clang/test/OpenMP/generic_loop_codegen.cpp

   1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name tmp2 --version 2
   2 // RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp %s
   3 // RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp -emit-llvm %s -o - | FileCheck %s --check-prefix=IR
   4
   5 // Check same results after serialization round-trip
   6 // RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp -emit-pch -o %t %s
   7 // RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp -include-pch %t -emit-llvm %s -o - | FileCheck %s --check-prefix=IR-PCH
   8
   9 // expected-no-diagnostics
  10
  11 #ifndef HEADER
  12 #define HEADER
  13
  14 void foo(int t) {
  15
  16   int i, j, z;
  17   #pragma omp loop collapse(2) reduction(+:z) lastprivate(j) bind(thread)
  18   for (int i = 0; i<t; ++i)
  19     for (j = 0; j<t; ++j)
  20       z += i+j;
  21 }
  22 #endif
  23 // IR-LABEL: define dso_local void @_Z3fooi
  24 // IR-SAME: (i32 noundef [[T:%.*]]) #[[ATTR0:[0-9]+]] {
  25 // IR-NEXT:  entry:
  26 // IR-NEXT:    [[T_ADDR:%.*]] = alloca i32, align 4
  27 // IR-NEXT:    [[I:%.*]] = alloca i32, align 4
  28 // IR-NEXT:    [[J:%.*]] = alloca i32, align 4
  29 // IR-NEXT:    [[Z:%.*]] = alloca i32, align 4
  30 // IR-NEXT:    [[TMP:%.*]] = alloca i32, align 4
  31 // IR-NEXT:    [[TMP2TMP1:%.*]] = alloca i32, align 4
  32 // IR-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
  33 // IR-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
  34 // IR-NEXT:    [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
  35 // IR-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
  36 // IR-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
  37 // IR-NEXT:    [[I8:%.*]] = alloca i32, align 4
  38 // IR-NEXT:    [[J9:%.*]] = alloca i32, align 4
  39 // IR-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
  40 // IR-NEXT:    [[I11:%.*]] = alloca i32, align 4
  41 // IR-NEXT:    [[J12:%.*]] = alloca i32, align 4
  42 // IR-NEXT:    [[Z13:%.*]] = alloca i32, align 4
  43 // IR-NEXT:    store i32 [[T]], ptr [[T_ADDR]], align 4
  44 // IR-NEXT:    [[TMP0:%.*]] = load i32, ptr [[T_ADDR]], align 4
  45 // IR-NEXT:    store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
  46 // IR-NEXT:    [[TMP1:%.*]] = load i32, ptr [[T_ADDR]], align 4
  47 // IR-NEXT:    store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_2]], align 4
  48 // IR-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
  49 // IR-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP2]], 0
  50 // IR-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
  51 // IR-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
  52 // IR-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
  53 // IR-NEXT:    [[SUB4:%.*]] = sub nsw i32 [[TMP3]], 0
  54 // IR-NEXT:    [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
  55 // IR-NEXT:    [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
  56 // IR-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
  57 // IR-NEXT:    [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
  58 // IR-NEXT:    store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
  59 // IR-NEXT:    store i64 0, ptr [[DOTOMP_LB]], align 8
  60 // IR-NEXT:    [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
  61 // IR-NEXT:    store i64 [[TMP4]], ptr [[DOTOMP_UB]], align 8
  62 // IR-NEXT:    store i32 0, ptr [[I8]], align 4
  63 // IR-NEXT:    store i32 0, ptr [[J9]], align 4
  64 // IR-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
  65 // IR-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
  66 // IR-NEXT:    br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[SIMD_IF_END:%.*]]
  67 // IR:       land.lhs.true:
  68 // IR-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
  69 // IR-NEXT:    [[CMP10:%.*]] = icmp slt i32 0, [[TMP6]]
  70 // IR-NEXT:    br i1 [[CMP10]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
  71 // IR:       simd.if.then:
  72 // IR-NEXT:    [[TMP7:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
  73 // IR-NEXT:    store i64 [[TMP7]], ptr [[DOTOMP_IV]], align 8
  74 // IR-NEXT:    store i32 0, ptr [[Z13]], align 4
  75 // IR-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
  76 // IR:       omp.inner.for.cond:
  77 // IR-NEXT:    [[TMP8:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3:![0-9]+]]
  78 // IR-NEXT:    [[TMP9:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP3]]
  79 // IR-NEXT:    [[CMP14:%.*]] = icmp sle i64 [[TMP8]], [[TMP9]]
  80 // IR-NEXT:    br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
  81 // IR:       omp.inner.for.body:
  82 // IR-NEXT:    [[TMP10:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
  83 // IR-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
  84 // IR-NEXT:    [[SUB15:%.*]] = sub nsw i32 [[TMP11]], 0
  85 // IR-NEXT:    [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
  86 // IR-NEXT:    [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
  87 // IR-NEXT:    [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
  88 // IR-NEXT:    [[DIV19:%.*]] = sdiv i64 [[TMP10]], [[CONV18]]
  89 // IR-NEXT:    [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
  90 // IR-NEXT:    [[ADD21:%.*]] = add nsw i64 0, [[MUL20]]
  91 // IR-NEXT:    [[CONV22:%.*]] = trunc i64 [[ADD21]] to i32
  92 // IR-NEXT:    store i32 [[CONV22]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP3]]
  93 // IR-NEXT:    [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
  94 // IR-NEXT:    [[TMP13:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
  95 // IR-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
  96 // IR-NEXT:    [[SUB23:%.*]] = sub nsw i32 [[TMP14]], 0
  97 // IR-NEXT:    [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
  98 // IR-NEXT:    [[MUL25:%.*]] = mul nsw i32 1, [[DIV24]]
  99 // IR-NEXT:    [[CONV26:%.*]] = sext i32 [[MUL25]] to i64
 100 // IR-NEXT:    [[DIV27:%.*]] = sdiv i64 [[TMP13]], [[CONV26]]
 101 // IR-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
 102 // IR-NEXT:    [[SUB28:%.*]] = sub nsw i32 [[TMP15]], 0
 103 // IR-NEXT:    [[DIV29:%.*]] = sdiv i32 [[SUB28]], 1
 104 // IR-NEXT:    [[MUL30:%.*]] = mul nsw i32 1, [[DIV29]]
 105 // IR-NEXT:    [[CONV31:%.*]] = sext i32 [[MUL30]] to i64
 106 // IR-NEXT:    [[MUL32:%.*]] = mul nsw i64 [[DIV27]], [[CONV31]]
 107 // IR-NEXT:    [[SUB33:%.*]] = sub nsw i64 [[TMP12]], [[MUL32]]
 108 // IR-NEXT:    [[MUL34:%.*]] = mul nsw i64 [[SUB33]], 1
 109 // IR-NEXT:    [[ADD35:%.*]] = add nsw i64 0, [[MUL34]]
 110 // IR-NEXT:    [[CONV36:%.*]] = trunc i64 [[ADD35]] to i32
 111 // IR-NEXT:    store i32 [[CONV36]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP3]]
 112 // IR-NEXT:    [[TMP16:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP3]]
 113 // IR-NEXT:    [[TMP17:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP3]]
 114 // IR-NEXT:    [[ADD37:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
 115 // IR-NEXT:    [[TMP18:%.*]] = load i32, ptr [[Z13]], align 4, !llvm.access.group [[ACC_GRP3]]
 116 // IR-NEXT:    [[ADD38:%.*]] = add nsw i32 [[TMP18]], [[ADD37]]
 117 // IR-NEXT:    store i32 [[ADD38]], ptr [[Z13]], align 4, !llvm.access.group [[ACC_GRP3]]
 118 // IR-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 119 // IR:       omp.body.continue:
 120 // IR-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 121 // IR:       omp.inner.for.inc:
 122 // IR-NEXT:    [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 123 // IR-NEXT:    [[ADD39:%.*]] = add nsw i64 [[TMP19]], 1
 124 // IR-NEXT:    store i64 [[ADD39]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 125 // IR-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
 126 // IR:       omp.inner.for.end:
 127 // IR-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
 128 // IR-NEXT:    [[SUB40:%.*]] = sub nsw i32 [[TMP20]], 0
 129 // IR-NEXT:    [[DIV41:%.*]] = sdiv i32 [[SUB40]], 1
 130 // IR-NEXT:    [[MUL42:%.*]] = mul nsw i32 [[DIV41]], 1
 131 // IR-NEXT:    [[ADD43:%.*]] = add nsw i32 0, [[MUL42]]
 132 // IR-NEXT:    store i32 [[ADD43]], ptr [[I11]], align 4
 133 // IR-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
 134 // IR-NEXT:    [[SUB44:%.*]] = sub nsw i32 [[TMP21]], 0
 135 // IR-NEXT:    [[DIV45:%.*]] = sdiv i32 [[SUB44]], 1
 136 // IR-NEXT:    [[MUL46:%.*]] = mul nsw i32 [[DIV45]], 1
 137 // IR-NEXT:    [[ADD47:%.*]] = add nsw i32 0, [[MUL46]]
 138 // IR-NEXT:    store i32 [[ADD47]], ptr [[J]], align 4
 139 // IR-NEXT:    [[TMP22:%.*]] = load i32, ptr [[Z]], align 4
 140 // IR-NEXT:    [[TMP23:%.*]] = load i32, ptr [[Z13]], align 4
 141 // IR-NEXT:    [[ADD48:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
 142 // IR-NEXT:    store i32 [[ADD48]], ptr [[Z]], align 4
 143 // IR-NEXT:    br label [[SIMD_IF_END]]
 144 // IR:       simd.if.end:
 145 // IR-NEXT:    ret void
 146 //
 147 //
 148 // IR-PCH-LABEL: define dso_local void @_Z3fooi
 149 // IR-PCH-SAME: (i32 noundef [[T:%.*]]) #[[ATTR0:[0-9]+]] {
 150 // IR-PCH-NEXT:  entry:
 151 // IR-PCH-NEXT:    [[T_ADDR:%.*]] = alloca i32, align 4
 152 // IR-PCH-NEXT:    [[I:%.*]] = alloca i32, align 4
 153 // IR-PCH-NEXT:    [[J:%.*]] = alloca i32, align 4
 154 // IR-PCH-NEXT:    [[Z:%.*]] = alloca i32, align 4
 155 // IR-PCH-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 156 // IR-PCH-NEXT:    [[TMP2TMP1:%.*]] = alloca i32, align 4
 157 // IR-PCH-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
 158 // IR-PCH-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
 159 // IR-PCH-NEXT:    [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
 160 // IR-PCH-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
 161 // IR-PCH-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
 162 // IR-PCH-NEXT:    [[I8:%.*]] = alloca i32, align 4
 163 // IR-PCH-NEXT:    [[J9:%.*]] = alloca i32, align 4
 164 // IR-PCH-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
 165 // IR-PCH-NEXT:    [[I11:%.*]] = alloca i32, align 4
 166 // IR-PCH-NEXT:    [[J12:%.*]] = alloca i32, align 4
 167 // IR-PCH-NEXT:    [[Z13:%.*]] = alloca i32, align 4
 168 // IR-PCH-NEXT:    store i32 [[T]], ptr [[T_ADDR]], align 4
 169 // IR-PCH-NEXT:    [[TMP0:%.*]] = load i32, ptr [[T_ADDR]], align 4
 170 // IR-PCH-NEXT:    store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4
 171 // IR-PCH-NEXT:    [[TMP1:%.*]] = load i32, ptr [[T_ADDR]], align 4
 172 // IR-PCH-NEXT:    store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_2]], align 4
 173 // IR-PCH-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
 174 // IR-PCH-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP2]], 0
 175 // IR-PCH-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
 176 // IR-PCH-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
 177 // IR-PCH-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
 178 // IR-PCH-NEXT:    [[SUB4:%.*]] = sub nsw i32 [[TMP3]], 0
 179 // IR-PCH-NEXT:    [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
 180 // IR-PCH-NEXT:    [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
 181 // IR-PCH-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
 182 // IR-PCH-NEXT:    [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
 183 // IR-PCH-NEXT:    store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
 184 // IR-PCH-NEXT:    store i64 0, ptr [[DOTOMP_LB]], align 8
 185 // IR-PCH-NEXT:    [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
 186 // IR-PCH-NEXT:    store i64 [[TMP4]], ptr [[DOTOMP_UB]], align 8
 187 // IR-PCH-NEXT:    store i32 0, ptr [[I8]], align 4
 188 // IR-PCH-NEXT:    store i32 0, ptr [[J9]], align 4
 189 // IR-PCH-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
 190 // IR-PCH-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
 191 // IR-PCH-NEXT:    br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[SIMD_IF_END:%.*]]
 192 // IR-PCH:       land.lhs.true:
 193 // IR-PCH-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
 194 // IR-PCH-NEXT:    [[CMP10:%.*]] = icmp slt i32 0, [[TMP6]]
 195 // IR-PCH-NEXT:    br i1 [[CMP10]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
 196 // IR-PCH:       simd.if.then:
 197 // IR-PCH-NEXT:    [[TMP7:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
 198 // IR-PCH-NEXT:    store i64 [[TMP7]], ptr [[DOTOMP_IV]], align 8
 199 // IR-PCH-NEXT:    store i32 0, ptr [[Z13]], align 4
 200 // IR-PCH-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 201 // IR-PCH:       omp.inner.for.cond:
 202 // IR-PCH-NEXT:    [[TMP8:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3:![0-9]+]]
 203 // IR-PCH-NEXT:    [[TMP9:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP3]]
 204 // IR-PCH-NEXT:    [[CMP14:%.*]] = icmp sle i64 [[TMP8]], [[TMP9]]
 205 // IR-PCH-NEXT:    br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 206 // IR-PCH:       omp.inner.for.body:
 207 // IR-PCH-NEXT:    [[TMP10:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 208 // IR-PCH-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
 209 // IR-PCH-NEXT:    [[SUB15:%.*]] = sub nsw i32 [[TMP11]], 0
 210 // IR-PCH-NEXT:    [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
 211 // IR-PCH-NEXT:    [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
 212 // IR-PCH-NEXT:    [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
 213 // IR-PCH-NEXT:    [[DIV19:%.*]] = sdiv i64 [[TMP10]], [[CONV18]]
 214 // IR-PCH-NEXT:    [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
 215 // IR-PCH-NEXT:    [[ADD21:%.*]] = add nsw i64 0, [[MUL20]]
 216 // IR-PCH-NEXT:    [[CONV22:%.*]] = trunc i64 [[ADD21]] to i32
 217 // IR-PCH-NEXT:    store i32 [[CONV22]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP3]]
 218 // IR-PCH-NEXT:    [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 219 // IR-PCH-NEXT:    [[TMP13:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 220 // IR-PCH-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
 221 // IR-PCH-NEXT:    [[SUB23:%.*]] = sub nsw i32 [[TMP14]], 0
 222 // IR-PCH-NEXT:    [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
 223 // IR-PCH-NEXT:    [[MUL25:%.*]] = mul nsw i32 1, [[DIV24]]
 224 // IR-PCH-NEXT:    [[CONV26:%.*]] = sext i32 [[MUL25]] to i64
 225 // IR-PCH-NEXT:    [[DIV27:%.*]] = sdiv i64 [[TMP13]], [[CONV26]]
 226 // IR-PCH-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
 227 // IR-PCH-NEXT:    [[SUB28:%.*]] = sub nsw i32 [[TMP15]], 0
 228 // IR-PCH-NEXT:    [[DIV29:%.*]] = sdiv i32 [[SUB28]], 1
 229 // IR-PCH-NEXT:    [[MUL30:%.*]] = mul nsw i32 1, [[DIV29]]
 230 // IR-PCH-NEXT:    [[CONV31:%.*]] = sext i32 [[MUL30]] to i64
 231 // IR-PCH-NEXT:    [[MUL32:%.*]] = mul nsw i64 [[DIV27]], [[CONV31]]
 232 // IR-PCH-NEXT:    [[SUB33:%.*]] = sub nsw i64 [[TMP12]], [[MUL32]]
 233 // IR-PCH-NEXT:    [[MUL34:%.*]] = mul nsw i64 [[SUB33]], 1
 234 // IR-PCH-NEXT:    [[ADD35:%.*]] = add nsw i64 0, [[MUL34]]
 235 // IR-PCH-NEXT:    [[CONV36:%.*]] = trunc i64 [[ADD35]] to i32
 236 // IR-PCH-NEXT:    store i32 [[CONV36]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP3]]
 237 // IR-PCH-NEXT:    [[TMP16:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP3]]
 238 // IR-PCH-NEXT:    [[TMP17:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP3]]
 239 // IR-PCH-NEXT:    [[ADD37:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
 240 // IR-PCH-NEXT:    [[TMP18:%.*]] = load i32, ptr [[Z13]], align 4, !llvm.access.group [[ACC_GRP3]]
 241 // IR-PCH-NEXT:    [[ADD38:%.*]] = add nsw i32 [[TMP18]], [[ADD37]]
 242 // IR-PCH-NEXT:    store i32 [[ADD38]], ptr [[Z13]], align 4, !llvm.access.group [[ACC_GRP3]]
 243 // IR-PCH-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 244 // IR-PCH:       omp.body.continue:
 245 // IR-PCH-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 246 // IR-PCH:       omp.inner.for.inc:
 247 // IR-PCH-NEXT:    [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 248 // IR-PCH-NEXT:    [[ADD39:%.*]] = add nsw i64 [[TMP19]], 1
 249 // IR-PCH-NEXT:    store i64 [[ADD39]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
 250 // IR-PCH-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
 251 // IR-PCH:       omp.inner.for.end:
 252 // IR-PCH-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
 253 // IR-PCH-NEXT:    [[SUB40:%.*]] = sub nsw i32 [[TMP20]], 0
 254 // IR-PCH-NEXT:    [[DIV41:%.*]] = sdiv i32 [[SUB40]], 1
 255 // IR-PCH-NEXT:    [[MUL42:%.*]] = mul nsw i32 [[DIV41]], 1
 256 // IR-PCH-NEXT:    [[ADD43:%.*]] = add nsw i32 0, [[MUL42]]
 257 // IR-PCH-NEXT:    store i32 [[ADD43]], ptr [[I11]], align 4
 258 // IR-PCH-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
 259 // IR-PCH-NEXT:    [[SUB44:%.*]] = sub nsw i32 [[TMP21]], 0
 260 // IR-PCH-NEXT:    [[DIV45:%.*]] = sdiv i32 [[SUB44]], 1
 261 // IR-PCH-NEXT:    [[MUL46:%.*]] = mul nsw i32 [[DIV45]], 1
 262 // IR-PCH-NEXT:    [[ADD47:%.*]] = add nsw i32 0, [[MUL46]]
 263 // IR-PCH-NEXT:    store i32 [[ADD47]], ptr [[J]], align 4
 264 // IR-PCH-NEXT:    [[TMP22:%.*]] = load i32, ptr [[Z]], align 4
 265 // IR-PCH-NEXT:    [[TMP23:%.*]] = load i32, ptr [[Z13]], align 4
 266 // IR-PCH-NEXT:    [[ADD48:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
 267 // IR-PCH-NEXT:    store i32 [[ADD48]], ptr [[Z]], align 4
 268 // IR-PCH-NEXT:    br label [[SIMD_IF_END]]
 269 // IR-PCH:       simd.if.end:
 270 // IR-PCH-NEXT:    ret void
 271 //