; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 3
; RUN: opt  -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=aarch64--linux-gnu -mattr=+sve < %s | FileCheck %s

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

define void @scalable_loads() {
; CHECK-LABEL: 'scalable_loads'
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %res.nxv8i8 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %res.nxv16i8 = load <vscale x 16 x i8>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:2 Lat:4 SizeLat:2 for: %res.nxv32i8 = load <vscale x 32 x i8>, ptr undef, align 32
; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:Invalid Lat:4 SizeLat:Invalid for: %res.nxv1i64 = load <vscale x 1 x i64>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:2 Lat:4 SizeLat:2 for: %res.nxv32i1 = load <vscale x 32 x i1>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %res.nxv16i1 = load <vscale x 16 x i1>, ptr undef, align 2
; CHECK-NEXT:  Cost Model: Found costs of RThru:Invalid CodeSize:Invalid Lat:4 SizeLat:Invalid for: %res.nxv4i1 = load <vscale x 4 x i1>, ptr undef, align 1
; CHECK-NEXT:  Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void
;
  %res.nxv8i8 = load <vscale x 8 x i8>, ptr undef
  %res.nxv16i8 = load <vscale x 16 x i8>, ptr undef
  %res.nxv32i8 = load <vscale x 32 x i8>, ptr undef
  %res.nxv1i64 = load <vscale x 1 x i64>, ptr undef
  %res.nxv32i1 = load <vscale x 32 x i1>, ptr undef
  %res.nxv16i1 = load <vscale x 16 x i1>, ptr undef
  %res.nxv4i1 = load <vscale x 4 x i1>, ptr undef
  ret void
}

define void @scalable_stores() {
; CHECK-LABEL: 'scalable_stores'
; CHECK-NEXT:  Cost Model: Found costs of 1 for: store <vscale x 8 x i8> undef, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 1 for: store <vscale x 16 x i8> undef, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:2 Lat:1 SizeLat:2 for: store <vscale x 32 x i8> undef, ptr undef, align 32
; CHECK-NEXT:  Cost Model: Found costs of Invalid for: store <vscale x 1 x i64> undef, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:2 Lat:1 SizeLat:2 for: store <vscale x 32 x i1> undef, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 1 for: store <vscale x 16 x i1> undef, ptr undef, align 2
; CHECK-NEXT:  Cost Model: Found costs of Invalid for: store <vscale x 4 x i1> undef, ptr undef, align 1
; CHECK-NEXT:  Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void
;
  store <vscale x 8 x i8> undef, ptr undef
  store <vscale x 16 x i8> undef, ptr undef
  store <vscale x 32 x i8> undef, ptr undef
  store <vscale x 1 x i64> undef, ptr undef
  store <vscale x 32 x i1> undef, ptr undef
  store <vscale x 16 x i1> undef, ptr undef
  store <vscale x 4 x i1> undef, ptr undef
  ret void
}

define void @scalable_ext_loads() {
; CHECK-LABEL: 'scalable_ext_loads'
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv16i8 = load <vscale x 16 x i8>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %zext.nxv16i8to16 = zext <vscale x 16 x i8> %load.nxv16i8 to <vscale x 16 x i16>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv8i8 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %zext.nxv8i8to16 = zext <vscale x 8 x i8> %load.nxv8i8 to <vscale x 8 x i16>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv4i8 = load <vscale x 4 x i8>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %zext.nxv4i8to32 = zext <vscale x 4 x i8> %load.nxv4i8 to <vscale x 4 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv2i8 = load <vscale x 2 x i8>, ptr undef, align 2
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %zext.nxv2i8to64 = zext <vscale x 2 x i8> %load.nxv2i8 to <vscale x 2 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv8i16 = load <vscale x 8 x i16>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %zext.nxv8i16to32 = zext <vscale x 8 x i16> %load.nxv8i16 to <vscale x 8 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv4i16 = load <vscale x 4 x i16>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %zext.nxv4i16to32 = zext <vscale x 4 x i16> %load.nxv4i16 to <vscale x 4 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv2i16 = load <vscale x 2 x i16>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %zext.nxv2i16to64 = zext <vscale x 2 x i16> %load.nxv2i16 to <vscale x 2 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv4i32 = load <vscale x 4 x i32>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %zext.nxv4i32to64 = zext <vscale x 4 x i32> %load.nxv4i32 to <vscale x 4 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv2i32 = load <vscale x 2 x i32>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %zext.nxv2i32to64 = zext <vscale x 2 x i32> %load.nxv2i32 to <vscale x 2 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv4i8.2 = load <vscale x 4 x i8>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 1 for: %zext.nxv4i8to64 = zext <vscale x 4 x i8> %load.nxv4i8.2 to <vscale x 4 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv4i16.2 = load <vscale x 4 x i16>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 1 for: %zext.nxv4i16to64 = zext <vscale x 4 x i16> %load.nxv4i16.2 to <vscale x 4 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv8i8.2 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 1 for: %zext.nxv8i8to32 = zext <vscale x 8 x i8> %load.nxv8i8.2 to <vscale x 8 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load.nxv8i8.3 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of RThru:3 CodeSize:1 Lat:1 SizeLat:1 for: %zext.nxv8i8to64 = zext <vscale x 8 x i8> %load.nxv8i8.3 to <vscale x 8 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv16i8 = load <vscale x 16 x i8>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %sext.nxv16i8to16 = sext <vscale x 16 x i8> %load2.nxv16i8 to <vscale x 16 x i16>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv8i8 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %sext.nxv8i8to16 = sext <vscale x 8 x i8> %load2.nxv8i8 to <vscale x 8 x i16>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv4i8 = load <vscale x 4 x i8>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %sext.nxv4i8to32 = sext <vscale x 4 x i8> %load2.nxv4i8 to <vscale x 4 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv2i8 = load <vscale x 2 x i8>, ptr undef, align 2
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %sext.nxv2i8to64 = sext <vscale x 2 x i8> %load2.nxv2i8 to <vscale x 2 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv8i16 = load <vscale x 8 x i16>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %sext.nxv8i16to32 = sext <vscale x 8 x i16> %load2.nxv8i16 to <vscale x 8 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv4i16 = load <vscale x 4 x i16>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %sext.nxv4i16to32 = sext <vscale x 4 x i16> %load2.nxv4i16 to <vscale x 4 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv2i16 = load <vscale x 2 x i16>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %sext.nxv2i16to64 = sext <vscale x 2 x i16> %load2.nxv2i16 to <vscale x 2 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv4i32 = load <vscale x 4 x i32>, ptr undef, align 16
; CHECK-NEXT:  Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %sext.nxv4i32to64 = sext <vscale x 4 x i32> %load2.nxv4i32 to <vscale x 4 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv2i32 = load <vscale x 2 x i32>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 0 for: %sext.nxv2i32to64 = sext <vscale x 2 x i32> %load2.nxv2i32 to <vscale x 2 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv4i8.2 = load <vscale x 4 x i8>, ptr undef, align 4
; CHECK-NEXT:  Cost Model: Found costs of 1 for: %sext.nxv4i8to64 = sext <vscale x 4 x i8> %load2.nxv4i8.2 to <vscale x 4 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv4i16.2 = load <vscale x 4 x i16>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 1 for: %sext.nxv4i16to64 = sext <vscale x 4 x i16> %load2.nxv4i16.2 to <vscale x 4 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv8i8.2 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of 1 for: %sext.nxv8i8to32 = sext <vscale x 8 x i8> %load2.nxv8i8.2 to <vscale x 8 x i32>
; CHECK-NEXT:  Cost Model: Found costs of RThru:1 CodeSize:1 Lat:4 SizeLat:1 for: %load2.nxv8i8.3 = load <vscale x 8 x i8>, ptr undef, align 8
; CHECK-NEXT:  Cost Model: Found costs of RThru:3 CodeSize:1 Lat:1 SizeLat:1 for: %sext.nxv8i8to64 = sext <vscale x 8 x i8> %load2.nxv8i8.3 to <vscale x 8 x i64>
; CHECK-NEXT:  Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void
;


  %load.nxv16i8 = load <vscale x 16 x i8>, ptr undef
  %zext.nxv16i8to16 = zext <vscale x 16 x i8> %load.nxv16i8 to <vscale x 16 x i16>
  %load.nxv8i8 = load <vscale x 8 x i8>, ptr undef
  %zext.nxv8i8to16 = zext <vscale x 8 x i8> %load.nxv8i8 to <vscale x 8 x i16>
  %load.nxv4i8 = load <vscale x 4 x i8>, ptr undef
  %zext.nxv4i8to32 = zext <vscale x 4 x i8> %load.nxv4i8 to <vscale x 4 x i32>
  %load.nxv2i8 = load <vscale x 2 x i8>, ptr undef
  %zext.nxv2i8to64 = zext <vscale x 2 x i8> %load.nxv2i8 to <vscale x 2 x i64>
  %load.nxv8i16 = load <vscale x 8 x i16>, ptr undef
  %zext.nxv8i16to32 = zext <vscale x 8 x i16> %load.nxv8i16 to <vscale x 8 x i32>
  %load.nxv4i16 = load <vscale x 4 x i16>, ptr undef
  %zext.nxv4i16to32 = zext <vscale x 4 x i16> %load.nxv4i16 to <vscale x 4 x i32>
  %load.nxv2i16 = load <vscale x 2 x i16>, ptr undef
  %zext.nxv2i16to64 = zext <vscale x 2 x i16> %load.nxv2i16 to <vscale x 2 x i64>
  %load.nxv4i32 = load <vscale x 4 x i32>, ptr undef
  %zext.nxv4i32to64 = zext <vscale x 4 x i32> %load.nxv4i32 to <vscale x 4 x i64>
  %load.nxv2i32 = load <vscale x 2 x i32>, ptr undef
  %zext.nxv2i32to64 = zext <vscale x 2 x i32> %load.nxv2i32 to <vscale x 2 x i64>
  %load.nxv4i8.2 = load <vscale x 4 x i8>, ptr undef
  %zext.nxv4i8to64 = zext <vscale x 4 x i8> %load.nxv4i8.2 to <vscale x 4 x i64>
  %load.nxv4i16.2 = load <vscale x 4 x i16>, ptr undef
  %zext.nxv4i16to64 = zext <vscale x 4 x i16> %load.nxv4i16.2 to <vscale x 4 x i64>
  %load.nxv8i8.2 = load <vscale x 8 x i8>, ptr undef
  %zext.nxv8i8to32 = zext <vscale x 8 x i8> %load.nxv8i8.2 to <vscale x 8 x i32>
  %load.nxv8i8.3 = load <vscale x 8 x i8>, ptr undef
  %zext.nxv8i8to64 = zext <vscale x 8 x i8> %load.nxv8i8.3 to <vscale x 8 x i64>

  %load2.nxv16i8 = load <vscale x 16 x i8>, ptr undef
  %sext.nxv16i8to16 = sext <vscale x 16 x i8> %load2.nxv16i8 to <vscale x 16 x i16>
  %load2.nxv8i8 = load <vscale x 8 x i8>, ptr undef
  %sext.nxv8i8to16 = sext <vscale x 8 x i8> %load2.nxv8i8 to <vscale x 8 x i16>
  %load2.nxv4i8 = load <vscale x 4 x i8>, ptr undef
  %sext.nxv4i8to32 = sext <vscale x 4 x i8> %load2.nxv4i8 to <vscale x 4 x i32>
  %load2.nxv2i8 = load <vscale x 2 x i8>, ptr undef
  %sext.nxv2i8to64 = sext <vscale x 2 x i8> %load2.nxv2i8 to <vscale x 2 x i64>
  %load2.nxv8i16 = load <vscale x 8 x i16>, ptr undef
  %sext.nxv8i16to32 = sext <vscale x 8 x i16> %load2.nxv8i16 to <vscale x 8 x i32>
  %load2.nxv4i16 = load <vscale x 4 x i16>, ptr undef
  %sext.nxv4i16to32 = sext <vscale x 4 x i16> %load2.nxv4i16 to <vscale x 4 x i32>
  %load2.nxv2i16 = load <vscale x 2 x i16>, ptr undef
  %sext.nxv2i16to64 = sext <vscale x 2 x i16> %load2.nxv2i16 to <vscale x 2 x i64>
  %load2.nxv4i32 = load <vscale x 4 x i32>, ptr undef
  %sext.nxv4i32to64 = sext <vscale x 4 x i32> %load2.nxv4i32 to <vscale x 4 x i64>
  %load2.nxv2i32 = load <vscale x 2 x i32>, ptr undef
  %sext.nxv2i32to64 = sext <vscale x 2 x i32> %load2.nxv2i32 to <vscale x 2 x i64>
  %load2.nxv4i8.2 = load <vscale x 4 x i8>, ptr undef
  %sext.nxv4i8to64 = sext <vscale x 4 x i8> %load2.nxv4i8.2 to <vscale x 4 x i64>
  %load2.nxv4i16.2 = load <vscale x 4 x i16>, ptr undef
  %sext.nxv4i16to64 = sext <vscale x 4 x i16> %load2.nxv4i16.2 to <vscale x 4 x i64>
  %load2.nxv8i8.2 = load <vscale x 8 x i8>, ptr undef
  %sext.nxv8i8to32 = sext <vscale x 8 x i8> %load2.nxv8i8.2 to <vscale x 8 x i32>
  %load2.nxv8i8.3 = load <vscale x 8 x i8>, ptr undef
  %sext.nxv8i8to64 = sext <vscale x 8 x i8> %load2.nxv8i8.3 to <vscale x 8 x i64>

  ret void
}