Lines Matching refs:O3
4 …%s -mcpu=corei7 -passes='default<O3>' -S -unroll-threshold=150 -unroll-allow-partial=0 | FileCheck…
5 ; RUN: opt < %s -mcpu=corei7 -passes='default<O3>' -S -unroll-allow-partial=0 | FileCheck %s --chec…
10 ; RUN: opt < %s -mcpu=corei7 -passes='default<O3>' -unroll-threshold=150 -vectorize-loops=false -S …
254 ; O3-LABEL: @enabled(
255 ; O3-NEXT: entry:
256 ; O3-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[N:%.*]], i64 0
257 ; O3-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32…
258 ; O3-NEXT: [[TMP0:%.*]] = bitcast i32* [[B:%.*]] to <4 x i32>*
259 ; O3-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, <4 x i32>* [[TMP0]], align 4
260 ; O3-NEXT: [[TMP1:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
261 ; O3-NEXT: [[TMP2:%.*]] = bitcast i32* [[A:%.*]] to <4 x i32>*
262 ; O3-NEXT: store <4 x i32> [[TMP1]], <4 x i32>* [[TMP2]], align 4
263 ; O3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 4
264 ; O3-NEXT: [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>*
265 ; O3-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, <4 x i32>* [[TMP4]], align 4
266 ; O3-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
267 ; O3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 4
268 ; O3-NEXT: [[TMP7:%.*]] = bitcast i32* [[TMP6]] to <4 x i32>*
269 ; O3-NEXT: store <4 x i32> [[TMP5]], <4 x i32>* [[TMP7]], align 4
270 ; O3-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 8
271 ; O3-NEXT: [[TMP9:%.*]] = bitcast i32* [[TMP8]] to <4 x i32>*
272 ; O3-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, <4 x i32>* [[TMP9]], align 4
273 ; O3-NEXT: [[TMP10:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
274 ; O3-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 8
275 ; O3-NEXT: [[TMP12:%.*]] = bitcast i32* [[TMP11]] to <4 x i32>*
276 ; O3-NEXT: store <4 x i32> [[TMP10]], <4 x i32>* [[TMP12]], align 4
277 ; O3-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 12
278 ; O3-NEXT: [[TMP14:%.*]] = bitcast i32* [[TMP13]] to <4 x i32>*
279 ; O3-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, <4 x i32>* [[TMP14]], align 4
280 ; O3-NEXT: [[TMP15:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
281 ; O3-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 12
282 ; O3-NEXT: [[TMP17:%.*]] = bitcast i32* [[TMP16]] to <4 x i32>*
283 ; O3-NEXT: store <4 x i32> [[TMP15]], <4 x i32>* [[TMP17]], align 4
284 ; O3-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 16
285 ; O3-NEXT: [[TMP19:%.*]] = bitcast i32* [[TMP18]] to <4 x i32>*
286 ; O3-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, <4 x i32>* [[TMP19]], align 4
287 ; O3-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
288 ; O3-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 16
289 ; O3-NEXT: [[TMP22:%.*]] = bitcast i32* [[TMP21]] to <4 x i32>*
290 ; O3-NEXT: store <4 x i32> [[TMP20]], <4 x i32>* [[TMP22]], align 4
291 ; O3-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 20
292 ; O3-NEXT: [[TMP24:%.*]] = bitcast i32* [[TMP23]] to <4 x i32>*
293 ; O3-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, <4 x i32>* [[TMP24]], align 4
294 ; O3-NEXT: [[TMP25:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
295 ; O3-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 20
296 ; O3-NEXT: [[TMP27:%.*]] = bitcast i32* [[TMP26]] to <4 x i32>*
297 ; O3-NEXT: store <4 x i32> [[TMP25]], <4 x i32>* [[TMP27]], align 4
298 ; O3-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 24
299 ; O3-NEXT: [[TMP29:%.*]] = bitcast i32* [[TMP28]] to <4 x i32>*
300 ; O3-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, <4 x i32>* [[TMP29]], align 4
301 ; O3-NEXT: [[TMP30:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
302 ; O3-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 24
303 ; O3-NEXT: [[TMP32:%.*]] = bitcast i32* [[TMP31]] to <4 x i32>*
304 ; O3-NEXT: store <4 x i32> [[TMP30]], <4 x i32>* [[TMP32]], align 4
305 ; O3-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 28
306 ; O3-NEXT: [[TMP34:%.*]] = bitcast i32* [[TMP33]] to <4 x i32>*
307 ; O3-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, <4 x i32>* [[TMP34]], align 4
308 ; O3-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
309 ; O3-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 28
310 ; O3-NEXT: [[TMP37:%.*]] = bitcast i32* [[TMP36]] to <4 x i32>*
311 ; O3-NEXT: store <4 x i32> [[TMP35]], <4 x i32>* [[TMP37]], align 4
312 ; O3-NEXT: [[TMP38:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 32
313 ; O3-NEXT: [[TMP39:%.*]] = bitcast i32* [[TMP38]] to <4 x i32>*
314 ; O3-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, <4 x i32>* [[TMP39]], align 4
315 ; O3-NEXT: [[TMP40:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
316 ; O3-NEXT: [[TMP41:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 32
317 ; O3-NEXT: [[TMP42:%.*]] = bitcast i32* [[TMP41]] to <4 x i32>*
318 ; O3-NEXT: store <4 x i32> [[TMP40]], <4 x i32>* [[TMP42]], align 4
319 ; O3-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 36
320 ; O3-NEXT: [[TMP44:%.*]] = bitcast i32* [[TMP43]] to <4 x i32>*
321 ; O3-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, <4 x i32>* [[TMP44]], align 4
322 ; O3-NEXT: [[TMP45:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
323 ; O3-NEXT: [[TMP46:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 36
324 ; O3-NEXT: [[TMP47:%.*]] = bitcast i32* [[TMP46]] to <4 x i32>*
325 ; O3-NEXT: store <4 x i32> [[TMP45]], <4 x i32>* [[TMP47]], align 4
326 ; O3-NEXT: [[TMP48:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 40
327 ; O3-NEXT: [[TMP49:%.*]] = bitcast i32* [[TMP48]] to <4 x i32>*
328 ; O3-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, <4 x i32>* [[TMP49]], align 4
329 ; O3-NEXT: [[TMP50:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
330 ; O3-NEXT: [[TMP51:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 40
331 ; O3-NEXT: [[TMP52:%.*]] = bitcast i32* [[TMP51]] to <4 x i32>*
332 ; O3-NEXT: store <4 x i32> [[TMP50]], <4 x i32>* [[TMP52]], align 4
333 ; O3-NEXT: [[TMP53:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 44
334 ; O3-NEXT: [[TMP54:%.*]] = bitcast i32* [[TMP53]] to <4 x i32>*
335 ; O3-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, <4 x i32>* [[TMP54]], align 4
336 ; O3-NEXT: [[TMP55:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
337 ; O3-NEXT: [[TMP56:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 44
338 ; O3-NEXT: [[TMP57:%.*]] = bitcast i32* [[TMP56]] to <4 x i32>*
339 ; O3-NEXT: store <4 x i32> [[TMP55]], <4 x i32>* [[TMP57]], align 4
340 ; O3-NEXT: [[TMP58:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 48
341 ; O3-NEXT: [[TMP59:%.*]] = bitcast i32* [[TMP58]] to <4 x i32>*
342 ; O3-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, <4 x i32>* [[TMP59]], align 4
343 ; O3-NEXT: [[TMP60:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
344 ; O3-NEXT: [[TMP61:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 48
345 ; O3-NEXT: [[TMP62:%.*]] = bitcast i32* [[TMP61]] to <4 x i32>*
346 ; O3-NEXT: store <4 x i32> [[TMP60]], <4 x i32>* [[TMP62]], align 4
347 ; O3-NEXT: [[TMP63:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 52
348 ; O3-NEXT: [[TMP64:%.*]] = bitcast i32* [[TMP63]] to <4 x i32>*
349 ; O3-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, <4 x i32>* [[TMP64]], align 4
350 ; O3-NEXT: [[TMP65:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
351 ; O3-NEXT: [[TMP66:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 52
352 ; O3-NEXT: [[TMP67:%.*]] = bitcast i32* [[TMP66]] to <4 x i32>*
353 ; O3-NEXT: store <4 x i32> [[TMP65]], <4 x i32>* [[TMP67]], align 4
354 ; O3-NEXT: [[TMP68:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 56
355 ; O3-NEXT: [[TMP69:%.*]] = bitcast i32* [[TMP68]] to <4 x i32>*
356 ; O3-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, <4 x i32>* [[TMP69]], align 4
357 ; O3-NEXT: [[TMP70:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
358 ; O3-NEXT: [[TMP71:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 56
359 ; O3-NEXT: [[TMP72:%.*]] = bitcast i32* [[TMP71]] to <4 x i32>*
360 ; O3-NEXT: store <4 x i32> [[TMP70]], <4 x i32>* [[TMP72]], align 4
361 ; O3-NEXT: [[TMP73:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 60
362 ; O3-NEXT: [[TMP74:%.*]] = bitcast i32* [[TMP73]] to <4 x i32>*
363 ; O3-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, <4 x i32>* [[TMP74]], align 4
364 ; O3-NEXT: [[TMP75:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
365 ; O3-NEXT: [[TMP76:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 60
366 ; O3-NEXT: [[TMP77:%.*]] = bitcast i32* [[TMP76]] to <4 x i32>*
367 ; O3-NEXT: store <4 x i32> [[TMP75]], <4 x i32>* [[TMP77]], align 4
368 ; O3-NEXT: [[TMP78:%.*]] = load i32, i32* [[A]], align 4
369 ; O3-NEXT: ret i32 [[TMP78]]
1227 ; O3-LABEL: @nopragma(
1228 ; O3-NEXT: entry:
1229 ; O3-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[N:%.*]], i64 0
1230 ; O3-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32…
1231 ; O3-NEXT: [[TMP0:%.*]] = bitcast i32* [[B:%.*]] to <4 x i32>*
1232 ; O3-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, <4 x i32>* [[TMP0]], align 4
1233 ; O3-NEXT: [[TMP1:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
1234 ; O3-NEXT: [[TMP2:%.*]] = bitcast i32* [[A:%.*]] to <4 x i32>*
1235 ; O3-NEXT: store <4 x i32> [[TMP1]], <4 x i32>* [[TMP2]], align 4
1236 ; O3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 4
1237 ; O3-NEXT: [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>*
1238 ; O3-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, <4 x i32>* [[TMP4]], align 4
1239 ; O3-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
1240 ; O3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 4
1241 ; O3-NEXT: [[TMP7:%.*]] = bitcast i32* [[TMP6]] to <4 x i32>*
1242 ; O3-NEXT: store <4 x i32> [[TMP5]], <4 x i32>* [[TMP7]], align 4
1243 ; O3-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 8
1244 ; O3-NEXT: [[TMP9:%.*]] = bitcast i32* [[TMP8]] to <4 x i32>*
1245 ; O3-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, <4 x i32>* [[TMP9]], align 4
1246 ; O3-NEXT: [[TMP10:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
1247 ; O3-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 8
1248 ; O3-NEXT: [[TMP12:%.*]] = bitcast i32* [[TMP11]] to <4 x i32>*
1249 ; O3-NEXT: store <4 x i32> [[TMP10]], <4 x i32>* [[TMP12]], align 4
1250 ; O3-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 12
1251 ; O3-NEXT: [[TMP14:%.*]] = bitcast i32* [[TMP13]] to <4 x i32>*
1252 ; O3-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, <4 x i32>* [[TMP14]], align 4
1253 ; O3-NEXT: [[TMP15:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
1254 ; O3-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 12
1255 ; O3-NEXT: [[TMP17:%.*]] = bitcast i32* [[TMP16]] to <4 x i32>*
1256 ; O3-NEXT: store <4 x i32> [[TMP15]], <4 x i32>* [[TMP17]], align 4
1257 ; O3-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 16
1258 ; O3-NEXT: [[TMP19:%.*]] = bitcast i32* [[TMP18]] to <4 x i32>*
1259 ; O3-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, <4 x i32>* [[TMP19]], align 4
1260 ; O3-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
1261 ; O3-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 16
1262 ; O3-NEXT: [[TMP22:%.*]] = bitcast i32* [[TMP21]] to <4 x i32>*
1263 ; O3-NEXT: store <4 x i32> [[TMP20]], <4 x i32>* [[TMP22]], align 4
1264 ; O3-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 20
1265 ; O3-NEXT: [[TMP24:%.*]] = bitcast i32* [[TMP23]] to <4 x i32>*
1266 ; O3-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, <4 x i32>* [[TMP24]], align 4
1267 ; O3-NEXT: [[TMP25:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
1268 ; O3-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 20
1269 ; O3-NEXT: [[TMP27:%.*]] = bitcast i32* [[TMP26]] to <4 x i32>*
1270 ; O3-NEXT: store <4 x i32> [[TMP25]], <4 x i32>* [[TMP27]], align 4
1271 ; O3-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 24
1272 ; O3-NEXT: [[TMP29:%.*]] = bitcast i32* [[TMP28]] to <4 x i32>*
1273 ; O3-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, <4 x i32>* [[TMP29]], align 4
1274 ; O3-NEXT: [[TMP30:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
1275 ; O3-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 24
1276 ; O3-NEXT: [[TMP32:%.*]] = bitcast i32* [[TMP31]] to <4 x i32>*
1277 ; O3-NEXT: store <4 x i32> [[TMP30]], <4 x i32>* [[TMP32]], align 4
1278 ; O3-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 28
1279 ; O3-NEXT: [[TMP34:%.*]] = bitcast i32* [[TMP33]] to <4 x i32>*
1280 ; O3-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, <4 x i32>* [[TMP34]], align 4
1281 ; O3-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
1282 ; O3-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 28
1283 ; O3-NEXT: [[TMP37:%.*]] = bitcast i32* [[TMP36]] to <4 x i32>*
1284 ; O3-NEXT: store <4 x i32> [[TMP35]], <4 x i32>* [[TMP37]], align 4
1285 ; O3-NEXT: [[TMP38:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 32
1286 ; O3-NEXT: [[TMP39:%.*]] = bitcast i32* [[TMP38]] to <4 x i32>*
1287 ; O3-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, <4 x i32>* [[TMP39]], align 4
1288 ; O3-NEXT: [[TMP40:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
1289 ; O3-NEXT: [[TMP41:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 32
1290 ; O3-NEXT: [[TMP42:%.*]] = bitcast i32* [[TMP41]] to <4 x i32>*
1291 ; O3-NEXT: store <4 x i32> [[TMP40]], <4 x i32>* [[TMP42]], align 4
1292 ; O3-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 36
1293 ; O3-NEXT: [[TMP44:%.*]] = bitcast i32* [[TMP43]] to <4 x i32>*
1294 ; O3-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, <4 x i32>* [[TMP44]], align 4
1295 ; O3-NEXT: [[TMP45:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
1296 ; O3-NEXT: [[TMP46:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 36
1297 ; O3-NEXT: [[TMP47:%.*]] = bitcast i32* [[TMP46]] to <4 x i32>*
1298 ; O3-NEXT: store <4 x i32> [[TMP45]], <4 x i32>* [[TMP47]], align 4
1299 ; O3-NEXT: [[TMP48:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 40
1300 ; O3-NEXT: [[TMP49:%.*]] = bitcast i32* [[TMP48]] to <4 x i32>*
1301 ; O3-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, <4 x i32>* [[TMP49]], align 4
1302 ; O3-NEXT: [[TMP50:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
1303 ; O3-NEXT: [[TMP51:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 40
1304 ; O3-NEXT: [[TMP52:%.*]] = bitcast i32* [[TMP51]] to <4 x i32>*
1305 ; O3-NEXT: store <4 x i32> [[TMP50]], <4 x i32>* [[TMP52]], align 4
1306 ; O3-NEXT: [[TMP53:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 44
1307 ; O3-NEXT: [[TMP54:%.*]] = bitcast i32* [[TMP53]] to <4 x i32>*
1308 ; O3-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, <4 x i32>* [[TMP54]], align 4
1309 ; O3-NEXT: [[TMP55:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
1310 ; O3-NEXT: [[TMP56:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 44
1311 ; O3-NEXT: [[TMP57:%.*]] = bitcast i32* [[TMP56]] to <4 x i32>*
1312 ; O3-NEXT: store <4 x i32> [[TMP55]], <4 x i32>* [[TMP57]], align 4
1313 ; O3-NEXT: [[TMP58:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 48
1314 ; O3-NEXT: [[TMP59:%.*]] = bitcast i32* [[TMP58]] to <4 x i32>*
1315 ; O3-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, <4 x i32>* [[TMP59]], align 4
1316 ; O3-NEXT: [[TMP60:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
1317 ; O3-NEXT: [[TMP61:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 48
1318 ; O3-NEXT: [[TMP62:%.*]] = bitcast i32* [[TMP61]] to <4 x i32>*
1319 ; O3-NEXT: store <4 x i32> [[TMP60]], <4 x i32>* [[TMP62]], align 4
1320 ; O3-NEXT: [[TMP63:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 52
1321 ; O3-NEXT: [[TMP64:%.*]] = bitcast i32* [[TMP63]] to <4 x i32>*
1322 ; O3-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, <4 x i32>* [[TMP64]], align 4
1323 ; O3-NEXT: [[TMP65:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
1324 ; O3-NEXT: [[TMP66:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 52
1325 ; O3-NEXT: [[TMP67:%.*]] = bitcast i32* [[TMP66]] to <4 x i32>*
1326 ; O3-NEXT: store <4 x i32> [[TMP65]], <4 x i32>* [[TMP67]], align 4
1327 ; O3-NEXT: [[TMP68:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 56
1328 ; O3-NEXT: [[TMP69:%.*]] = bitcast i32* [[TMP68]] to <4 x i32>*
1329 ; O3-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, <4 x i32>* [[TMP69]], align 4
1330 ; O3-NEXT: [[TMP70:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
1331 ; O3-NEXT: [[TMP71:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 56
1332 ; O3-NEXT: [[TMP72:%.*]] = bitcast i32* [[TMP71]] to <4 x i32>*
1333 ; O3-NEXT: store <4 x i32> [[TMP70]], <4 x i32>* [[TMP72]], align 4
1334 ; O3-NEXT: [[TMP73:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 60
1335 ; O3-NEXT: [[TMP74:%.*]] = bitcast i32* [[TMP73]] to <4 x i32>*
1336 ; O3-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, <4 x i32>* [[TMP74]], align 4
1337 ; O3-NEXT: [[TMP75:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
1338 ; O3-NEXT: [[TMP76:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 60
1339 ; O3-NEXT: [[TMP77:%.*]] = bitcast i32* [[TMP76]] to <4 x i32>*
1340 ; O3-NEXT: store <4 x i32> [[TMP75]], <4 x i32>* [[TMP77]], align 4
1341 ; O3-NEXT: [[TMP78:%.*]] = load i32, i32* [[A]], align 4
1342 ; O3-NEXT: ret i32 [[TMP78]]
1750 ; O3-LABEL: @disabled(
1751 ; O3-NEXT: entry:
1752 ; O3-NEXT: br label [[FOR_BODY:%.*]]
1753 ; O3: for.body:
1754 ; O3-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_B…
1755 ; O3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDVARS_IV]]
1756 ; O3-NEXT: [[TMP0:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
1757 ; O3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
1758 ; O3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 [[INDVARS_IV]]
1759 ; O3-NEXT: store i32 [[ADD]], i32* [[ARRAYIDX2]], align 4
1760 ; O3-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
1761 ; O3-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
1762 ; O3-NEXT: br i1 [[EXITCOND]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP0:![0…
1763 ; O3: for.end:
1764 ; O3-NEXT: [[TMP1:%.*]] = load i32, i32* [[A]], align 4
1765 ; O3-NEXT: ret i32 [[TMP1]]