llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+experimental-v,+experimental-zfh,+f,+d -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV32
; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+experimental-v,+experimental-zfh,+f,+d -riscv-v-vector-bits-min=128  -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV64

declare half @llvm.vector.reduce.fadd.v1f16(half, <1 x half>)

define half @vreduce_fadd_v1f16(<1 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_fadd_v1f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 1, e16, mf4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.h fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <1 x half>, <1 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v1f16(half %s, <1 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v1f16(<1 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v1f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 1, e16, mf4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 1, e16, mf4, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <1 x half>, <1 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v1f16(half %s, <1 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v2f16(half, <2 x half>)

define half @vreduce_fadd_v2f16(<2 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_fadd_v2f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI2_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI2_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.h fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <2 x half>, <2 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v2f16(half %s, <2 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v2f16(<2 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v2f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x half>, <2 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v2f16(half %s, <2 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v4f16(half, <4 x half>)

define half @vreduce_fadd_v4f16(<4 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_fadd_v4f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI4_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI4_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.h fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v4f16(half %s, <4 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v4f16(<4 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v4f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v4f16(half %s, <4 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v8f16(half, <8 x half>)

define half @vreduce_fadd_v8f16(<8 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_fadd_v8f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI6_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI6_0)(a1)
; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.h fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <8 x half>, <8 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v8f16(half %s, <8 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v8f16(<8 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v8f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <8 x half>, <8 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v8f16(half %s, <8 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v16f16(half, <16 x half>)

define half @vreduce_fadd_v16f16(<16 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_fadd_v16f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI8_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI8_0)(a1)
; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.h fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <16 x half>, <16 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v16f16(half %s, <16 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v16f16(<16 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v16f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, fa0
; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <16 x half>, <16 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v16f16(half %s, <16 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v32f16(half, <32 x half>)

define half @vreduce_fadd_v32f16(<32 x half>* %x, half %s) {
; RV32-LABEL: vreduce_fadd_v32f16:
; RV32:       # %bb.0:
; RV32-NEXT:    li a1, 32
; RV32-NEXT:    lui a2, %hi(.LCPI10_0)
; RV32-NEXT:    flh ft0, %lo(.LCPI10_0)(a2)
; RV32-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
; RV32-NEXT:    vle16.v v8, (a0)
; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; RV32-NEXT:    vfmv.v.f v12, ft0
; RV32-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
; RV32-NEXT:    vfredusum.vs v8, v8, v12
; RV32-NEXT:    vfmv.f.s ft0, v8
; RV32-NEXT:    fadd.h fa0, fa0, ft0
; RV32-NEXT:    ret
;
; RV64-LABEL: vreduce_fadd_v32f16:
; RV64:       # %bb.0:
; RV64-NEXT:    lui a1, %hi(.LCPI10_0)
; RV64-NEXT:    flh ft0, %lo(.LCPI10_0)(a1)
; RV64-NEXT:    li a1, 32
; RV64-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
; RV64-NEXT:    vle16.v v8, (a0)
; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; RV64-NEXT:    vfmv.v.f v12, ft0
; RV64-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
; RV64-NEXT:    vfredusum.vs v8, v8, v12
; RV64-NEXT:    vfmv.f.s ft0, v8
; RV64-NEXT:    fadd.h fa0, fa0, ft0
; RV64-NEXT:    ret
  %v = load <32 x half>, <32 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v32f16(half %s, <32 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v32f16(<32 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v32f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 32
; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v12, fa0
; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v12
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <32 x half>, <32 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v32f16(half %s, <32 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v64f16(half, <64 x half>)

define half @vreduce_fadd_v64f16(<64 x half>* %x, half %s) {
; RV32-LABEL: vreduce_fadd_v64f16:
; RV32:       # %bb.0:
; RV32-NEXT:    li a1, 64
; RV32-NEXT:    lui a2, %hi(.LCPI12_0)
; RV32-NEXT:    flh ft0, %lo(.LCPI12_0)(a2)
; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; RV32-NEXT:    vle16.v v8, (a0)
; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; RV32-NEXT:    vfmv.v.f v16, ft0
; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; RV32-NEXT:    vfredusum.vs v8, v8, v16
; RV32-NEXT:    vfmv.f.s ft0, v8
; RV32-NEXT:    fadd.h fa0, fa0, ft0
; RV32-NEXT:    ret
;
; RV64-LABEL: vreduce_fadd_v64f16:
; RV64:       # %bb.0:
; RV64-NEXT:    lui a1, %hi(.LCPI12_0)
; RV64-NEXT:    flh ft0, %lo(.LCPI12_0)(a1)
; RV64-NEXT:    li a1, 64
; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; RV64-NEXT:    vle16.v v8, (a0)
; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; RV64-NEXT:    vfmv.v.f v16, ft0
; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; RV64-NEXT:    vfredusum.vs v8, v8, v16
; RV64-NEXT:    vfmv.f.s ft0, v8
; RV64-NEXT:    fadd.h fa0, fa0, ft0
; RV64-NEXT:    ret
  %v = load <64 x half>, <64 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v64f16(half %s, <64 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v64f16(<64 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v64f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 64
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, fa0
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <64 x half>, <64 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v64f16(half %s, <64 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fadd.v128f16(half, <128 x half>)

define half @vreduce_fadd_v128f16(<128 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_fadd_v128f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 64
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle16.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI14_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI14_0)(a0)
; CHECK-NEXT:    vfadd.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.h fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <128 x half>, <128 x half>* %x
  %red = call reassoc half @llvm.vector.reduce.fadd.v128f16(half %s, <128 x half> %v)
  ret half %red
}

define half @vreduce_ord_fadd_v128f16(<128 x half>* %x, half %s) {
; CHECK-LABEL: vreduce_ord_fadd_v128f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    addi a1, a0, 128
; CHECK-NEXT:    li a2, 64
; CHECK-NEXT:    vsetvli zero, a2, e16, m8, ta, mu
; CHECK-NEXT:    vle16.v v8, (a1)
; CHECK-NEXT:    vle16.v v16, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v24, fa0
; CHECK-NEXT:    vsetvli zero, a2, e16, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v16, v16, v24
; CHECK-NEXT:    vfmv.f.s ft0, v16
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a2, e16, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <128 x half>, <128 x half>* %x
  %red = call half @llvm.vector.reduce.fadd.v128f16(half %s, <128 x half> %v)
  ret half %red
}

declare float @llvm.vector.reduce.fadd.v1f32(float, <1 x float>)

define float @vreduce_fadd_v1f32(<1 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_fadd_v1f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 1, e32, mf2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.s fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <1 x float>, <1 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v1f32(float %s, <1 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v1f32(<1 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v1f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 1, e32, mf2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 1, e32, mf2, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <1 x float>, <1 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v1f32(float %s, <1 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fadd.v2f32(float, <2 x float>)

define float @vreduce_fadd_v2f32(<2 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_fadd_v2f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI18_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI18_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.s fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <2 x float>, <2 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v2f32(float %s, <2 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v2f32(<2 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v2f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x float>, <2 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v2f32(float %s, <2 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fadd.v4f32(float, <4 x float>)

define float @vreduce_fadd_v4f32(<4 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_fadd_v4f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI20_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI20_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.s fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v4f32(float %s, <4 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v4f32(<4 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v4f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v4f32(float %s, <4 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fadd.v8f32(float, <8 x float>)

define float @vreduce_fadd_v8f32(<8 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_fadd_v8f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI22_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI22_0)(a1)
; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.s fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <8 x float>, <8 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v8f32(float %s, <8 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v8f32(<8 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v8f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, fa0
; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <8 x float>, <8 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v8f32(float %s, <8 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fadd.v16f32(float, <16 x float>)

define float @vreduce_fadd_v16f32(<16 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_fadd_v16f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI24_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI24_0)(a1)
; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v12, ft0
; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v12
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.s fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <16 x float>, <16 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v16f32(float %s, <16 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v16f32(<16 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v16f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v12, fa0
; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v12
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <16 x float>, <16 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v16f32(float %s, <16 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fadd.v32f32(float, <32 x float>)

define float @vreduce_fadd_v32f32(<32 x float>* %x, float %s) {
; RV32-LABEL: vreduce_fadd_v32f32:
; RV32:       # %bb.0:
; RV32-NEXT:    li a1, 32
; RV32-NEXT:    lui a2, %hi(.LCPI26_0)
; RV32-NEXT:    flw ft0, %lo(.LCPI26_0)(a2)
; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; RV32-NEXT:    vle32.v v8, (a0)
; RV32-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; RV32-NEXT:    vfmv.v.f v16, ft0
; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; RV32-NEXT:    vfredusum.vs v8, v8, v16
; RV32-NEXT:    vfmv.f.s ft0, v8
; RV32-NEXT:    fadd.s fa0, fa0, ft0
; RV32-NEXT:    ret
;
; RV64-LABEL: vreduce_fadd_v32f32:
; RV64:       # %bb.0:
; RV64-NEXT:    lui a1, %hi(.LCPI26_0)
; RV64-NEXT:    flw ft0, %lo(.LCPI26_0)(a1)
; RV64-NEXT:    li a1, 32
; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; RV64-NEXT:    vle32.v v8, (a0)
; RV64-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; RV64-NEXT:    vfmv.v.f v16, ft0
; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; RV64-NEXT:    vfredusum.vs v8, v8, v16
; RV64-NEXT:    vfmv.f.s ft0, v8
; RV64-NEXT:    fadd.s fa0, fa0, ft0
; RV64-NEXT:    ret
  %v = load <32 x float>, <32 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v32f32(float %s, <32 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v32f32(<32 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v32f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 32
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, fa0
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <32 x float>, <32 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v32f32(float %s, <32 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fadd.v64f32(float, <64 x float>)

define float @vreduce_fadd_v64f32(<64 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_fadd_v64f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 32
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle32.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI28_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI28_0)(a0)
; CHECK-NEXT:    vfadd.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.s fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <64 x float>, <64 x float>* %x
  %red = call reassoc float @llvm.vector.reduce.fadd.v64f32(float %s, <64 x float> %v)
  ret float %red
}

define float @vreduce_ord_fadd_v64f32(<64 x float>* %x, float %s) {
; CHECK-LABEL: vreduce_ord_fadd_v64f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    addi a1, a0, 128
; CHECK-NEXT:    li a2, 32
; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
; CHECK-NEXT:    vle32.v v8, (a1)
; CHECK-NEXT:    vle32.v v16, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v24, fa0
; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v16, v16, v24
; CHECK-NEXT:    vfmv.f.s ft0, v16
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <64 x float>, <64 x float>* %x
  %red = call float @llvm.vector.reduce.fadd.v64f32(float %s, <64 x float> %v)
  ret float %red
}

declare double @llvm.vector.reduce.fadd.v1f64(double, <1 x double>)

define double @vreduce_fadd_v1f64(<1 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_fadd_v1f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.d fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <1 x double>, <1 x double>* %x
  %red = call reassoc double @llvm.vector.reduce.fadd.v1f64(double %s, <1 x double> %v)
  ret double %red
}

define double @vreduce_ord_fadd_v1f64(<1 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_v1f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <1 x double>, <1 x double>* %x
  %red = call double @llvm.vector.reduce.fadd.v1f64(double %s, <1 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fadd.v2f64(double, <2 x double>)

define double @vreduce_fadd_v2f64(<2 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_fadd_v2f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI32_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI32_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.d fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <2 x double>, <2 x double>* %x
  %red = call reassoc double @llvm.vector.reduce.fadd.v2f64(double %s, <2 x double> %v)
  ret double %red
}

define double @vreduce_ord_fadd_v2f64(<2 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_v2f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, fa0
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x double>, <2 x double>* %x
  %red = call double @llvm.vector.reduce.fadd.v2f64(double %s, <2 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fadd.v4f64(double, <4 x double>)

define double @vreduce_fadd_v4f64(<4 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_fadd_v4f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI34_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI34_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.d fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call reassoc double @llvm.vector.reduce.fadd.v4f64(double %s, <4 x double> %v)
  ret double %red
}

define double @vreduce_ord_fadd_v4f64(<4 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_v4f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, fa0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call double @llvm.vector.reduce.fadd.v4f64(double %s, <4 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fadd.v8f64(double, <8 x double>)

define double @vreduce_fadd_v8f64(<8 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_fadd_v8f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI36_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI36_0)(a1)
; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v12, ft0
; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v12
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.d fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <8 x double>, <8 x double>* %x
  %red = call reassoc double @llvm.vector.reduce.fadd.v8f64(double %s, <8 x double> %v)
  ret double %red
}

define double @vreduce_ord_fadd_v8f64(<8 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_v8f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v12, fa0
; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v12
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <8 x double>, <8 x double>* %x
  %red = call double @llvm.vector.reduce.fadd.v8f64(double %s, <8 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fadd.v16f64(double, <16 x double>)

define double @vreduce_fadd_v16f64(<16 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_fadd_v16f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI38_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI38_0)(a1)
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.d fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <16 x double>, <16 x double>* %x
  %red = call reassoc double @llvm.vector.reduce.fadd.v16f64(double %s, <16 x double> %v)
  ret double %red
}

define double @vreduce_ord_fadd_v16f64(<16 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_v16f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, fa0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <16 x double>, <16 x double>* %x
  %red = call double @llvm.vector.reduce.fadd.v16f64(double %s, <16 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fadd.v32f64(double, <32 x double>)

define double @vreduce_fadd_v32f64(<32 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_fadd_v32f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle64.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI40_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI40_0)(a0)
; CHECK-NEXT:    vfadd.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredusum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s ft0, v8
; CHECK-NEXT:    fadd.d fa0, fa0, ft0
; CHECK-NEXT:    ret
  %v = load <32 x double>, <32 x double>* %x
  %red = call reassoc double @llvm.vector.reduce.fadd.v32f64(double %s, <32 x double> %v)
  ret double %red
}

define double @vreduce_ord_fadd_v32f64(<32 x double>* %x, double %s) {
; CHECK-LABEL: vreduce_ord_fadd_v32f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    addi a1, a0, 128
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vle64.v v8, (a1)
; CHECK-NEXT:    vle64.v v16, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v24, fa0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v16, v16, v24
; CHECK-NEXT:    vfmv.f.s ft0, v16
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredosum.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <32 x double>, <32 x double>* %x
  %red = call double @llvm.vector.reduce.fadd.v32f64(double %s, <32 x double> %v)
  ret double %red
}

declare half @llvm.vector.reduce.fmin.v2f16(<2 x half>)

define half @vreduce_fmin_v2f16(<2 x half>* %x) {
; CHECK-LABEL: vreduce_fmin_v2f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI42_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI42_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x half>, <2 x half>* %x
  %red = call half @llvm.vector.reduce.fmin.v2f16(<2 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fmin.v4f16(<4 x half>)

define half @vreduce_fmin_v4f16(<4 x half>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI43_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI43_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call half @llvm.vector.reduce.fmin.v4f16(<4 x half> %v)
  ret half %red
}

define half @vreduce_fmin_v4f16_nonans(<4 x half>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f16_nonans:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI44_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI44_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call nnan half @llvm.vector.reduce.fmin.v4f16(<4 x half> %v)
  ret half %red
}

define half @vreduce_fmin_v4f16_nonans_noinfs(<4 x half>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f16_nonans_noinfs:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI45_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI45_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call nnan ninf half @llvm.vector.reduce.fmin.v4f16(<4 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fmin.v128f16(<128 x half>)

define half @vreduce_fmin_v128f16(<128 x half>* %x) {
; CHECK-LABEL: vreduce_fmin_v128f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 64
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle16.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI46_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI46_0)(a0)
; CHECK-NEXT:    vfmin.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <128 x half>, <128 x half>* %x
  %red = call half @llvm.vector.reduce.fmin.v128f16(<128 x half> %v)
  ret half %red
}

declare float @llvm.vector.reduce.fmin.v2f32(<2 x float>)

define float @vreduce_fmin_v2f32(<2 x float>* %x) {
; CHECK-LABEL: vreduce_fmin_v2f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI47_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI47_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x float>, <2 x float>* %x
  %red = call float @llvm.vector.reduce.fmin.v2f32(<2 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fmin.v4f32(<4 x float>)

define float @vreduce_fmin_v4f32(<4 x float>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI48_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI48_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> %v)
  ret float %red
}

define float @vreduce_fmin_v4f32_nonans(<4 x float>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f32_nonans:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI49_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI49_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call nnan float @llvm.vector.reduce.fmin.v4f32(<4 x float> %v)
  ret float %red
}

define float @vreduce_fmin_v4f32_nonans_noinfs(<4 x float>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f32_nonans_noinfs:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI50_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI50_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call nnan ninf float @llvm.vector.reduce.fmin.v4f32(<4 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fmin.v128f32(<128 x float>)

define float @vreduce_fmin_v128f32(<128 x float>* %x) {
; CHECK-LABEL: vreduce_fmin_v128f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 32
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    addi a2, a0, 384
; CHECK-NEXT:    vle32.v v8, (a2)
; CHECK-NEXT:    addi a2, a0, 128
; CHECK-NEXT:    vle32.v v16, (a2)
; CHECK-NEXT:    vle32.v v24, (a0)
; CHECK-NEXT:    addi a0, a0, 256
; CHECK-NEXT:    vle32.v v0, (a0)
; CHECK-NEXT:    vfmin.vv v8, v16, v8
; CHECK-NEXT:    lui a0, %hi(.LCPI51_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI51_0)(a0)
; CHECK-NEXT:    vfmin.vv v16, v24, v0
; CHECK-NEXT:    vfmin.vv v8, v16, v8
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <128 x float>, <128 x float>* %x
  %red = call float @llvm.vector.reduce.fmin.v128f32(<128 x float> %v)
  ret float %red
}

declare double @llvm.vector.reduce.fmin.v2f64(<2 x double>)

define double @vreduce_fmin_v2f64(<2 x double>* %x) {
; CHECK-LABEL: vreduce_fmin_v2f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI52_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI52_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x double>, <2 x double>* %x
  %red = call double @llvm.vector.reduce.fmin.v2f64(<2 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fmin.v4f64(<4 x double>)

define double @vreduce_fmin_v4f64(<4 x double>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI53_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI53_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> %v)
  ret double %red
}

define double @vreduce_fmin_v4f64_nonans(<4 x double>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f64_nonans:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI54_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI54_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call nnan double @llvm.vector.reduce.fmin.v4f64(<4 x double> %v)
  ret double %red
}

define double @vreduce_fmin_v4f64_nonans_noinfs(<4 x double>* %x) {
; CHECK-LABEL: vreduce_fmin_v4f64_nonans_noinfs:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI55_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI55_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call nnan ninf double @llvm.vector.reduce.fmin.v4f64(<4 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fmin.v32f64(<32 x double>)

define double @vreduce_fmin_v32f64(<32 x double>* %x) {
; CHECK-LABEL: vreduce_fmin_v32f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle64.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI56_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI56_0)(a0)
; CHECK-NEXT:    vfmin.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredmin.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <32 x double>, <32 x double>* %x
  %red = call double @llvm.vector.reduce.fmin.v32f64(<32 x double> %v)
  ret double %red
}

declare half @llvm.vector.reduce.fmax.v2f16(<2 x half>)

define half @vreduce_fmax_v2f16(<2 x half>* %x) {
; CHECK-LABEL: vreduce_fmax_v2f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI57_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI57_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x half>, <2 x half>* %x
  %red = call half @llvm.vector.reduce.fmax.v2f16(<2 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fmax.v4f16(<4 x half>)

define half @vreduce_fmax_v4f16(<4 x half>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI58_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI58_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call half @llvm.vector.reduce.fmax.v4f16(<4 x half> %v)
  ret half %red
}

define half @vreduce_fmax_v4f16_nonans(<4 x half>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f16_nonans:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI59_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI59_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call nnan half @llvm.vector.reduce.fmax.v4f16(<4 x half> %v)
  ret half %red
}

define half @vreduce_fmax_v4f16_nonans_noinfs(<4 x half>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f16_nonans_noinfs:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI60_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI60_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x half>, <4 x half>* %x
  %red = call nnan ninf half @llvm.vector.reduce.fmax.v4f16(<4 x half> %v)
  ret half %red
}

declare half @llvm.vector.reduce.fmax.v128f16(<128 x half>)

define half @vreduce_fmax_v128f16(<128 x half>* %x) {
; CHECK-LABEL: vreduce_fmax_v128f16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 64
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vle16.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle16.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI61_0)
; CHECK-NEXT:    flh ft0, %lo(.LCPI61_0)(a0)
; CHECK-NEXT:    vfmax.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <128 x half>, <128 x half>* %x
  %red = call half @llvm.vector.reduce.fmax.v128f16(<128 x half> %v)
  ret half %red
}

declare float @llvm.vector.reduce.fmax.v2f32(<2 x float>)

define float @vreduce_fmax_v2f32(<2 x float>* %x) {
; CHECK-LABEL: vreduce_fmax_v2f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI62_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI62_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x float>, <2 x float>* %x
  %red = call float @llvm.vector.reduce.fmax.v2f32(<2 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fmax.v4f32(<4 x float>)

define float @vreduce_fmax_v4f32(<4 x float>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI63_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI63_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> %v)
  ret float %red
}

define float @vreduce_fmax_v4f32_nonans(<4 x float>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f32_nonans:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI64_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI64_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call nnan float @llvm.vector.reduce.fmax.v4f32(<4 x float> %v)
  ret float %red
}

define float @vreduce_fmax_v4f32_nonans_noinfs(<4 x float>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f32_nonans_noinfs:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI65_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI65_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x float>, <4 x float>* %x
  %red = call nnan ninf float @llvm.vector.reduce.fmax.v4f32(<4 x float> %v)
  ret float %red
}

declare float @llvm.vector.reduce.fmax.v128f32(<128 x float>)

define float @vreduce_fmax_v128f32(<128 x float>* %x) {
; CHECK-LABEL: vreduce_fmax_v128f32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    li a1, 32
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    addi a2, a0, 384
; CHECK-NEXT:    vle32.v v8, (a2)
; CHECK-NEXT:    addi a2, a0, 128
; CHECK-NEXT:    vle32.v v16, (a2)
; CHECK-NEXT:    vle32.v v24, (a0)
; CHECK-NEXT:    addi a0, a0, 256
; CHECK-NEXT:    vle32.v v0, (a0)
; CHECK-NEXT:    vfmax.vv v8, v16, v8
; CHECK-NEXT:    lui a0, %hi(.LCPI66_0)
; CHECK-NEXT:    flw ft0, %lo(.LCPI66_0)(a0)
; CHECK-NEXT:    vfmax.vv v16, v24, v0
; CHECK-NEXT:    vfmax.vv v8, v16, v8
; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <128 x float>, <128 x float>* %x
  %red = call float @llvm.vector.reduce.fmax.v128f32(<128 x float> %v)
  ret float %red
}

declare double @llvm.vector.reduce.fmax.v2f64(<2 x double>)

define double @vreduce_fmax_v2f64(<2 x double>* %x) {
; CHECK-LABEL: vreduce_fmax_v2f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI67_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI67_0)(a1)
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v9, ft0
; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v9
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <2 x double>, <2 x double>* %x
  %red = call double @llvm.vector.reduce.fmax.v2f64(<2 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fmax.v4f64(<4 x double>)

define double @vreduce_fmax_v4f64(<4 x double>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI68_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI68_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> %v)
  ret double %red
}

define double @vreduce_fmax_v4f64_nonans(<4 x double>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f64_nonans:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI69_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI69_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call nnan double @llvm.vector.reduce.fmax.v4f64(<4 x double> %v)
  ret double %red
}

define double @vreduce_fmax_v4f64_nonans_noinfs(<4 x double>* %x) {
; CHECK-LABEL: vreduce_fmax_v4f64_nonans_noinfs:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI70_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI70_0)(a1)
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v10, ft0
; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v10
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <4 x double>, <4 x double>* %x
  %red = call nnan ninf double @llvm.vector.reduce.fmax.v4f64(<4 x double> %v)
  ret double %red
}

declare double @llvm.vector.reduce.fmax.v32f64(<32 x double>)

define double @vreduce_fmax_v32f64(<32 x double>* %x) {
; CHECK-LABEL: vreduce_fmax_v32f64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vle64.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    vle64.v v16, (a0)
; CHECK-NEXT:    lui a0, %hi(.LCPI71_0)
; CHECK-NEXT:    fld ft0, %lo(.LCPI71_0)(a0)
; CHECK-NEXT:    vfmax.vv v8, v8, v16
; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
; CHECK-NEXT:    vfmv.v.f v16, ft0
; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
; CHECK-NEXT:    vfredmax.vs v8, v8, v16
; CHECK-NEXT:    vfmv.f.s fa0, v8
; CHECK-NEXT:    ret
  %v = load <32 x double>, <32 x double>* %x
  %red = call double @llvm.vector.reduce.fmax.v32f64(<32 x double> %v)
  ret double %red
}