llvm/llvm/test/Analysis/CostModel/ARM/reduce-fp.ll

; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
; RUN: opt < %s -mtriple=armv8a-linux-gnueabihf -mattr=+fp64 -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefix=CHECK-V8
; RUN: opt < %s -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefix=CHECK-MVEFP
; RUN: opt < %s -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefix=CHECK-MVEI

target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"

define void @fadd_strict() {
; CHECK-V8-LABEL: 'fadd_strict'
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v2f16 = call half @llvm.vector.reduce.fadd.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fadd_v4f16 = call half @llvm.vector.reduce.fadd.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fadd_v8f16 = call half @llvm.vector.reduce.fadd.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %fadd_v16f16 = call half @llvm.vector.reduce.fadd.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f32 = call float @llvm.vector.reduce.fadd.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v4f32 = call float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fadd_v8f32 = call float @llvm.vector.reduce.fadd.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f64 = call double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v4f64 = call double @llvm.vector.reduce.fadd.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %fadd_v4f128 = call fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fadd_strict'
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %fadd_v2f16 = call half @llvm.vector.reduce.fadd.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fadd_v4f16 = call half @llvm.vector.reduce.fadd.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %fadd_v8f16 = call half @llvm.vector.reduce.fadd.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %fadd_v16f16 = call half @llvm.vector.reduce.fadd.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %fadd_v2f32 = call float @llvm.vector.reduce.fadd.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v4f32 = call float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v8f32 = call float @llvm.vector.reduce.fadd.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f64 = call double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v4f64 = call double @llvm.vector.reduce.fadd.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fadd_v4f128 = call fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fadd_strict'
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f16 = call half @llvm.vector.reduce.fadd.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v4f16 = call half @llvm.vector.reduce.fadd.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fadd_v8f16 = call half @llvm.vector.reduce.fadd.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fadd_v16f16 = call half @llvm.vector.reduce.fadd.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f32 = call float @llvm.vector.reduce.fadd.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v4f32 = call float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fadd_v8f32 = call float @llvm.vector.reduce.fadd.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f64 = call double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v4f64 = call double @llvm.vector.reduce.fadd.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fadd_v4f128 = call fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
  %fadd_v2f16 = call half @llvm.vector.reduce.fadd.v2f16(half 0.0, <2 x half> undef)
  %fadd_v4f16 = call half @llvm.vector.reduce.fadd.v4f16(half 0.0, <4 x half> undef)
  %fadd_v8f16 = call half @llvm.vector.reduce.fadd.v8f16(half 0.0, <8 x half> undef)
  %fadd_v16f16 = call half @llvm.vector.reduce.fadd.v16f16(half 0.0, <16 x half> undef)
  %fadd_v2f32 = call float @llvm.vector.reduce.fadd.v2f32(float 0.0, <2 x float> undef)
  %fadd_v4f32 = call float @llvm.vector.reduce.fadd.v4f32(float 0.0, <4 x float> undef)
  %fadd_v8f32 = call float @llvm.vector.reduce.fadd.v8f32(float 0.0, <8 x float> undef)
  %fadd_v2f64 = call double @llvm.vector.reduce.fadd.v2f64(double 0.0, <2 x double> undef)
  %fadd_v4f64 = call double @llvm.vector.reduce.fadd.v4f64(double 0.0, <4 x double> undef)
  %fadd_v4f128 = call fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
  ret void
}


define void @fadd_unordered() {
; CHECK-V8-LABEL: 'fadd_unordered'
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v2f16 = call reassoc half @llvm.vector.reduce.fadd.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %fadd_v4f16 = call reassoc half @llvm.vector.reduce.fadd.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %fadd_v8f16 = call reassoc half @llvm.vector.reduce.fadd.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %fadd_v16f16 = call reassoc half @llvm.vector.reduce.fadd.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f32 = call reassoc float @llvm.vector.reduce.fadd.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fadd_v4f32 = call reassoc float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v8f32 = call reassoc float @llvm.vector.reduce.fadd.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v2f64 = call reassoc double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %fadd_v4f64 = call reassoc double @llvm.vector.reduce.fadd.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %fadd_v4f128 = call reassoc fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fadd_unordered'
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %fadd_v2f16 = call reassoc half @llvm.vector.reduce.fadd.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fadd_v4f16 = call reassoc half @llvm.vector.reduce.fadd.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fadd_v8f16 = call reassoc half @llvm.vector.reduce.fadd.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %fadd_v16f16 = call reassoc half @llvm.vector.reduce.fadd.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %fadd_v2f32 = call reassoc float @llvm.vector.reduce.fadd.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fadd_v4f32 = call reassoc float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fadd_v8f32 = call reassoc float @llvm.vector.reduce.fadd.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fadd_v2f64 = call reassoc double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %fadd_v4f64 = call reassoc double @llvm.vector.reduce.fadd.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %fadd_v4f128 = call reassoc fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fadd_unordered'
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fadd_v2f16 = call reassoc half @llvm.vector.reduce.fadd.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %fadd_v4f16 = call reassoc half @llvm.vector.reduce.fadd.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 169 for instruction: %fadd_v8f16 = call reassoc half @llvm.vector.reduce.fadd.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 225 for instruction: %fadd_v16f16 = call reassoc half @llvm.vector.reduce.fadd.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fadd_v2f32 = call reassoc float @llvm.vector.reduce.fadd.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %fadd_v4f32 = call reassoc float @llvm.vector.reduce.fadd.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: %fadd_v8f32 = call reassoc float @llvm.vector.reduce.fadd.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fadd_v2f64 = call reassoc double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %fadd_v4f64 = call reassoc double @llvm.vector.reduce.fadd.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %fadd_v4f128 = call reassoc fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
  %fadd_v2f16 = call reassoc half @llvm.vector.reduce.fadd.v2f16(half 0.0, <2 x half> undef)
  %fadd_v4f16 = call reassoc half @llvm.vector.reduce.fadd.v4f16(half 0.0, <4 x half> undef)
  %fadd_v8f16 = call reassoc half @llvm.vector.reduce.fadd.v8f16(half 0.0, <8 x half> undef)
  %fadd_v16f16 = call reassoc half @llvm.vector.reduce.fadd.v16f16(half 0.0, <16 x half> undef)
  %fadd_v2f32 = call reassoc float @llvm.vector.reduce.fadd.v2f32(float 0.0, <2 x float> undef)
  %fadd_v4f32 = call reassoc float @llvm.vector.reduce.fadd.v4f32(float 0.0, <4 x float> undef)
  %fadd_v8f32 = call reassoc float @llvm.vector.reduce.fadd.v8f32(float 0.0, <8 x float> undef)
  %fadd_v2f64 = call reassoc double @llvm.vector.reduce.fadd.v2f64(double 0.0, <2 x double> undef)
  %fadd_v4f64 = call reassoc double @llvm.vector.reduce.fadd.v4f64(double 0.0, <4 x double> undef)
  %fadd_v4f128 = call reassoc fp128 @llvm.vector.reduce.fadd.v4f128(fp128 undef, <4 x fp128> undef)
  ret void
}

define void @fmul_strict() {
; CHECK-V8-LABEL: 'fmul_strict'
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v2f16 = call half @llvm.vector.reduce.fmul.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fmul_v4f16 = call half @llvm.vector.reduce.fmul.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fmul_v8f16 = call half @llvm.vector.reduce.fmul.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %fmul_v16f16 = call half @llvm.vector.reduce.fmul.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f32 = call float @llvm.vector.reduce.fmul.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v4f32 = call float @llvm.vector.reduce.fmul.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fmul_v8f32 = call float @llvm.vector.reduce.fmul.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f64 = call double @llvm.vector.reduce.fmul.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v4f64 = call double @llvm.vector.reduce.fmul.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %fmul_v4f128 = call fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fmul_strict'
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %fmul_v2f16 = call half @llvm.vector.reduce.fmul.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fmul_v4f16 = call half @llvm.vector.reduce.fmul.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %fmul_v8f16 = call half @llvm.vector.reduce.fmul.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %fmul_v16f16 = call half @llvm.vector.reduce.fmul.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %fmul_v2f32 = call float @llvm.vector.reduce.fmul.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v4f32 = call float @llvm.vector.reduce.fmul.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v8f32 = call float @llvm.vector.reduce.fmul.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f64 = call double @llvm.vector.reduce.fmul.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v4f64 = call double @llvm.vector.reduce.fmul.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fmul_v4f128 = call fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fmul_strict'
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f16 = call half @llvm.vector.reduce.fmul.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v4f16 = call half @llvm.vector.reduce.fmul.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fmul_v8f16 = call half @llvm.vector.reduce.fmul.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fmul_v16f16 = call half @llvm.vector.reduce.fmul.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f32 = call float @llvm.vector.reduce.fmul.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v4f32 = call float @llvm.vector.reduce.fmul.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %fmul_v8f32 = call float @llvm.vector.reduce.fmul.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f64 = call double @llvm.vector.reduce.fmul.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v4f64 = call double @llvm.vector.reduce.fmul.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %fmul_v4f128 = call fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
  %fmul_v2f16 = call half @llvm.vector.reduce.fmul.v2f16(half 0.0, <2 x half> undef)
  %fmul_v4f16 = call half @llvm.vector.reduce.fmul.v4f16(half 0.0, <4 x half> undef)
  %fmul_v8f16 = call half @llvm.vector.reduce.fmul.v8f16(half 0.0, <8 x half> undef)
  %fmul_v16f16 = call half @llvm.vector.reduce.fmul.v16f16(half 0.0, <16 x half> undef)
  %fmul_v2f32 = call float @llvm.vector.reduce.fmul.v2f32(float 0.0, <2 x float> undef)
  %fmul_v4f32 = call float @llvm.vector.reduce.fmul.v4f32(float 0.0, <4 x float> undef)
  %fmul_v8f32 = call float @llvm.vector.reduce.fmul.v8f32(float 0.0, <8 x float> undef)
  %fmul_v2f64 = call double @llvm.vector.reduce.fmul.v2f64(double 0.0, <2 x double> undef)
  %fmul_v4f64 = call double @llvm.vector.reduce.fmul.v4f64(double 0.0, <4 x double> undef)
  %fmul_v4f128 = call fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
  ret void
}


define void @fmul_unordered() {
; CHECK-V8-LABEL: 'fmul_unordered'
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v2f16 = call reassoc half @llvm.vector.reduce.fmul.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %fmul_v4f16 = call reassoc half @llvm.vector.reduce.fmul.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %fmul_v8f16 = call reassoc half @llvm.vector.reduce.fmul.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %fmul_v16f16 = call reassoc half @llvm.vector.reduce.fmul.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f32 = call reassoc float @llvm.vector.reduce.fmul.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fmul_v4f32 = call reassoc float @llvm.vector.reduce.fmul.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v8f32 = call reassoc float @llvm.vector.reduce.fmul.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v2f64 = call reassoc double @llvm.vector.reduce.fmul.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %fmul_v4f64 = call reassoc double @llvm.vector.reduce.fmul.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %fmul_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-V8-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fmul_unordered'
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %fmul_v2f16 = call reassoc half @llvm.vector.reduce.fmul.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fmul_v4f16 = call reassoc half @llvm.vector.reduce.fmul.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %fmul_v8f16 = call reassoc half @llvm.vector.reduce.fmul.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %fmul_v16f16 = call reassoc half @llvm.vector.reduce.fmul.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %fmul_v2f32 = call reassoc float @llvm.vector.reduce.fmul.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %fmul_v4f32 = call reassoc float @llvm.vector.reduce.fmul.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %fmul_v8f32 = call reassoc float @llvm.vector.reduce.fmul.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fmul_v2f64 = call reassoc double @llvm.vector.reduce.fmul.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %fmul_v4f64 = call reassoc double @llvm.vector.reduce.fmul.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %fmul_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEFP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fmul_unordered'
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fmul_v2f16 = call reassoc half @llvm.vector.reduce.fmul.v2f16(half 0xH0000, <2 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %fmul_v4f16 = call reassoc half @llvm.vector.reduce.fmul.v4f16(half 0xH0000, <4 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 169 for instruction: %fmul_v8f16 = call reassoc half @llvm.vector.reduce.fmul.v8f16(half 0xH0000, <8 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 225 for instruction: %fmul_v16f16 = call reassoc half @llvm.vector.reduce.fmul.v16f16(half 0xH0000, <16 x half> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fmul_v2f32 = call reassoc float @llvm.vector.reduce.fmul.v2f32(float 0.000000e+00, <2 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %fmul_v4f32 = call reassoc float @llvm.vector.reduce.fmul.v4f32(float 0.000000e+00, <4 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: %fmul_v8f32 = call reassoc float @llvm.vector.reduce.fmul.v8f32(float 0.000000e+00, <8 x float> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %fmul_v2f64 = call reassoc double @llvm.vector.reduce.fmul.v2f64(double 0.000000e+00, <2 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %fmul_v4f64 = call reassoc double @llvm.vector.reduce.fmul.v4f64(double 0.000000e+00, <4 x double> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %fmul_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
; CHECK-MVEI-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
;
  %fmul_v2f16 = call reassoc half @llvm.vector.reduce.fmul.v2f16(half 0.0, <2 x half> undef)
  %fmul_v4f16 = call reassoc half @llvm.vector.reduce.fmul.v4f16(half 0.0, <4 x half> undef)
  %fmul_v8f16 = call reassoc half @llvm.vector.reduce.fmul.v8f16(half 0.0, <8 x half> undef)
  %fmul_v16f16 = call reassoc half @llvm.vector.reduce.fmul.v16f16(half 0.0, <16 x half> undef)
  %fmul_v2f32 = call reassoc float @llvm.vector.reduce.fmul.v2f32(float 0.0, <2 x float> undef)
  %fmul_v4f32 = call reassoc float @llvm.vector.reduce.fmul.v4f32(float 0.0, <4 x float> undef)
  %fmul_v8f32 = call reassoc float @llvm.vector.reduce.fmul.v8f32(float 0.0, <8 x float> undef)
  %fmul_v2f64 = call reassoc double @llvm.vector.reduce.fmul.v2f64(double 0.0, <2 x double> undef)
  %fmul_v4f64 = call reassoc double @llvm.vector.reduce.fmul.v4f64(double 0.0, <4 x double> undef)
  %fmul_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmul.v4f128(fp128 undef, <4 x fp128> undef)
  ret void
}


declare half @llvm.vector.reduce.fadd.v2f16(half, <2 x half>)
declare half @llvm.vector.reduce.fadd.v4f16(half, <4 x half>)
declare half @llvm.vector.reduce.fadd.v8f16(half, <8 x half>)
declare half @llvm.vector.reduce.fadd.v16f16(half, <16 x half>)
declare float @llvm.vector.reduce.fadd.v2f32(float, <2 x float>)
declare float @llvm.vector.reduce.fadd.v4f32(float, <4 x float>)
declare float @llvm.vector.reduce.fadd.v8f32(float, <8 x float>)
declare double @llvm.vector.reduce.fadd.v2f64(double, <2 x double>)
declare double @llvm.vector.reduce.fadd.v4f64(double, <4 x double>)
declare fp128 @llvm.vector.reduce.fadd.v4f128(fp128, <4 x fp128>)


declare half @llvm.vector.reduce.fmul.v2f16(half, <2 x half>)
declare half @llvm.vector.reduce.fmul.v4f16(half, <4 x half>)
declare half @llvm.vector.reduce.fmul.v8f16(half, <8 x half>)
declare half @llvm.vector.reduce.fmul.v16f16(half, <16 x half>)
declare float @llvm.vector.reduce.fmul.v2f32(float, <2 x float>)
declare float @llvm.vector.reduce.fmul.v4f32(float, <4 x float>)
declare float @llvm.vector.reduce.fmul.v8f32(float, <8 x float>)
declare double @llvm.vector.reduce.fmul.v2f64(double, <2 x double>)
declare double @llvm.vector.reduce.fmul.v4f64(double, <4 x double>)
declare fp128 @llvm.vector.reduce.fmul.v4f128(fp128, <4 x fp128>)