; Function Attrs: nounwind
define void @__OpenCL_k02g_kernel(i32 addrspace(1)* noalias %g_output, %2 addrspace(2)* %g_header, %3 addrspace(1)* noalias %g_dag, i64 %start_nonce, i64 %target, i32 %isolate, i32 %dagsize, i32 %dm, i32 %dv, i32 %dmask, i32 addrspace(2)* noalias %d_header, i32 %d_ExtraNonce, i32 %DcrPerItem) #0 {
entry:
  %g_output.addr = alloca i32 addrspace(1)*, align 8
  %g_header.addr = alloca %2 addrspace(2)*, align 8
  %g_dag.addr = alloca %3 addrspace(1)*, align 8
  %start_nonce.addr = alloca i64, align 8
  %target.addr = alloca i64, align 8
  %isolate.addr = alloca i32, align 4
  %dagsize.addr = alloca i32, align 4
  %dm.addr = alloca i32, align 4
  %dv.addr = alloca i32, align 4
  %dmask.addr = alloca i32, align 4
  %d_header.addr = alloca i32 addrspace(2)*, align 8
  %d_ExtraNonce.addr = alloca i32, align 4
  %DcrPerItem.addr = alloca i32, align 4
  %loc_arr = alloca [576 x i64] addrspace(3)*, align 8
  %state = alloca [25 x i64], align 8
  %gid = alloca i32, align 4
  %lar = alloca i64 addrspace(3)*, align 8
  %n = alloca i32, align 4
  %dpi = alloca i32, align 4
  %curstep = alloca i32, align 4
  %i = alloca i32, align 4
  %share = alloca %0 addrspace(3)*, align 8
  %i49 = alloca i32, align 4
  %i74 = alloca i32, align 4
  %morebit = alloca i32, align 4
  %thread_id = alloca i32, align 4
  %hash_id = alloca i32, align 4
  %i105 = alloca i32, align 4
  %i117 = alloca i32, align 4
  %mix = alloca <4 x i32>, align 16
  %share0 = alloca i32 addrspace(3)*, align 8
  %init0 = alloca i32, align 4
  %a = alloca i32, align 4
  %update_share = alloca i32, align 4
  %i181 = alloca i32, align 4
  %v = alloca i32, align 4
  %ml = alloca i32, align 4
  %aml = alloca i32, align 4
  %res = alloca i32, align 4
  %i266 = alloca i32, align 4
  %i294 = alloca i32, align 4
  %tmp319 = alloca <8 x i8>, align 8
  %slot = alloca i32, align 4
  store i32 addrspace(1)* %g_output, i32 addrspace(1)** %g_output.addr, align 8
  store %2 addrspace(2)* %g_header, %2 addrspace(2)** %g_header.addr, align 8
  store %3 addrspace(1)* %g_dag, %3 addrspace(1)** %g_dag.addr, align 8
  store i64 %start_nonce, i64* %start_nonce.addr, align 8
  store i64 %target, i64* %target.addr, align 8
  store i32 %isolate, i32* %isolate.addr, align 4
  store i32 %dagsize, i32* %dagsize.addr, align 4
  store i32 %dm, i32* %dm.addr, align 4
  store i32 %dv, i32* %dv.addr, align 4
  store i32 %dmask, i32* %dmask.addr, align 4
  store i32 addrspace(2)* %d_header, i32 addrspace(2)** %d_header.addr, align 8
  store i32 %d_ExtraNonce, i32* %d_ExtraNonce.addr, align 4
  store i32 %DcrPerItem, i32* %DcrPerItem.addr, align 4
  store [576 x i64] addrspace(3)* @k02g_cllocal_loc_arr, [576 x i64] addrspace(3)** %loc_arr, align 8
  call void @barrier(i32 2) #0
  %tmp = load i32 addrspace(1)** %g_output.addr, align 8
  %arrayidx = getelementptr i32 addrspace(1)* %tmp, i64 0
  %tmp1 = load volatile i32 addrspace(1)* %arrayidx, align 4
  %tmp2 = and i32 %tmp1, 15
  %tobool = icmp ne i32 %tmp2, 0
  br i1 %tobool, label %if.then, label %if.end

return:                                           ; preds = %if.end341, %for.exit, %if.then
  ret void

if.end:                                           ; preds = %entry
  %call = call i64 @get_global_id(i32 0) #0
  %conv = trunc i64 %call to i32
  store i32 %conv, i32* %gid, align 4
  %tmp3 = load [576 x i64] addrspace(3)** %loc_arr, align 8
  %ptr = getelementptr inbounds [576 x i64] addrspace(3)* %tmp3, i32 0, i32 0
  %call4 = call i64 @get_local_id(i32 0) #0
  %tmp5 = mul i64 9, %call4
  %tmp6 = getelementptr i64 addrspace(3)* %ptr, i64 %tmp5
  store i64 addrspace(3)* %tmp6, i64 addrspace(3)** %lar, align 8
  %tmp7 = load i32* %gid, align 4
  %tmp8 = udiv i32 %tmp7, 64
  store i32 %tmp8, i32* %n, align 4
  %tmp9 = load i32* %n, align 4
  %tmp10 = urem i32 %tmp9, 3
  %cmp = icmp eq i32 %tmp10, 0
  br i1 %cmp, label %if.then12, label %if.end11

if.then:                                          ; preds = %entry
  br label %return

if.end11:                                         ; preds = %if.end
  %tmp46 = load [576 x i64] addrspace(3)** %loc_arr, align 8
  %ptr47 = getelementptr inbounds [576 x i64] addrspace(3)* %tmp46, i32 0, i32 0
  %conv48 = bitcast i64 addrspace(3)* %ptr47 to %0 addrspace(3)*
  store %0 addrspace(3)* %conv48, %0 addrspace(3)** %share, align 8
  store i32 0, i32* %i49, align 4
  br label %for.cond50

if.then12:                                        ; preds = %if.end
  %tmp13 = load i32* %DcrPerItem.addr, align 4
  %tmp14 = add i32 %tmp13, 16
  %tmp15 = udiv i32 %tmp14, 16
  store i32 %tmp15, i32* %dpi, align 4
  %tmp16 = load i32* %gid, align 4
  %tmp17 = mul i32 16, %tmp16
  %conv18 = zext i32 %tmp17 to i64
  %call19 = call i64 @get_global_size(i32 0) #0
  %tmp20 = udiv i64 %conv18, %call19
  %conv21 = trunc i64 %tmp20 to i32
  store i32 %conv21, i32* %curstep, align 4
  %tmp22 = load i32* %curstep, align 4
  %tmp23 = load i32* %DcrPerItem.addr, align 4
  %tmp24 = and i32 %tmp23, 15
  %cmp25 = icmp ult i32 %tmp22, %tmp24
  br i1 %cmp25, label %if.then27, label %if.end26

if.end26:                                         ; preds = %if.then27, %if.then12
  store i32 0, i32* %i, align 4
  br label %for.cond

if.then27:                                        ; preds = %if.then12
  %tmp28 = load i32* %dpi, align 4
  %tmp29 = add i32 %tmp28, 1
  store i32 %tmp29, i32* %dpi, align 4
  br label %if.end26

for.cond:                                         ; preds = %for.inc, %if.end26
  %tmp30 = load i32* %i, align 4
  %tmp31 = load i32* %dpi, align 4
  %cmp32 = icmp ult i32 %tmp30, %tmp31
  br i1 %cmp32, label %for.body, label %for.exit

for.exit:                                         ; preds = %for.cond
  br label %return

for.body:                                         ; preds = %for.cond
  %tmp33 = load i32 addrspace(1)** %g_output.addr, align 8
  %tmp34 = load i32 addrspace(2)** %d_header.addr, align 8
  %tmp35 = load i32* %gid, align 4
  %conv36 = zext i32 %tmp35 to i64
  %tmp37 = load i32* %i, align 4
  %conv38 = zext i32 %tmp37 to i64
  %call39 = call i64 @get_global_size(i32 0) #0
  %tmp40 = mul i64 %conv38, %call39
  %tmp41 = add i64 %conv36, %tmp40
  %conv42 = trunc i64 %tmp41 to i32
  %tmp43 = load i64 addrspace(3)** %lar, align 8
  call void @lbc_search(i32 addrspace(1)* noalias %tmp33, i32 addrspace(2)* %tmp34, i32 %conv42, i64 addrspace(3)* %tmp43) #0
  br label %for.inc

for.inc:                                          ; preds = %for.body
  %tmp44 = load i32* %i, align 4
  %tmp45 = add i32 %tmp44, 1
  store i32 %tmp45, i32* %i, align 4
  br label %for.cond

for.cond50:                                       ; preds = %for.inc55, %if.end11
  %tmp52 = load i32* %i49, align 4
  %cmp53 = icmp ne i32 %tmp52, 4
  br i1 %cmp53, label %for.body54, label %for.exit51

for.exit51:                                       ; preds = %for.cond50
  %ptr68 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %arrayidx69 = getelementptr i64* %ptr68, i64 4
  %tmp70 = load i64* %start_nonce.addr, align 8
  %tmp71 = load i32* %gid, align 4
  %conv72 = zext i32 %tmp71 to i64
  %tmp73 = add i64 %tmp70, %conv72
  store i64 %tmp73, i64* %arrayidx69, align 8
  store i32 6, i32* %i74, align 4
  br label %for.cond75

for.body54:                                       ; preds = %for.cond50
  %ptr56 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %tmp57 = load i32* %i49, align 4
  %conv58 = zext i32 %tmp57 to i64
  %arrayidx59 = getelementptr i64* %ptr56, i64 %conv58
  %tmp60 = load %2 addrspace(2)** %g_header.addr, align 8
  %structele = getelementptr inbounds %2 addrspace(2)* %tmp60, i32 0, i32 0
  %ptr61 = getelementptr inbounds [4 x i64] addrspace(2)* %structele, i32 0, i32 0
  %tmp62 = load i32* %i49, align 4
  %conv63 = zext i32 %tmp62 to i64
  %arrayidx64 = getelementptr i64 addrspace(2)* %ptr61, i64 %conv63
  %tmp65 = load i64 addrspace(2)* %arrayidx64, align 8
  store i64 %tmp65, i64* %arrayidx59, align 8
  br label %for.inc55

for.inc55:                                        ; preds = %for.body54
  %tmp66 = load i32* %i49, align 4
  %tmp67 = add i32 %tmp66, 1
  store i32 %tmp67, i32* %i49, align 4
  br label %for.cond50

for.cond75:                                       ; preds = %for.inc80, %for.exit51
  %tmp77 = load i32* %i74, align 4
  %cmp78 = icmp ne i32 %tmp77, 25
  br i1 %cmp78, label %for.body79, label %for.exit76

for.exit76:                                       ; preds = %for.cond75
  %ptr87 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %arrayidx88 = getelementptr i64* %ptr87, i64 5
  store i64 1, i64* %arrayidx88, align 8
  %ptr89 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %arrayidx90 = getelementptr i64* %ptr89, i64 8
  store i64 -9223372036854775808, i64* %arrayidx90, align 8
  %ptr91 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %conv92 = bitcast i64* %ptr91 to <2 x i32>*
  %tmp93 = load i32* %isolate.addr, align 4
  call void @keccak_f1600_no_absorb(<2 x i32>* %conv92, i32 8, i32 %tmp93) #0
  %tmp94 = load i32* %dm.addr, align 4
  %tmp95 = sub i32 32, %tmp94
  %tmp96 = and i32 %tmp95, 63
  %tmp97 = zext i32 %tmp96 to i64
  %tmp98 = shl i64 1, %tmp97
  %conv99 = trunc i64 %tmp98 to i32
  store i32 %conv99, i32* %morebit, align 4
  %tmp100 = load i32* %gid, align 4
  %tmp101 = and i32 %tmp100, 7
  store i32 %tmp101, i32* %thread_id, align 4
  %tmp102 = load i32* %gid, align 4
  %tmp103 = urem i32 %tmp102, 64
  %tmp104 = lshr i32 %tmp103, 3
  store i32 %tmp104, i32* %hash_id, align 4
  store i32 0, i32* %i105, align 4
  br label %for.cond106

for.body79:                                       ; preds = %for.cond75
  %ptr81 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %tmp82 = load i32* %i74, align 4
  %conv83 = zext i32 %tmp82 to i64
  %arrayidx84 = getelementptr i64* %ptr81, i64 %conv83
  store i64 0, i64* %arrayidx84, align 8
  br label %for.inc80

for.inc80:                                        ; preds = %for.body79
  %tmp85 = load i32* %i74, align 4
  %tmp86 = add i32 %tmp85, 1
  store i32 %tmp86, i32* %i74, align 4
  br label %for.cond75

for.cond106:                                      ; preds = %for.inc111, %for.exit76
  %tmp108 = load i32* %i105, align 4
  %cmp109 = icmp slt i32 %tmp108, 8
  br i1 %cmp109, label %for.body110, label %for.exit107

for.exit107:                                      ; preds = %for.cond106
  store i32 13, i32* %i294, align 4
  br label %for.cond295

for.body110:                                      ; preds = %for.cond106
  %tmp112 = load i32* %i105, align 4
  %tmp113 = load i32* %thread_id, align 4
  %cmp114 = icmp eq i32 %tmp112, %tmp113
  br i1 %cmp114, label %if.then116, label %if.end115

for.inc111:                                       ; preds = %if.end264
  %tmp292 = load i32* %i105, align 4
  %tmp293 = add nsw i32 %tmp292, 1
  store i32 %tmp293, i32* %i105, align 4
  br label %for.cond106

if.end115:                                        ; preds = %for.exit119, %for.body110
  call void @barrier(i32 1) #0
  %tmp141 = load %0 addrspace(3)** %share, align 8
  %tmp142 = load i32* %hash_id, align 4
  %conv143 = zext i32 %tmp142 to i64
  %arrayidx144 = getelementptr %0 addrspace(3)* %tmp141, i64 %conv143
  %structele145 = getelementptr inbounds %0 addrspace(3)* %arrayidx144, i32 0, i32 0
  %ptr146 = getelementptr inbounds [4 x <4 x i32>] addrspace(3)* %structele145, i32 0, i32 0
  %tmp147 = load i32* %thread_id, align 4
  %tmp148 = and i32 %tmp147, 3
  %conv149 = zext i32 %tmp148 to i64
  %arrayidx150 = getelementptr <4 x i32> addrspace(3)* %ptr146, i64 %conv149
  %tmp151 = load <4 x i32> addrspace(3)* %arrayidx150, align 16
  store <4 x i32> %tmp151, <4 x i32>* %mix, align 16
  call void @barrier(i32 1) #0
  %tmp152 = load %0 addrspace(3)** %share, align 8
  %tmp153 = load i32* %hash_id, align 4
  %conv154 = zext i32 %tmp153 to i64
  %arrayidx155 = getelementptr %0 addrspace(3)* %tmp152, i64 %conv154
  %structele156 = getelementptr inbounds %0 addrspace(3)* %arrayidx155, i32 0, i32 0
  %tmp157 = bitcast [4 x <4 x i32>] addrspace(3)* %structele156 to [16 x i32] addrspace(3)*
  %ptr158 = getelementptr inbounds [16 x i32] addrspace(3)* %tmp157, i32 0, i32 0
  store i32 addrspace(3)* %ptr158, i32 addrspace(3)** %share0, align 8
  %tmp159 = load i32* %thread_id, align 4
  %cmp160 = icmp eq i32 %tmp159, 0
  br i1 %cmp160, label %if.then162, label %if.end161

if.then116:                                       ; preds = %for.body110
  store i32 0, i32* %i117, align 4
  br label %for.cond118

for.cond118:                                      ; preds = %for.inc123, %if.then116
  %tmp120 = load i32* %i117, align 4
  %cmp121 = icmp ne i32 %tmp120, 8
  br i1 %cmp121, label %for.body122, label %for.exit119

for.exit119:                                      ; preds = %for.cond118
  br label %if.end115

for.body122:                                      ; preds = %for.cond118
  %tmp124 = load %0 addrspace(3)** %share, align 8
  %tmp125 = load i32* %hash_id, align 4
  %conv126 = zext i32 %tmp125 to i64
  %arrayidx127 = getelementptr %0 addrspace(3)* %tmp124, i64 %conv126
  %structele128 = getelementptr inbounds %0 addrspace(3)* %arrayidx127, i32 0, i32 0
  %tmp129 = bitcast [4 x <4 x i32>] addrspace(3)* %structele128 to [8 x i64] addrspace(3)*
  %ptr130 = getelementptr inbounds [8 x i64] addrspace(3)* %tmp129, i32 0, i32 0
  %tmp131 = load i32* %i117, align 4
  %conv132 = zext i32 %tmp131 to i64
  %arrayidx133 = getelementptr i64 addrspace(3)* %ptr130, i64 %conv132
  %ptr134 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %tmp135 = load i32* %i117, align 4
  %conv136 = zext i32 %tmp135 to i64
  %arrayidx137 = getelementptr i64* %ptr134, i64 %conv136
  %tmp138 = load i64* %arrayidx137, align 8
  store i64 %tmp138, i64 addrspace(3)* %arrayidx133, align 8
  br label %for.inc123

for.inc123:                                       ; preds = %for.body122
  %tmp139 = load i32* %i117, align 4
  %tmp140 = add i32 %tmp139, 1
  store i32 %tmp140, i32* %i117, align 4
  br label %for.cond118

if.end161:                                        ; preds = %if.then162, %if.end115
  call void @barrier(i32 1) #0
  %tmp166 = load i32 addrspace(3)** %share0, align 8
  %tmp167 = load i32 addrspace(3)* %tmp166, align 4
  store i32 %tmp167, i32* %init0, align 4
  store i32 0, i32* %a, align 4
  br label %for.cond168

if.then162:                                       ; preds = %if.end115
  %tmp163 = load i32 addrspace(3)** %share0, align 8
  %tmp164 = load <4 x i32>* %mix, align 16
  %tmp165 = extractelement <4 x i32> %tmp164, i32 0
  store i32 %tmp165, i32 addrspace(3)* %tmp163, align 4
  br label %if.end161

for.cond168:                                      ; preds = %for.inc173, %if.end161
  %tmp170 = load i32* %a, align 4
  %cmp171 = icmp ult i32 %tmp170, 64
  br i1 %cmp171, label %for.body172, label %for.exit169

for.exit169:                                      ; preds = %for.cond168
  %tmp248 = load %0 addrspace(3)** %share, align 8
  %tmp249 = load i32* %hash_id, align 4
  %conv250 = zext i32 %tmp249 to i64
  %arrayidx251 = getelementptr %0 addrspace(3)* %tmp248, i64 %conv250
  %structele252 = getelementptr inbounds %0 addrspace(3)* %arrayidx251, i32 0, i32 0
  %tmp253 = bitcast [4 x <4 x i32>] addrspace(3)* %structele252 to [16 x i32] addrspace(3)*
  %ptr254 = getelementptr inbounds [16 x i32] addrspace(3)* %tmp253, i32 0, i32 0
  %tmp255 = load i32* %thread_id, align 4
  %conv256 = zext i32 %tmp255 to i64
  %arrayidx257 = getelementptr i32 addrspace(3)* %ptr254, i64 %conv256
  %tmp258 = load <4 x i32>* %mix, align 16
  %call259 = call i32 @fnv_reduce(<4 x i32> %tmp258) #0
  store i32 %call259, i32 addrspace(3)* %arrayidx257, align 4
  call void @barrier(i32 1) #0
  %tmp260 = load i32* %i105, align 4
  %tmp261 = load i32* %thread_id, align 4
  %cmp262 = icmp eq i32 %tmp260, %tmp261
  br i1 %cmp262, label %if.then265, label %if.end264

for.body172:                                      ; preds = %for.cond168
  %tmp174 = load i32* %thread_id, align 4
  %tmp175 = load i32* %a, align 4
  %tmp176 = lshr i32 %tmp175, 2
  %tmp177 = and i32 %tmp176, 7
  %cmp178 = icmp eq i32 %tmp174, %tmp177
  %cmp.ext = zext i1 %cmp178 to i32
  %cmp179 = icmp ne i32 %cmp.ext, 0
  %cmp.ext180 = zext i1 %cmp179 to i32
  store i32 %cmp.ext180, i32* %update_share, align 4
  store i32 0, i32* %i181, align 4
  br label %for.cond182

for.inc173:                                       ; preds = %for.exit183
  %tmp246 = load i32* %a, align 4
  %tmp247 = add i32 %tmp246, 4
  store i32 %tmp247, i32* %a, align 4
  br label %for.cond168

for.cond182:                                      ; preds = %for.inc188, %for.body172
  %tmp184 = load i32* %i181, align 4
  %cmp185 = icmp ne i32 %tmp184, 4
  br i1 %cmp185, label %for.body187, label %for.exit183

for.exit183:                                      ; preds = %for.cond182
  br label %for.inc173

for.body187:                                      ; preds = %for.cond182
  %tmp189 = load i32* %update_share, align 4
  %tobool190 = icmp ne i32 %tmp189, 0
  br i1 %tobool190, label %if.then192, label %if.end191

for.inc188:                                       ; preds = %if.end191
  %tmp244 = load i32* %i181, align 4
  %tmp245 = add i32 %tmp244, 1
  store i32 %tmp245, i32* %i181, align 4
  br label %for.cond182

if.end191:                                        ; preds = %if.end220, %for.body187
  call void @barrier(i32 1) #0
  %tmp231 = load <4 x i32>* %mix, align 16
  %tmp232 = load %3 addrspace(1)** %g_dag.addr, align 8
  %tmp233 = load i32 addrspace(3)** %share0, align 8
  %tmp234 = load i32 addrspace(3)* %tmp233, align 4
  %conv235 = zext i32 %tmp234 to i64
  %arrayidx236 = getelementptr %3 addrspace(1)* %tmp232, i64 %conv235
  %structele237 = getelementptr inbounds %3 addrspace(1)* %arrayidx236, i32 0, i32 0
  %ptr238 = getelementptr inbounds [8 x <4 x i32>] addrspace(1)* %structele237, i32 0, i32 0
  %tmp239 = load i32* %thread_id, align 4
  %conv240 = zext i32 %tmp239 to i64
  %arrayidx241 = getelementptr <4 x i32> addrspace(1)* %ptr238, i64 %conv240
  %tmp242 = load <4 x i32> addrspace(1)* %arrayidx241, align 16
  %call243 = call <4 x i32> @fnv4(<4 x i32> %tmp231, <4 x i32> %tmp242) #0
  store <4 x i32> %call243, <4 x i32>* %mix, align 16
  br label %for.inc188

if.then192:                                       ; preds = %for.body187
  %tmp193 = load i32* %init0, align 4
  %tmp194 = load i32* %a, align 4
  %tmp195 = load i32* %i181, align 4
  %tmp196 = add i32 %tmp194, %tmp195
  %tmp197 = xor i32 %tmp193, %tmp196
  %conv198 = bitcast <4 x i32>* %mix to i32*
  %tmp199 = load i32* %i181, align 4
  %conv200 = zext i32 %tmp199 to i64
  %arrayidx201 = getelementptr i32* %conv198, i64 %conv200
  %tmp202 = load i32* %arrayidx201, align 4
  %call203 = call i32 @fnv(i32 %tmp197, i32 %tmp202) #0
  store i32 %call203, i32* %v, align 4
  %tmp204 = load i32* %v, align 4
  %tmp205 = load i32* %dv.addr, align 4
  %call206 = call i32 @__mul_hi_u32(i32 %tmp204, i32 %tmp205) #0
  store i32 %call206, i32* %ml, align 4
  %tmp207 = load i32* %ml, align 4
  %tmp208 = load i32* %v, align 4
  %tmp209 = load i32* %dmask.addr, align 4
  %tmp210 = and i32 %tmp208, %tmp209
  %tmp211 = add i32 %tmp207, %tmp210
  store i32 %tmp211, i32* %aml, align 4
  %tmp212 = load i32* %aml, align 4
  %tmp213 = load i32* %dm.addr, align 4
  %tmp214 = and i32 %tmp213, 31
  %tmp215 = lshr i32 %tmp212, %tmp214
  store i32 %tmp215, i32* %res, align 4
  %tmp216 = load i32* %aml, align 4
  %tmp217 = load i32* %ml, align 4
  %cmp218 = icmp ult i32 %tmp216, %tmp217
  br i1 %cmp218, label %if.then221, label %if.end220

if.end220:                                        ; preds = %if.then221, %if.then192
  %tmp225 = load i32 addrspace(3)** %share0, align 8
  %tmp226 = load i32* %v, align 4
  %tmp227 = load i32* %dagsize.addr, align 4
  %tmp228 = load i32* %res, align 4
  %tmp229 = mul i32 %tmp227, %tmp228
  %tmp230 = sub i32 %tmp226, %tmp229
  store i32 %tmp230, i32 addrspace(3)* %tmp225, align 4
  br label %if.end191

if.then221:                                       ; preds = %if.then192
  %tmp222 = load i32* %res, align 4
  %tmp223 = load i32* %morebit, align 4
  %tmp224 = or i32 %tmp222, %tmp223
  store i32 %tmp224, i32* %res, align 4
  br label %if.end220

if.end264:                                        ; preds = %for.exit268, %for.exit169
  call void @barrier(i32 1) #0
  br label %for.inc111

if.then265:                                       ; preds = %for.exit169
  store i32 0, i32* %i266, align 4
  br label %for.cond267

for.cond267:                                      ; preds = %for.inc273, %if.then265
  %tmp269 = load i32* %i266, align 4
  %cmp270 = icmp ne i32 %tmp269, 4
  br i1 %cmp270, label %for.body272, label %for.exit268

for.exit268:                                      ; preds = %for.cond267
  br label %if.end264

for.body272:                                      ; preds = %for.cond267
  %ptr274 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %tmp275 = getelementptr i64* %ptr274, i64 8
  %tmp276 = load i32* %i266, align 4
  %conv277 = zext i32 %tmp276 to i64
  %arrayidx278 = getelementptr i64* %tmp275, i64 %conv277
  %tmp279 = load %0 addrspace(3)** %share, align 8
  %tmp280 = load i32* %hash_id, align 4
  %conv281 = zext i32 %tmp280 to i64
  %arrayidx282 = getelementptr %0 addrspace(3)* %tmp279, i64 %conv281
  %structele283 = getelementptr inbounds %0 addrspace(3)* %arrayidx282, i32 0, i32 0
  %tmp284 = bitcast [4 x <4 x i32>] addrspace(3)* %structele283 to [8 x i64] addrspace(3)*
  %ptr285 = getelementptr inbounds [8 x i64] addrspace(3)* %tmp284, i32 0, i32 0
  %tmp286 = load i32* %i266, align 4
  %conv287 = zext i32 %tmp286 to i64
  %arrayidx288 = getelementptr i64 addrspace(3)* %ptr285, i64 %conv287
  %tmp289 = load i64 addrspace(3)* %arrayidx288, align 8
  store i64 %tmp289, i64* %arrayidx278, align 8
  br label %for.inc273

for.inc273:                                       ; preds = %for.body272
  %tmp290 = load i32* %i266, align 4
  %tmp291 = add i32 %tmp290, 1
  store i32 %tmp291, i32* %i266, align 4
  br label %for.cond267

for.cond295:                                      ; preds = %for.inc301, %for.exit107
  %tmp297 = load i32* %i294, align 4
  %cmp298 = icmp ne i32 %tmp297, 25
  br i1 %cmp298, label %for.body300, label %for.exit296

for.exit296:                                      ; preds = %for.cond295
  %ptr308 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %arrayidx309 = getelementptr i64* %ptr308, i64 12
  store i64 1, i64* %arrayidx309, align 8
  %ptr310 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %arrayidx311 = getelementptr i64* %ptr310, i64 16
  store i64 -9223372036854775808, i64* %arrayidx311, align 8
  %ptr312 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %conv313 = bitcast i64* %ptr312 to <2 x i32>*
  %tmp314 = load i32* %isolate.addr, align 4
  call void @keccak_f1600_no_absorb(<2 x i32>* %conv313, i32 1, i32 %tmp314) #0
  %ptr315 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %arrayidx316 = getelementptr i64* %ptr315, i64 0
  %tmp317 = load i64* %arrayidx316, align 8
  %conv318 = bitcast i64 %tmp317 to <8 x i8>
  store <8 x i8> %conv318, <8 x i8>* %tmp319, align 8
  %tmp320 = load <8 x i8>* %tmp319, align 8
  %tmp321 = extractelement <8 x i8> %tmp320, i32 7
  %tmp322 = insertelement <8 x i8> undef, i8 %tmp321, i32 0
  %tmp323 = extractelement <8 x i8> %tmp320, i32 6
  %tmp324 = insertelement <8 x i8> %tmp322, i8 %tmp323, i32 1
  %tmp325 = extractelement <8 x i8> %tmp320, i32 5
  %tmp326 = insertelement <8 x i8> %tmp324, i8 %tmp325, i32 2
  %tmp327 = extractelement <8 x i8> %tmp320, i32 4
  %tmp328 = insertelement <8 x i8> %tmp326, i8 %tmp327, i32 3
  %tmp329 = extractelement <8 x i8> %tmp320, i32 3
  %tmp330 = insertelement <8 x i8> %tmp328, i8 %tmp329, i32 4
  %tmp331 = extractelement <8 x i8> %tmp320, i32 2
  %tmp332 = insertelement <8 x i8> %tmp330, i8 %tmp331, i32 5
  %tmp333 = extractelement <8 x i8> %tmp320, i32 1
  %tmp334 = insertelement <8 x i8> %tmp332, i8 %tmp333, i32 6
  %tmp335 = extractelement <8 x i8> %tmp320, i32 0
  %tmp336 = insertelement <8 x i8> %tmp334, i8 %tmp335, i32 7
  %conv337 = bitcast <8 x i8> %tmp336 to i64
  %tmp338 = load i64* %target.addr, align 8
  %cmp339 = icmp ult i64 %conv337, %tmp338
  br i1 %cmp339, label %if.then342, label %if.end341

for.body300:                                      ; preds = %for.cond295
  %ptr302 = getelementptr inbounds [25 x i64]* %state, i32 0, i32 0
  %tmp303 = load i32* %i294, align 4
  %conv304 = zext i32 %tmp303 to i64
  %arrayidx305 = getelementptr i64* %ptr302, i64 %conv304
  store i64 0, i64* %arrayidx305, align 8
  br label %for.inc301

for.inc301:                                       ; preds = %for.body300
  %tmp306 = load i32* %i294, align 4
  %tmp307 = add i32 %tmp306, 1
  store i32 %tmp307, i32* %i294, align 4
  br label %for.cond295

if.end341:                                        ; preds = %if.then342, %for.exit296
  br label %return

if.then342:                                       ; preds = %for.exit296
  %tmp343 = load i32 addrspace(1)** %g_output.addr, align 8
  %tmp344 = getelementptr i32 addrspace(1)* %tmp343, i64 0
  %call345 = call i32 @__atomic_inc_gu32(i32 addrspace(1)* %tmp344) #0
  %call346 = call i32 @__min_u32(i32 3, i32 %call345) #0
  store i32 %call346, i32* %slot, align 4
  %tmp347 = load i32 addrspace(1)** %g_output.addr, align 8
  %tmp348 = load i32* %slot, align 4
  %tmp349 = mul i32 9, %tmp348
  %tmp350 = add i32 1, %tmp349
  %conv351 = zext i32 %tmp350 to i64
  %arrayidx352 = getelementptr i32 addrspace(1)* %tmp347, i64 %conv351
  %tmp353 = load i32* %gid, align 4
  store volatile i32 %tmp353, i32 addrspace(1)* %arrayidx352, align 4
  br label %if.end341
}