diff options
Diffstat (limited to 'llvm/test/Transforms/LoopVectorize/induction.ll')
| -rw-r--r-- | llvm/test/Transforms/LoopVectorize/induction.ll | 114 |
1 files changed, 57 insertions, 57 deletions
diff --git a/llvm/test/Transforms/LoopVectorize/induction.ll b/llvm/test/Transforms/LoopVectorize/induction.ll index 45674acaae53..08d05a1e2db6 100644 --- a/llvm/test/Transforms/LoopVectorize/induction.ll +++ b/llvm/test/Transforms/LoopVectorize/induction.ll @@ -90,7 +90,7 @@ define void @multi_int_induction(ptr %A, i32 %N) { ; IND-NEXT: [[INC]] = add nsw i32 [[COUNT_09]], 1 ; IND-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1 ; IND-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32 -; IND-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]] +; IND-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[N]], [[LFTR_WIDEIV]] ; IND-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]] ; IND: for.end: ; IND-NEXT: ret void @@ -134,7 +134,7 @@ define void @multi_int_induction(ptr %A, i32 %N) { ; UNROLL-NEXT: [[INC]] = add nsw i32 [[COUNT_09]], 1 ; UNROLL-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1 ; UNROLL-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32 -; UNROLL-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]] +; UNROLL-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[N]], [[LFTR_WIDEIV]] ; UNROLL-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]] ; UNROLL: for.end: ; UNROLL-NEXT: ret void @@ -227,7 +227,7 @@ define void @multi_int_induction(ptr %A, i32 %N) { ; INTERLEAVE-NEXT: [[INC]] = add nsw i32 [[COUNT_09]], 1 ; INTERLEAVE-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1 ; INTERLEAVE-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32 -; INTERLEAVE-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]] +; INTERLEAVE-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[N]], [[LFTR_WIDEIV]] ; INTERLEAVE-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]] ; INTERLEAVE: for.end: ; INTERLEAVE-NEXT: ret void @@ -361,7 +361,7 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) { ; IND-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; IND-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]] ; IND: middle.block: -; IND-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]] +; IND-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]] ; IND-NEXT: br i1 [[CMP_N]], label [[LOOPEXIT:%.*]], label [[SCALAR_PH]] ; IND: scalar.ph: ; IND-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ] @@ -374,7 +374,7 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) { ; IND-NEXT: [[TMP13:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]] ; IND-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP13]], i64 [[OFFSET2]] ; IND-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4 -; IND-NEXT: [[M:%.*]] = fmul fast float [[L2]], [[B]] +; IND-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]] ; IND-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]] ; IND-NEXT: store float [[AD]], ptr [[ARR_IDX]], align 4 ; IND-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 @@ -428,7 +428,7 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) { ; UNROLL-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]] ; UNROLL: middle.block: -; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]] +; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[CMP_N]], label [[LOOPEXIT:%.*]], label [[SCALAR_PH]] ; UNROLL: scalar.ph: ; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ] @@ -441,7 +441,7 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) { ; UNROLL-NEXT: [[TMP17:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]] ; UNROLL-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP17]], i64 [[OFFSET2]] ; UNROLL-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4 -; UNROLL-NEXT: [[M:%.*]] = fmul fast float [[L2]], [[B]] +; UNROLL-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]] ; UNROLL-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]] ; UNROLL-NEXT: store float [[AD]], ptr [[ARR_IDX]], align 4 ; UNROLL-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 @@ -571,7 +571,7 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) { ; INTERLEAVE-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]] ; INTERLEAVE: middle.block: -; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]] +; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[CMP_N]], label [[LOOPEXIT:%.*]], label [[SCALAR_PH]] ; INTERLEAVE: scalar.ph: ; INTERLEAVE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ] @@ -584,7 +584,7 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) { ; INTERLEAVE-NEXT: [[TMP17:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]] ; INTERLEAVE-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP17]], i64 [[OFFSET2]] ; INTERLEAVE-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4 -; INTERLEAVE-NEXT: [[M:%.*]] = fmul fast float [[L2]], [[B]] +; INTERLEAVE-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]] ; INTERLEAVE-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]] ; INTERLEAVE-NEXT: store float [[AD]], ptr [[ARR_IDX]], align 4 ; INTERLEAVE-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 @@ -1636,7 +1636,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) { ; IND-NEXT: [[TMP8:%.*]] = or disjoint i64 [[TMP7]], 4 ; IND-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP8]] ; IND-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[SCEVGEP]], [[SCEVGEP2]] -; IND-NEXT: [[BOUND1:%.*]] = icmp ugt ptr [[SCEVGEP1]], [[A]] +; IND-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[A]], [[SCEVGEP1]] ; IND-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] ; IND-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]] ; IND: vector.ph: @@ -1676,7 +1676,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) { ; IND-NEXT: store i32 [[TMP21]], ptr [[TMP22]], align 1 ; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1 ; IND-NEXT: [[TMP23:%.*]] = trunc i64 [[I_NEXT]] to i32 -; IND-NEXT: [[COND:%.*]] = icmp eq i32 [[TMP23]], [[N]] +; IND-NEXT: [[COND:%.*]] = icmp eq i32 [[N]], [[TMP23]] ; IND-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]] ; IND: for.end: ; IND-NEXT: ret void @@ -1699,7 +1699,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) { ; UNROLL-NEXT: [[TMP8:%.*]] = or disjoint i64 [[TMP7]], 4 ; UNROLL-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP8]] ; UNROLL-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[SCEVGEP]], [[SCEVGEP2]] -; UNROLL-NEXT: [[BOUND1:%.*]] = icmp ugt ptr [[SCEVGEP1]], [[A]] +; UNROLL-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[A]], [[SCEVGEP1]] ; UNROLL-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] ; UNROLL-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]] ; UNROLL: vector.ph: @@ -1753,7 +1753,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) { ; UNROLL-NEXT: store i32 [[TMP32]], ptr [[TMP33]], align 1 ; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1 ; UNROLL-NEXT: [[TMP34:%.*]] = trunc i64 [[I_NEXT]] to i32 -; UNROLL-NEXT: [[COND:%.*]] = icmp eq i32 [[TMP34]], [[N]] +; UNROLL-NEXT: [[COND:%.*]] = icmp eq i32 [[N]], [[TMP34]] ; UNROLL-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]] ; UNROLL: for.end: ; UNROLL-NEXT: ret void @@ -1855,7 +1855,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) { ; INTERLEAVE-NEXT: [[TMP8:%.*]] = or disjoint i64 [[TMP7]], 4 ; INTERLEAVE-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP8]] ; INTERLEAVE-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[SCEVGEP]], [[SCEVGEP2]] -; INTERLEAVE-NEXT: [[BOUND1:%.*]] = icmp ugt ptr [[SCEVGEP1]], [[A]] +; INTERLEAVE-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[A]], [[SCEVGEP1]] ; INTERLEAVE-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]] ; INTERLEAVE-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]] ; INTERLEAVE: vector.ph: @@ -1920,7 +1920,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) { ; INTERLEAVE-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 1 ; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1 ; INTERLEAVE-NEXT: [[TMP40:%.*]] = trunc i64 [[I_NEXT]] to i32 -; INTERLEAVE-NEXT: [[COND:%.*]] = icmp eq i32 [[TMP40]], [[N]] +; INTERLEAVE-NEXT: [[COND:%.*]] = icmp eq i32 [[N]], [[TMP40]] ; INTERLEAVE-NEXT: br i1 [[COND]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]] ; INTERLEAVE: for.end: ; INTERLEAVE-NEXT: ret void @@ -2535,13 +2535,13 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) { ; IND: for.body: ; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; IND-NEXT: [[TMP11:%.*]] = trunc i64 [[I]] to i32 -; IND-NEXT: [[TMP12:%.*]] = add i32 [[TMP11]], [[A]] +; IND-NEXT: [[TMP12:%.*]] = add i32 [[A]], [[TMP11]] ; IND-NEXT: [[TMP13:%.*]] = trunc i32 [[TMP12]] to i16 ; IND-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[PAIR_I16]], ptr [[P]], i64 [[I]], i32 1 ; IND-NEXT: store i16 [[TMP13]], ptr [[TMP14]], align 2 ; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1 ; IND-NEXT: [[TMP15:%.*]] = trunc i64 [[I_NEXT]] to i32 -; IND-NEXT: [[COND:%.*]] = icmp eq i32 [[TMP15]], [[N]] +; IND-NEXT: [[COND:%.*]] = icmp eq i32 [[N]], [[TMP15]] ; IND-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP27:![0-9]+]] ; IND: for.end: ; IND-NEXT: ret void @@ -2594,13 +2594,13 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) { ; UNROLL: for.body: ; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; UNROLL-NEXT: [[TMP19:%.*]] = trunc i64 [[I]] to i32 -; UNROLL-NEXT: [[TMP20:%.*]] = add i32 [[TMP19]], [[A]] +; UNROLL-NEXT: [[TMP20:%.*]] = add i32 [[A]], [[TMP19]] ; UNROLL-NEXT: [[TMP21:%.*]] = trunc i32 [[TMP20]] to i16 ; UNROLL-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[PAIR_I16]], ptr [[P]], i64 [[I]], i32 1 ; UNROLL-NEXT: store i16 [[TMP21]], ptr [[TMP22]], align 2 ; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1 ; UNROLL-NEXT: [[TMP23:%.*]] = trunc i64 [[I_NEXT]] to i32 -; UNROLL-NEXT: [[COND:%.*]] = icmp eq i32 [[TMP23]], [[N]] +; UNROLL-NEXT: [[COND:%.*]] = icmp eq i32 [[N]], [[TMP23]] ; UNROLL-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP27:![0-9]+]] ; UNROLL: for.end: ; UNROLL-NEXT: ret void @@ -2730,13 +2730,13 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) { ; INTERLEAVE: for.body: ; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; INTERLEAVE-NEXT: [[TMP31:%.*]] = trunc i64 [[I]] to i32 -; INTERLEAVE-NEXT: [[TMP32:%.*]] = add i32 [[TMP31]], [[A]] +; INTERLEAVE-NEXT: [[TMP32:%.*]] = add i32 [[A]], [[TMP31]] ; INTERLEAVE-NEXT: [[TMP33:%.*]] = trunc i32 [[TMP32]] to i16 ; INTERLEAVE-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[PAIR_I16]], ptr [[P]], i64 [[I]], i32 1 ; INTERLEAVE-NEXT: store i16 [[TMP33]], ptr [[TMP34]], align 2 ; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1 ; INTERLEAVE-NEXT: [[TMP35:%.*]] = trunc i64 [[I_NEXT]] to i32 -; INTERLEAVE-NEXT: [[COND:%.*]] = icmp eq i32 [[TMP35]], [[N]] +; INTERLEAVE-NEXT: [[COND:%.*]] = icmp eq i32 [[N]], [[TMP35]] ; INTERLEAVE-NEXT: br i1 [[COND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP27:![0-9]+]] ; INTERLEAVE: for.end: ; INTERLEAVE-NEXT: ret void @@ -3516,7 +3516,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; IND-NEXT: [[TMP2:%.*]] = xor i8 [[T]], -1 ; IND-NEXT: [[TMP3:%.*]] = icmp ult i8 [[TMP2]], [[TMP1]] ; IND-NEXT: [[TMP4:%.*]] = trunc i32 [[LEN]] to i8 -; IND-NEXT: [[TMP5:%.*]] = add i8 [[TMP4]], [[T]] +; IND-NEXT: [[TMP5:%.*]] = add i8 [[T]], [[TMP4]] ; IND-NEXT: [[TMP6:%.*]] = icmp slt i8 [[TMP5]], [[T]] ; IND-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[LEN]], 255 ; IND-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]] @@ -3525,7 +3525,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; IND: vector.ph: ; IND-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], 510 ; IND-NEXT: [[DOTCAST:%.*]] = trunc i32 [[N_VEC]] to i8 -; IND-NEXT: [[IND_END:%.*]] = add i8 [[DOTCAST]], [[T]] +; IND-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[DOTCAST]] ; IND-NEXT: [[IND_END2:%.*]] = add nuw nsw i32 [[N_VEC]], [[EXT]] ; IND-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT]], i64 0 ; IND-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer @@ -3535,7 +3535,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; IND-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] ; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; IND-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8 -; IND-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST4]], [[T]] +; IND-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]] ; IND-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64 ; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]] ; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP11]], align 4 @@ -3582,7 +3582,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; UNROLL-NEXT: [[TMP2:%.*]] = xor i8 [[T]], -1 ; UNROLL-NEXT: [[TMP3:%.*]] = icmp ult i8 [[TMP2]], [[TMP1]] ; UNROLL-NEXT: [[TMP4:%.*]] = trunc i32 [[LEN]] to i8 -; UNROLL-NEXT: [[TMP5:%.*]] = add i8 [[TMP4]], [[T]] +; UNROLL-NEXT: [[TMP5:%.*]] = add i8 [[T]], [[TMP4]] ; UNROLL-NEXT: [[TMP6:%.*]] = icmp slt i8 [[TMP5]], [[T]] ; UNROLL-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[LEN]], 255 ; UNROLL-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]] @@ -3591,7 +3591,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; UNROLL: vector.ph: ; UNROLL-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], 508 ; UNROLL-NEXT: [[DOTCAST:%.*]] = trunc i32 [[N_VEC]] to i8 -; UNROLL-NEXT: [[IND_END:%.*]] = add i8 [[DOTCAST]], [[T]] +; UNROLL-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[DOTCAST]] ; UNROLL-NEXT: [[IND_END2:%.*]] = add nuw nsw i32 [[N_VEC]], [[EXT]] ; UNROLL-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT]], i64 0 ; UNROLL-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer @@ -3602,7 +3602,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; UNROLL-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2> ; UNROLL-NEXT: [[DOTCAST5:%.*]] = trunc i32 [[INDEX]] to i8 -; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]] +; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST5]] ; UNROLL-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64 ; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]] ; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 8 @@ -3726,7 +3726,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; INTERLEAVE-NEXT: [[TMP2:%.*]] = xor i8 [[T]], -1 ; INTERLEAVE-NEXT: [[TMP3:%.*]] = icmp ult i8 [[TMP2]], [[TMP1]] ; INTERLEAVE-NEXT: [[TMP4:%.*]] = trunc i32 [[LEN]] to i8 -; INTERLEAVE-NEXT: [[TMP5:%.*]] = add i8 [[TMP4]], [[T]] +; INTERLEAVE-NEXT: [[TMP5:%.*]] = add i8 [[T]], [[TMP4]] ; INTERLEAVE-NEXT: [[TMP6:%.*]] = icmp slt i8 [[TMP5]], [[T]] ; INTERLEAVE-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[LEN]], 255 ; INTERLEAVE-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]] @@ -3735,7 +3735,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; INTERLEAVE: vector.ph: ; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], 504 ; INTERLEAVE-NEXT: [[DOTCAST:%.*]] = trunc i32 [[N_VEC]] to i8 -; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i8 [[DOTCAST]], [[T]] +; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[DOTCAST]] ; INTERLEAVE-NEXT: [[IND_END2:%.*]] = add nuw nsw i32 [[N_VEC]], [[EXT]] ; INTERLEAVE-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[EXT]], i64 0 ; INTERLEAVE-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer @@ -3746,7 +3746,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) { ; INTERLEAVE-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4> ; INTERLEAVE-NEXT: [[DOTCAST5:%.*]] = trunc i32 [[INDEX]] to i8 -; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]] +; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST5]] ; INTERLEAVE-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64 ; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]] ; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 16 @@ -3900,7 +3900,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; IND-NEXT: [[TMP2:%.*]] = xor i8 [[T]], -1 ; IND-NEXT: [[TMP3:%.*]] = icmp ult i8 [[TMP2]], [[TMP1]] ; IND-NEXT: [[TMP4:%.*]] = trunc i32 [[LEN]] to i8 -; IND-NEXT: [[TMP5:%.*]] = add i8 [[TMP4]], [[T]] +; IND-NEXT: [[TMP5:%.*]] = add i8 [[T]], [[TMP4]] ; IND-NEXT: [[TMP6:%.*]] = icmp slt i8 [[TMP5]], [[T]] ; IND-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[LEN]], 255 ; IND-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]] @@ -3909,7 +3909,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; IND: vector.ph: ; IND-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], 510 ; IND-NEXT: [[DOTCAST:%.*]] = trunc i32 [[N_VEC]] to i8 -; IND-NEXT: [[IND_END:%.*]] = add i8 [[DOTCAST]], [[T]] +; IND-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[DOTCAST]] ; IND-NEXT: [[EXT_MUL5:%.*]] = add nuw nsw i32 [[N_VEC]], [[EXT]] ; IND-NEXT: [[IND_END1:%.*]] = shl nuw nsw i32 [[EXT_MUL5]], 2 ; IND-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT_MUL]], i64 0 @@ -3920,7 +3920,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; IND-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] ; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; IND-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8 -; IND-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST4]], [[T]] +; IND-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]] ; IND-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64 ; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]] ; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP11]], align 4 @@ -3969,7 +3969,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; UNROLL-NEXT: [[TMP2:%.*]] = xor i8 [[T]], -1 ; UNROLL-NEXT: [[TMP3:%.*]] = icmp ult i8 [[TMP2]], [[TMP1]] ; UNROLL-NEXT: [[TMP4:%.*]] = trunc i32 [[LEN]] to i8 -; UNROLL-NEXT: [[TMP5:%.*]] = add i8 [[TMP4]], [[T]] +; UNROLL-NEXT: [[TMP5:%.*]] = add i8 [[T]], [[TMP4]] ; UNROLL-NEXT: [[TMP6:%.*]] = icmp slt i8 [[TMP5]], [[T]] ; UNROLL-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[LEN]], 255 ; UNROLL-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]] @@ -3978,7 +3978,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; UNROLL: vector.ph: ; UNROLL-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], 508 ; UNROLL-NEXT: [[DOTCAST:%.*]] = trunc i32 [[N_VEC]] to i8 -; UNROLL-NEXT: [[IND_END:%.*]] = add i8 [[DOTCAST]], [[T]] +; UNROLL-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[DOTCAST]] ; UNROLL-NEXT: [[EXT_MUL6:%.*]] = add nuw nsw i32 [[N_VEC]], [[EXT]] ; UNROLL-NEXT: [[IND_END1:%.*]] = shl nuw nsw i32 [[EXT_MUL6]], 2 ; UNROLL-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[EXT_MUL]], i64 0 @@ -3990,7 +3990,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; UNROLL-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], <i32 8, i32 8> ; UNROLL-NEXT: [[DOTCAST5:%.*]] = trunc i32 [[INDEX]] to i8 -; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]] +; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST5]] ; UNROLL-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64 ; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]] ; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 8 @@ -4119,7 +4119,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; INTERLEAVE-NEXT: [[TMP2:%.*]] = xor i8 [[T]], -1 ; INTERLEAVE-NEXT: [[TMP3:%.*]] = icmp ult i8 [[TMP2]], [[TMP1]] ; INTERLEAVE-NEXT: [[TMP4:%.*]] = trunc i32 [[LEN]] to i8 -; INTERLEAVE-NEXT: [[TMP5:%.*]] = add i8 [[TMP4]], [[T]] +; INTERLEAVE-NEXT: [[TMP5:%.*]] = add i8 [[T]], [[TMP4]] ; INTERLEAVE-NEXT: [[TMP6:%.*]] = icmp slt i8 [[TMP5]], [[T]] ; INTERLEAVE-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[LEN]], 255 ; INTERLEAVE-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]] @@ -4128,7 +4128,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; INTERLEAVE: vector.ph: ; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], 504 ; INTERLEAVE-NEXT: [[DOTCAST:%.*]] = trunc i32 [[N_VEC]] to i8 -; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i8 [[DOTCAST]], [[T]] +; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i8 [[T]], [[DOTCAST]] ; INTERLEAVE-NEXT: [[EXT_MUL6:%.*]] = add nuw nsw i32 [[N_VEC]], [[EXT]] ; INTERLEAVE-NEXT: [[IND_END1:%.*]] = shl nuw nsw i32 [[EXT_MUL6]], 2 ; INTERLEAVE-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[EXT_MUL]], i64 0 @@ -4140,7 +4140,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) { ; INTERLEAVE-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 16, i32 16, i32 16, i32 16> ; INTERLEAVE-NEXT: [[DOTCAST5:%.*]] = trunc i32 [[INDEX]] to i8 -; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]] +; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST5]] ; INTERLEAVE-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64 ; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]] ; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 16 @@ -4262,7 +4262,7 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) { ; IND-NEXT: [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] ; IND-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP40:![0-9]+]] ; IND: middle.block: -; IND-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N_VEC]], [[K]] +; IND-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[K]], [[N_VEC]] ; IND-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; IND: scalar.ph: ; IND-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER:%.*]] ] @@ -4299,7 +4299,7 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) { ; UNROLL-NEXT: [[TMP3:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP40:![0-9]+]] ; UNROLL: middle.block: -; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N_VEC]], [[K]] +; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[K]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; UNROLL: scalar.ph: ; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER:%.*]] ] @@ -4376,7 +4376,7 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) { ; INTERLEAVE-NEXT: [[TMP3:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP40:![0-9]+]] ; INTERLEAVE: middle.block: -; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N_VEC]], [[K]] +; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[K]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; INTERLEAVE: scalar.ph: ; INTERLEAVE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER:%.*]] ] @@ -4474,7 +4474,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) { ; IND-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; IND-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP42:![0-9]+]] ; IND: middle.block: -; IND-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[K]] +; IND-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[K]], [[N_VEC]] ; IND-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; IND: scalar.ph: ; IND-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] @@ -4517,7 +4517,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) { ; UNROLL-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP42:![0-9]+]] ; UNROLL: middle.block: -; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[K]] +; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[K]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; UNROLL: scalar.ph: ; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] @@ -4609,7 +4609,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) { ; INTERLEAVE-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP42:![0-9]+]] ; INTERLEAVE: middle.block: -; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[K]] +; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[K]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; INTERLEAVE: scalar.ph: ; INTERLEAVE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] @@ -4694,7 +4694,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) { ; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]] ; IND: vector.ph: ; IND-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], -2 -; IND-NEXT: [[IND_END:%.*]] = add i32 [[N_VEC]], [[I]] +; IND-NEXT: [[IND_END:%.*]] = add i32 [[I]], [[N_VEC]] ; IND-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[I]], i64 0 ; IND-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer ; IND-NEXT: [[INDUCTION:%.*]] = add <2 x i32> [[DOTSPLAT]], <i32 0, i32 1> @@ -4702,7 +4702,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) { ; IND: vector.body: ; IND-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] ; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] -; IND-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[INDEX]], [[I]] +; IND-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[I]], [[INDEX]] ; IND-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64 ; IND-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]] ; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP2]], align 4 @@ -4734,7 +4734,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) { ; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]] ; UNROLL: vector.ph: ; UNROLL-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], -4 -; UNROLL-NEXT: [[IND_END:%.*]] = add i32 [[N_VEC]], [[I]] +; UNROLL-NEXT: [[IND_END:%.*]] = add i32 [[I]], [[N_VEC]] ; UNROLL-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[I]], i64 0 ; UNROLL-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer ; UNROLL-NEXT: [[INDUCTION:%.*]] = add <2 x i32> [[DOTSPLAT]], <i32 0, i32 1> @@ -4743,7 +4743,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) { ; UNROLL-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] ; UNROLL-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2> -; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[INDEX]], [[I]] +; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[I]], [[INDEX]] ; UNROLL-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64 ; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]] ; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8 @@ -4823,7 +4823,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) { ; INTERLEAVE-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]] ; INTERLEAVE: vector.ph: ; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i32 [[TMP0]], -8 -; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i32 [[N_VEC]], [[I]] +; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i32 [[I]], [[N_VEC]] ; INTERLEAVE-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[I]], i64 0 ; INTERLEAVE-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer ; INTERLEAVE-NEXT: [[INDUCTION:%.*]] = add <4 x i32> [[DOTSPLAT]], <i32 0, i32 1, i32 2, i32 3> @@ -4832,7 +4832,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) { ; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] ; INTERLEAVE-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4> -; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[INDEX]], [[I]] +; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[I]], [[INDEX]] ; INTERLEAVE-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64 ; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]] ; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16 @@ -6307,7 +6307,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n ; IND: vector.ph: ; IND-NEXT: [[N_VEC:%.*]] = and i64 [[N]], -2 ; IND-NEXT: [[DOTCAST:%.*]] = trunc i64 [[N_VEC]] to i32 -; IND-NEXT: [[IND_END:%.*]] = mul i32 [[DOTCAST]], [[STEP]] +; IND-NEXT: [[IND_END:%.*]] = mul i32 [[STEP]], [[DOTCAST]] ; IND-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[STEP]], i64 0 ; IND-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer ; IND-NEXT: [[TMP15:%.*]] = mul nuw <2 x i32> [[DOTSPLAT]], <i32 0, i32 1> @@ -6328,7 +6328,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n ; IND-NEXT: br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP54:![0-9]+]] ; IND: middle.block: ; IND-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <2 x i32> [[VEC_IND]], i64 1 -; IND-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]] +; IND-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]] ; IND-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; IND: scalar.ph: ; IND-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] @@ -6378,7 +6378,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n ; UNROLL: vector.ph: ; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[N]], -4 ; UNROLL-NEXT: [[DOTCAST:%.*]] = trunc i64 [[N_VEC]] to i32 -; UNROLL-NEXT: [[IND_END:%.*]] = mul i32 [[DOTCAST]], [[STEP]] +; UNROLL-NEXT: [[IND_END:%.*]] = mul i32 [[STEP]], [[DOTCAST]] ; UNROLL-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[STEP]], i64 0 ; UNROLL-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x i32> [[DOTSPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer ; UNROLL-NEXT: [[TMP15:%.*]] = mul nuw <2 x i32> [[DOTSPLAT]], <i32 0, i32 1> @@ -6403,7 +6403,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n ; UNROLL-NEXT: br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP54:![0-9]+]] ; UNROLL: middle.block: ; UNROLL-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <2 x i32> [[STEP_ADD]], i64 1 -; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]] +; UNROLL-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]] ; UNROLL-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; UNROLL: scalar.ph: ; UNROLL-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] @@ -6536,7 +6536,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n ; INTERLEAVE: vector.ph: ; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[N]], -8 ; INTERLEAVE-NEXT: [[DOTCAST:%.*]] = trunc i64 [[N_VEC]] to i32 -; INTERLEAVE-NEXT: [[IND_END:%.*]] = mul i32 [[DOTCAST]], [[STEP]] +; INTERLEAVE-NEXT: [[IND_END:%.*]] = mul i32 [[STEP]], [[DOTCAST]] ; INTERLEAVE-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[STEP]], i64 0 ; INTERLEAVE-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x i32> [[DOTSPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer ; INTERLEAVE-NEXT: [[TMP15:%.*]] = mul <4 x i32> [[DOTSPLAT]], <i32 0, i32 1, i32 2, i32 3> @@ -6561,7 +6561,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n ; INTERLEAVE-NEXT: br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP54:![0-9]+]] ; INTERLEAVE: middle.block: ; INTERLEAVE-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i32> [[STEP_ADD]], i64 3 -; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]] +; INTERLEAVE-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]] ; INTERLEAVE-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] ; INTERLEAVE: scalar.ph: ; INTERLEAVE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] |