel x11, x13, x8, lt
; CHECK-CVT-NEXT: fcvt s2, h2
; CHECK-CVT-NEXT: cmp x11, x12
; CHECK-CVT-NEXT: fcvt s1, h1
; CHECK-CVT-NEXT: csel x6, x11, x12, gt
; CHECK-CVT-NEXT: cmp x9, x8
; CHECK-CVT-NEXT: csel x9, x9, x8, lt
; CHECK-CVT-NEXT: fcvt s0, h0
; CHECK-CVT-NEXT: cmp x9, x12
; CHECK-CVT-NEXT: fcvtzs x11, s2
; CHECK-CVT-NEXT: csel x7, x9, x12, gt
; CHECK-CVT-NEXT: cmp x10, x8
; CHECK-CVT-NEXT: csel x9, x10, x8, lt
; CHECK-CVT-NEXT: fcvtzs x10, s1
; CHECK-CVT-NEXT: cmp x9, x12
; CHECK-CVT-NEXT: csel x0, x9, x12, gt
; CHECK-CVT-NEXT: cmp x11, x8
; CHECK-CVT-NEXT: csel x9, x11, x8, lt
; CHECK-CVT-NEXT: fcvtzs x11, s0
; CHECK-CVT-NEXT: cmp x9, x12
; CHECK-CVT-NEXT: csel x1, x9, x12, gt
; CHECK-CVT-NEXT: cmp x10, x8
; CHECK-CVT-NEXT: csel x9, x10, x8, lt
; CHECK-CVT-NEXT: cmp x9, x12
; CHECK-CVT-NEXT: csel x2, x9, x12, gt
; CHECK-CVT-NEXT: cmp x11, x8
; CHECK-CVT-NEXT: csel x8, x11, x8, lt
; CHECK-CVT-NEXT: cmp x8, x12
; CHECK-CVT-NEXT: csel x3, x8, x12, gt
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v8f16_v8i50:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: ext v1.16b, v0.16b, v0.16b, #8
; CHECK-FP16-NEXT: mov x8, #562949953421311
; CHECK-FP16-NEXT: mov x11, #-562949953421312
; CHECK-FP16-NEXT: mov h2, v1.h[1]
; CHECK-FP16-NEXT: fcvtzs x9, h1
; CHECK-FP16-NEXT: mov h3, v1.h[2]
; CHECK-FP16-NEXT: mov h1, v1.h[3]
; CHECK-FP16-NEXT: cmp x9, x8
; CHECK-FP16-NEXT: fcvtzs x10, h2
; CHECK-FP16-NEXT: csel x9, x9, x8, lt
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: fcvtzs x12, h3
; CHECK-FP16-NEXT: csel x4, x9, x11, gt
; CHECK-FP16-NEXT: mov h2, v0.h[2]
; CHECK-FP16-NEXT: cmp x10, x8
; CHECK-FP16-NEXT: csel x9, x10, x8, lt
; CHECK-FP16-NEXT: fcvtzs x10, h1
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: mov h1, v0.h[1]
; CHECK-FP16-NEXT: csel x5, x9, x11, gt
; CHECK-FP16-NEXT: cmp x12, x8
; CHECK-FP16-NEXT: csel x9, x12, x8, lt
; CHECK-FP16-NEXT: fcvtzs x12, h0
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: mov h0, v0.h[3]
; CHECK-FP16-NEXT: csel x6, x9, x11, gt
; CHECK-FP16-NEXT: cmp x10, x8
; CHECK-FP16-NEXT: csel x9, x10, x8, lt
; CHECK-FP16-NEXT: fcvtzs x10, h1
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: csel x7, x9, x11, gt
; CHECK-FP16-NEXT: cmp x12, x8
; CHECK-FP16-NEXT: csel x9, x12, x8, lt
; CHECK-FP16-NEXT: fcvtzs x12, h2
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: csel x0, x9, x11, gt
; CHECK-FP16-NEXT: cmp x10, x8
; CHECK-FP16-NEXT: csel x9, x10, x8, lt
; CHECK-FP16-NEXT: fcvtzs x10, h0
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: csel x1, x9, x11, gt
; CHECK-FP16-NEXT: cmp x12, x8
; CHECK-FP16-NEXT: csel x9, x12, x8, lt
; CHECK-FP16-NEXT: cmp x9, x11
; CHECK-FP16-NEXT: csel x2, x9, x11, gt
; CHECK-FP16-NEXT: cmp x10, x8
; CHECK-FP16-NEXT: csel x8, x10, x8, lt
; CHECK-FP16-NEXT: cmp x8, x11
; CHECK-FP16-NEXT: csel x3, x8, x11, gt
; CHECK-FP16-NEXT: ret
%x = call <8 x i50> @llvm.fptosi.sat.v8f16.v8i50(<8 x half> %f)
ret <8 x i50> %x
}
define <8 x i64> @test_signed_v8f16_v8i64(<8 x half> %f) {
; CHECK-CVT-LABEL: test_signed_v8f16_v8i64:
; CHECK-CVT: // %bb.0:
; CHECK-CVT-NEXT: ext v1.16b, v0.16b, v0.16b, #8
; CHECK-CVT-NEXT: mov h4, v0.h[2]
; CHECK-CVT-NEXT: fcvt s5, h0
; CHECK-CVT-NEXT: fcvt s2, h1
; CHECK-CVT-NEXT: mov h3, v1.h[1]
; CHECK-CVT-NEXT: mov h6, v1.h[2]
; CHECK-CVT-NEXT: fcvt s4, h4
; CHECK-CVT-NEXT: mov h1, v1.h[3]
; CHECK-CVT-NEXT: fcvtzs x9, s5
; CHECK-CVT-NEXT: fcvtzs x8, s2
; CHECK-CVT-NEXT: fcvt s2, h3
; CHECK-CVT-NEXT: mov h3, v0.h[1]
; CHECK-CVT-NEXT: mov h0, v0.h[3]
; CHECK-CVT-NEXT: fcvt s5, h6
; CHECK-CVT-NEXT: fcvt s6, h1
; CHECK-CVT-NEXT: fcvtzs x10, s2
; CHECK-CVT-NEXT: fmov d2, x8
; CHECK-CVT-NEXT: fcvtzs x8, s4
; CHECK-CVT-NEXT: fcvt s3, h3
; CHECK-CVT-NEXT: fcvt s4, h0
; CHECK-CVT-NEXT: fmov d0, x9
; CHECK-CVT-NEXT: mov v2.d[1], x10
; CHECK-CVT-NEXT: fcvtzs x10, s5
; CHECK-CVT-NEXT: fmov d1, x8
; CHECK-CVT-NEXT: fcvtzs x9, s3
; CHECK-CVT-NEXT: fcvtzs x8, s4
; CHECK-CVT-NEXT: fmov d3, x10
; CHECK-CVT-NEXT: fcvtzs x10, s6
; CHECK-CVT-NEXT: mov v0.d[1], x9
; CHECK-CVT-NEXT: mov v1.d[1], x8
; CHECK-CVT-NEXT: mov v3.d[1], x10
; CHECK-CVT-NEXT: ret
;
; CHECK-FP16-LABEL: test_signed_v8f16_v8i64:
; CHECK-FP16: // %bb.0:
; CHECK-FP16-NEXT: ext v1.16b, v0.16b, v0.16b, #8
; CHECK-FP16-NEXT: mov h3, v0.h[2]
; CHECK-FP16-NEXT: mov h5, v0.h[3]
; CHECK-FP16-NEXT: fcvtzs x9, h0
; CHECK-FP16-NEXT: mov h2, v1.h[1]
; CHECK-FP16-NEXT: fcvtzs x8, h1
; CHECK-FP16-NEXT: mov h4, v1.h[2]
; CHECK-FP16-NEXT: mov h6, v1.h[3]
; CHECK-FP16-NEXT: fcvtzs x10, h2
; CHECK-FP16-NEXT: fmov d2, x8
; CHECK-FP16-NEXT: fcvtzs x8, h3
; CHECK-FP16-NEXT: mov h3, v0.h[1]
; CHECK-FP16-NEXT: fmov d0, x9
; CHECK-FP16-NEXT: mov v2.d[1], x10
; CHECK-FP16-NEXT: fcvtzs x10, h4
; CHECK-FP16-NEXT: fmov d1, x8
; CHECK-FP16-NEXT: fcvtzs x9, h3
; CHECK-FP16-NEXT: fcvtzs x8, h5
; CHECK-FP16-NEXT: fmov d3, x10
; CHECK-FP16-NEXT: fcvtzs x10, h6
; CHECK-FP16-NEXT: mov v0.d[1], x9
; CHECK-FP16-NEXT: mov v1.d[1], x8
; CHECK-FP16-NEXT: mov v3.d[1], x10
; CHECK-FP16-NEXT: ret
%x = call <8 x i64> @llvm.fptosi.sat.v8f16.v8i64(<8 x half> %f)
ret <8 x i64> %x
}
define <8 x i100> @test_signed_v8f16_v8i100(<8 x half> %f) {
; CHECK-LABEL: test_signed_v8f16_v8i100:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #192
; CHECK-NEXT: str d10, [sp, #64] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x29, x30, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: stp x28, x27, [sp, #112] // 16-byte Folded Spill
; CHECK-NEXT: stp x26, x25, [sp, #128] // 16-byte Folded Spill
; CHECK-NEXT: stp x24, x23, [sp, #144] // 16-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #160] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #176] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 192
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w23, -40
; CHECK-NEXT: .cfi_offset w24, -48
; CHECK-NEXT: .cfi_offset w25, -56
; CHECK-NEXT: .cfi_offset w26, -64
; CHECK-NEXT: .cfi_offset w27, -72
; CHECK-NEXT: .cfi_offset w28, -80
; CHECK-NEXT: .cfi_offset w30, -88
; CHECK-NEXT: .cfi_offset w29, -96
; CHECK-NEXT: .cfi_offset b8, -104
; CHECK-NEXT: .cfi_offset b9, -112
; CHECK-NEXT: .cfi_offset b10, -128
; CHECK-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: mov x19, x8
; CHECK-NEXT: ext v0.16b, v0.16b, v0.16b, #8
; CHECK-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: mov h0, v0.h[1]
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-251658240
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: mov x25, #-34359738368
; CHECK-NEXT: mov x23, #34359738367
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: mov w8, #1895825407
; CHECK-NEXT: mov h0, v0.h[3]
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x23, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp, #24] // 8-byte Folded Spill
; CHECK-NEXT: csel x8, xzr, x9, vs
; CHECK-NEXT: str x8, [sp, #72] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x23, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: csel x22, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp, #16] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[2]
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x23, x8, gt
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x24, xzr, x8, vs
; CHECK-NEXT: csel x8, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp, #32] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[1]
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x23, x8, gt
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x27, xzr, x8, vs
; CHECK-NEXT: csel x8, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp, #8] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[3]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x23, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: csel x29, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x25, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x23, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x20, xzr, x8, vs
; CHECK-NEXT: csel x28, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[2]
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x23, x8, gt
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x21, xzr, x8, vs
; CHECK-NEXT: csel x26, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fmov d0, x20
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: ldr x11, [sp, #8] // 8-byte Folded Reload
; CHECK-NEXT: lsr x10, x28, #28
; CHECK-NEXT: ldr d1, [sp] // 8-byte Folded Reload
; CHECK-NEXT: lsr x12, x29, #28
; CHECK-NEXT: mov v0.d[1], x28
; CHECK-NEXT: csel x8, x25, x1, lt
; CHECK-NEXT: csel x9, xzr, x0, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: stur x11, [x19, #75]
; CHECK-NEXT: ldr x13, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: csinv x9, x9, xzr, le
; CHECK-NEXT: csel x8, x23, x8, gt
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fmov x11, d0
; CHECK-NEXT: stur x13, [x19, #50]
; CHECK-NEXT: mov v1.d[1], x29
; CHECK-NEXT: ldr d0, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: strb w10, [x19, #49]
; CHECK-NEXT: extr x10, x28, x11, #28
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: bfi x8, x11, #36, #28
; CHECK-NEXT: strb w12, [x19, #24]
; CHECK-NEXT: stur x9, [x19, #25]
; CHECK-NEXT: fmov x12, d1
; CHECK-NEXT: stur x10, [x19, #41]
; CHECK-NEXT: lsr x9, x22, #28
; CHECK-NEXT: ldr d1, [sp, #24] // 8-byte Folded Reload
; CHECK-NEXT: stur x8, [x19, #33]
; CHECK-NEXT: ldr x11, [sp, #72] // 8-byte Folded Reload
; CHECK-NEXT: extr x18, x29, x12, #28
; CHECK-NEXT: mov v0.d[1], x22
; CHECK-NEXT: bfi x21, x12, #36, #28
; CHECK-NEXT: str x26, [x19]
; CHECK-NEXT: mov v1.d[1], x11
; CHECK-NEXT: lsr x10, x11, #28
; CHECK-NEXT: mov x13, x11
; CHECK-NEXT: stp x21, x18, [x19, #8]
; CHECK-NEXT: fmov x8, d0
; CHECK-NEXT: strb w9, [x19, #99]
; CHECK-NEXT: strb w10, [x19, #74]
; CHECK-NEXT: fmov x11, d1
; CHECK-NEXT: extr x12, x22, x8, #28
; CHECK-NEXT: bfi x27, x8, #36, #28
; CHECK-NEXT: extr x8, x13, x11, #28
; CHECK-NEXT: bfi x24, x11, #36, #28
; CHECK-NEXT: stur x12, [x19, #91]
; CHECK-NEXT: stur x27, [x19, #83]
; CHECK-NEXT: stur x8, [x19, #66]
; CHECK-NEXT: stur x24, [x19, #58]
; CHECK-NEXT: ldp x20, x19, [sp, #176] // 16-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #160] // 16-byte Folded Reload
; CHECK-NEXT: ldp x24, x23, [sp, #144] // 16-byte Folded Reload
; CHECK-NEXT: ldp x26, x25, [sp, #128] // 16-byte Folded Reload
; CHECK-NEXT: ldp x28, x27, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: ldp x29, x30, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: ldr d10, [sp, #64] // 8-byte Folded Reload
; CHECK-NEXT: add sp, sp, #192
; CHECK-NEXT: ret
%x = call <8 x i100> @llvm.fptosi.sat.v8f16.v8i100(<8 x half> %f)
ret <8 x i100> %x
}
define <8 x i128> @test_signed_v8f16_v8i128(<8 x half> %f) {
; CHECK-LABEL: test_signed_v8f16_v8i128:
; CHECK: // %bb.0:
; CHECK-NEXT: sub sp, sp, #192
; CHECK-NEXT: str d10, [sp, #64] // 8-byte Folded Spill
; CHECK-NEXT: stp d9, d8, [sp, #80] // 16-byte Folded Spill
; CHECK-NEXT: stp x29, x30, [sp, #96] // 16-byte Folded Spill
; CHECK-NEXT: stp x28, x27, [sp, #112] // 16-byte Folded Spill
; CHECK-NEXT: stp x26, x25, [sp, #128] // 16-byte Folded Spill
; CHECK-NEXT: stp x24, x23, [sp, #144] // 16-byte Folded Spill
; CHECK-NEXT: stp x22, x21, [sp, #160] // 16-byte Folded Spill
; CHECK-NEXT: stp x20, x19, [sp, #176] // 16-byte Folded Spill
; CHECK-NEXT: .cfi_def_cfa_offset 192
; CHECK-NEXT: .cfi_offset w19, -8
; CHECK-NEXT: .cfi_offset w20, -16
; CHECK-NEXT: .cfi_offset w21, -24
; CHECK-NEXT: .cfi_offset w22, -32
; CHECK-NEXT: .cfi_offset w23, -40
; CHECK-NEXT: .cfi_offset w24, -48
; CHECK-NEXT: .cfi_offset w25, -56
; CHECK-NEXT: .cfi_offset w26, -64
; CHECK-NEXT: .cfi_offset w27, -72
; CHECK-NEXT: .cfi_offset w28, -80
; CHECK-NEXT: .cfi_offset w30, -88
; CHECK-NEXT: .cfi_offset w29, -96
; CHECK-NEXT: .cfi_offset b8, -104
; CHECK-NEXT: .cfi_offset b9, -112
; CHECK-NEXT: .cfi_offset b10, -128
; CHECK-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-NEXT: mov x19, x8
; CHECK-NEXT: ext v0.16b, v0.16b, v0.16b, #8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: mov w8, #-16777216
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: mov x21, #-9223372036854775808
; CHECK-NEXT: mov x22, #9223372036854775807
; CHECK-NEXT: fmov s10, w8
; CHECK-NEXT: mov w8, #2130706431
; CHECK-NEXT: mov h0, v0.h[1]
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: fmov s9, w8
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp, #72] // 8-byte Folded Spill
; CHECK-NEXT: csel x8, xzr, x9, vs
; CHECK-NEXT: str x8, [sp, #24] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[2]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x10, xzr, x8, vs
; CHECK-NEXT: csel x8, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: stp x8, x10, [sp, #8] // 16-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: mov h0, v0.h[3]
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: str x8, [sp, #32] // 8-byte Folded Spill
; CHECK-NEXT: csel x8, xzr, x9, vs
; CHECK-NEXT: str x8, [sp] // 8-byte Folded Spill
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x28, xzr, x8, vs
; CHECK-NEXT: csel x29, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[1]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x20, xzr, x8, vs
; CHECK-NEXT: csel x23, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[2]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x24, xzr, x8, vs
; CHECK-NEXT: csel x25, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: mov h0, v0.h[3]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: fcvt s8, h0
; CHECK-NEXT: csel x26, xzr, x8, vs
; CHECK-NEXT: csel x27, xzr, x9, vs
; CHECK-NEXT: fmov s0, s8
; CHECK-NEXT: bl __fixsfti
; CHECK-NEXT: fcmp s8, s10
; CHECK-NEXT: stp x26, x27, [x19, #32]
; CHECK-NEXT: stp x24, x25, [x19, #16]
; CHECK-NEXT: stp x20, x23, [x19]
; CHECK-NEXT: csel x8, xzr, x0, lt
; CHECK-NEXT: csel x9, x21, x1, lt
; CHECK-NEXT: fcmp s8, s9
; CHECK-NEXT: stp x28, x29, [x19, #112]
; CHECK-NEXT: ldr x10, [sp] // 8-byte Folded Reload
; CHECK-NEXT: csel x9, x22, x9, gt
; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: fcmp s8, s8
; CHECK-NEXT: str x10, [x19, #104]
; CHECK-NEXT: ldr x10, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: csel x9, xzr, x9, vs
; CHECK-NEXT: csel x8, xzr, x8, vs
; CHECK-NEXT: str x10, [x19, #96]
; CHECK-NEXT: stp x8, x9, [x19, #48]
; CHECK-NEXT: ldr x8, [sp, #8] // 8-byte Folded Reload
; CHECK-NEXT: str x8, [x19, #88]
; CHECK-NEXT: ldr x8, [sp, #16] // 8-byte Folded Reload
; CHECK-NEXT: str x8, [x19, #80]
; CHECK-NEXT: ldr x8, [sp, #24] // 8-byte Folded Reload
; CHECK-NEXT: str x8, [x19, #72]
; CHECK-NEXT: ldr x8, [sp, #72] // 8-byte Folded Reload
; CHECK-NEXT: str x8, [x19, #64]
; CHECK-NEXT: ldp x20, x19, [sp, #176] // 16-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #160] // 16-byte Folded Reload
; CHECK-NEXT: ldp x24, x23, [sp, #144] // 16-byte Folded Reload
; CHECK-NEXT: ldp x26, x25, [sp, #128] // 16-byte Folded Reload
; CHECK-NEXT: ldp x28, x27, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: ldp x29, x30, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: ldr d10, [sp, #64] // 8-byte Folded Reload
; CHECK-NEXT: add sp, sp, #192
; CHECK-NEXT: ret
%x = call <8 x i128> @llvm.fptosi.sat.v8f16.v8i128(<8 x half> %f)
ret <8 x i128> %x
}