diff options
Diffstat (limited to 'llvm/test/Transforms/LoopVectorize/float-induction.ll')
| -rw-r--r-- | llvm/test/Transforms/LoopVectorize/float-induction.ll | 208 |
1 files changed, 104 insertions, 104 deletions
diff --git a/llvm/test/Transforms/LoopVectorize/float-induction.ll b/llvm/test/Transforms/LoopVectorize/float-induction.ll index 89ac22b054e2..601c475df56c 100644 --- a/llvm/test/Transforms/LoopVectorize/float-induction.ll +++ b/llvm/test/Transforms/LoopVectorize/float-induction.ll @@ -61,7 +61,7 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N) ; VEC4_INTERL1: for.body: ; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL1-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL1-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]] ; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -102,7 +102,7 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N) ; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fsub fast <4 x float> [[VEC_IND]], [[DOTSPLAT5]] ; VEC4_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP4]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP4]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP5]], align 4 ; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8 @@ -119,7 +119,7 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N) ; VEC4_INTERL2: for.body: ; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL2-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]] ; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -170,7 +170,7 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N) ; VEC1_INTERL2: for.body: ; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC1_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4 ; VEC1_INTERL2-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]] ; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -308,7 +308,7 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32 ; VEC4_INTERL1: for.body: ; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL1-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL1-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]] ; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -349,7 +349,7 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32 ; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fsub reassoc <4 x float> [[VEC_IND]], [[DOTSPLAT5]] ; VEC4_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP4]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP4]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP5]], align 4 ; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8 @@ -366,7 +366,7 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32 ; VEC4_INTERL2: for.body: ; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL2-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]] ; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -419,7 +419,7 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32 ; VEC1_INTERL2: for.body: ; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC1_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4 ; VEC1_INTERL2-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]] ; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -552,7 +552,7 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 { ; VEC4_INTERL1: for.body: ; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC4_INTERL1-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL1-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01 ; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -586,7 +586,7 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 { ; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fadd fast <4 x float> [[VEC_IND]], splat (float 2.000000e+00) ; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP2]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP3]], align 4 ; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8 @@ -603,7 +603,7 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 { ; VEC4_INTERL2: for.body: ; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC4_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01 ; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -653,7 +653,7 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 { ; VEC1_INTERL2: for.body: ; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC1_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4 ; VEC1_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01 ; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -806,14 +806,14 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca ; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL1-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[CONV1:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL1-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL1-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]] ; VEC4_INTERL1-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01 ; VEC4_INTERL1-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]] -; VEC4_INTERL1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4 -; VEC4_INTERL1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4 ; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 ; VEC4_INTERL1-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32 @@ -857,7 +857,7 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca ; VEC4_INTERL2-NEXT: [[VEC_IND10:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT13:%.*]], [[VECTOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[STEP_ADD11:%.*]] = fadd fast <4 x float> [[VEC_IND10]], [[DOTSPLAT5]] ; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP6]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND10]], ptr [[TMP6]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD11]], ptr [[TMP7]], align 4 ; VEC4_INTERL2-NEXT: [[TMP8:%.*]] = fadd fast <4 x float> [[VEC_IND10]], [[BROADCAST]] @@ -867,11 +867,11 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca ; VEC4_INTERL2-NEXT: [[TMP12:%.*]] = fadd fast <4 x float> [[TMP10]], [[TMP8]] ; VEC4_INTERL2-NEXT: [[TMP13:%.*]] = fadd fast <4 x float> [[TMP11]], [[TMP9]] ; VEC4_INTERL2-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[TMP12]], ptr [[TMP14]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[TMP13]], ptr [[TMP15]], align 4 ; VEC4_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP16]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP16]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[TMP10]], ptr [[TMP16]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[TMP11]], ptr [[TMP17]], align 4 ; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8 @@ -891,14 +891,14 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca ; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[CONV1:%.*]], [[FOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL2-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]] ; VEC4_INTERL2-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01 ; VEC4_INTERL2-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]] -; VEC4_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4 -; VEC4_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4 ; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 ; VEC4_INTERL2-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32 @@ -969,14 +969,14 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca ; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ] ; VEC1_INTERL2-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[CONV1:%.*]], [[FOR_BODY]] ] ; VEC1_INTERL2-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ] -; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4 ; VEC1_INTERL2-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]] ; VEC1_INTERL2-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01 ; VEC1_INTERL2-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]] -; VEC1_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4 -; VEC1_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4 ; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 ; VEC1_INTERL2-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32 @@ -1134,7 +1134,7 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) { ; VEC4_INTERL1: for.body: ; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC4_INTERL1-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL1-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL1-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01 ; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -1165,7 +1165,7 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) { ; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ <float 1.000000e+00, float 1.500000e+00, float 2.000000e+00, float 2.500000e+00>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ] ; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fadd fast <4 x float> [[VEC_IND]], splat (float 2.000000e+00) ; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 16 ; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP2]], align 4 ; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP3]], align 4 ; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8 @@ -1182,7 +1182,7 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) { ; VEC4_INTERL2: for.body: ; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC4_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC4_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4 ; VEC4_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01 ; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -1232,7 +1232,7 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) { ; VEC1_INTERL2: for.body: ; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC1_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]] +; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]] ; VEC1_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4 ; VEC1_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01 ; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1 @@ -1320,7 +1320,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC4_INTERL1-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float ; VEC4_INTERL1-NEXT: br label [[VECTOR_BODY:%.*]] ; VEC4_INTERL1: vector.body: -; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE8:%.*]] ] +; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE7:%.*]] ] ; VEC4_INTERL1-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float ; VEC4_INTERL1-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] ; VEC4_INTERL1-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4 @@ -1333,32 +1333,32 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE]] ; VEC4_INTERL1: pred.store.continue: ; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1 -; VEC4_INTERL1-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]] -; VEC4_INTERL1: pred.store.if3: +; VEC4_INTERL1-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF2:%.*]], label [[PRED_STORE_CONTINUE3:%.*]] +; VEC4_INTERL1: pred.store.if2: ; VEC4_INTERL1-NEXT: [[TMP5:%.*]] = or disjoint i64 [[INDEX]], 1 ; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP5]] ; VEC4_INTERL1-NEXT: [[TMP7:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00 ; VEC4_INTERL1-NEXT: store float [[TMP7]], ptr [[TMP6]], align 4 -; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE4]] -; VEC4_INTERL1: pred.store.continue4: +; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE3]] +; VEC4_INTERL1: pred.store.continue3: ; VEC4_INTERL1-NEXT: [[TMP8:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2 -; VEC4_INTERL1-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6:%.*]] -; VEC4_INTERL1: pred.store.if5: +; VEC4_INTERL1-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF4:%.*]], label [[PRED_STORE_CONTINUE5:%.*]] +; VEC4_INTERL1: pred.store.if4: ; VEC4_INTERL1-NEXT: [[TMP9:%.*]] = or disjoint i64 [[INDEX]], 2 ; VEC4_INTERL1-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP9]] ; VEC4_INTERL1-NEXT: [[TMP11:%.*]] = fadd fast float [[DOTCAST2]], 2.000000e+00 ; VEC4_INTERL1-NEXT: store float [[TMP11]], ptr [[TMP10]], align 4 -; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE6]] -; VEC4_INTERL1: pred.store.continue6: +; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE5]] +; VEC4_INTERL1: pred.store.continue5: ; VEC4_INTERL1-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3 -; VEC4_INTERL1-NEXT: br i1 [[TMP12]], label [[PRED_STORE_IF7:%.*]], label [[PRED_STORE_CONTINUE8]] -; VEC4_INTERL1: pred.store.if7: +; VEC4_INTERL1-NEXT: br i1 [[TMP12]], label [[PRED_STORE_IF6:%.*]], label [[PRED_STORE_CONTINUE7]] +; VEC4_INTERL1: pred.store.if6: ; VEC4_INTERL1-NEXT: [[TMP13:%.*]] = or disjoint i64 [[INDEX]], 3 ; VEC4_INTERL1-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP13]] ; VEC4_INTERL1-NEXT: [[TMP15:%.*]] = fadd fast float [[DOTCAST2]], 3.000000e+00 ; VEC4_INTERL1-NEXT: store float [[TMP15]], ptr [[TMP14]], align 4 -; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE8]] -; VEC4_INTERL1: pred.store.continue8: +; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE7]] +; VEC4_INTERL1: pred.store.continue7: ; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4 ; VEC4_INTERL1-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; VEC4_INTERL1-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]] @@ -1372,7 +1372,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC4_INTERL1: for.body: ; VEC4_INTERL1-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC4_INTERL1-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC4_INTERL1-NEXT: [[VAR0:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]] +; VEC4_INTERL1-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I]] ; VEC4_INTERL1-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4 ; VEC4_INTERL1-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00 ; VEC4_INTERL1-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]] @@ -1397,10 +1397,10 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC4_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float ; VEC4_INTERL2-NEXT: br label [[VECTOR_BODY:%.*]] ; VEC4_INTERL2: vector.body: -; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE17:%.*]] ] +; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE16:%.*]] ] ; VEC4_INTERL2-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float ; VEC4_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16 +; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16 ; VEC4_INTERL2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP1]], align 4 ; VEC4_INTERL2-NEXT: [[WIDE_LOAD3:%.*]] = load <4 x float>, ptr [[TMP2]], align 4 ; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = fcmp fast oeq <4 x float> [[WIDE_LOAD]], zeroinitializer @@ -1408,73 +1408,73 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP3]], i64 0 ; VEC4_INTERL2-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]] ; VEC4_INTERL2: pred.store.if: -; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]] -; VEC4_INTERL2-NEXT: store float [[DOTCAST2]], ptr [[TMP6]], align 4 +; VEC4_INTERL2-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]] +; VEC4_INTERL2-NEXT: store float [[DOTCAST2]], ptr [[TMP35]], align 4 ; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE]] ; VEC4_INTERL2: pred.store.continue: -; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1 -; VEC4_INTERL2-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF4:%.*]], label [[PRED_STORE_CONTINUE5:%.*]] -; VEC4_INTERL2: pred.store.if4: -; VEC4_INTERL2-NEXT: [[TMP8:%.*]] = or disjoint i64 [[INDEX]], 1 -; VEC4_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP8]] -; VEC4_INTERL2-NEXT: [[TMP10:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00 -; VEC4_INTERL2-NEXT: store float [[TMP10]], ptr [[TMP9]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE5]] -; VEC4_INTERL2: pred.store.continue5: -; VEC4_INTERL2-NEXT: [[TMP11:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2 -; VEC4_INTERL2-NEXT: br i1 [[TMP11]], label [[PRED_STORE_IF6:%.*]], label [[PRED_STORE_CONTINUE7:%.*]] -; VEC4_INTERL2: pred.store.if6: -; VEC4_INTERL2-NEXT: [[TMP12:%.*]] = or disjoint i64 [[INDEX]], 2 -; VEC4_INTERL2-NEXT: [[TMP13:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP12]] -; VEC4_INTERL2-NEXT: [[TMP14:%.*]] = fadd fast float [[DOTCAST2]], 2.000000e+00 -; VEC4_INTERL2-NEXT: store float [[TMP14]], ptr [[TMP13]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE7]] -; VEC4_INTERL2: pred.store.continue7: -; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3 -; VEC4_INTERL2-NEXT: br i1 [[TMP15]], label [[PRED_STORE_IF8:%.*]], label [[PRED_STORE_CONTINUE9:%.*]] -; VEC4_INTERL2: pred.store.if8: -; VEC4_INTERL2-NEXT: [[TMP16:%.*]] = or disjoint i64 [[INDEX]], 3 -; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP16]] -; VEC4_INTERL2-NEXT: [[TMP18:%.*]] = fadd fast float [[DOTCAST2]], 3.000000e+00 -; VEC4_INTERL2-NEXT: store float [[TMP18]], ptr [[TMP17]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE9]] -; VEC4_INTERL2: pred.store.continue9: -; VEC4_INTERL2-NEXT: [[TMP19:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0 -; VEC4_INTERL2-NEXT: br i1 [[TMP19]], label [[PRED_STORE_IF10:%.*]], label [[PRED_STORE_CONTINUE11:%.*]] -; VEC4_INTERL2: pred.store.if10: -; VEC4_INTERL2-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 4 -; VEC4_INTERL2-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP0]] +; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1 +; VEC4_INTERL2-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]] +; VEC4_INTERL2: pred.store.if3: +; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = or disjoint i64 [[INDEX]], 1 +; VEC4_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP7]] +; VEC4_INTERL2-NEXT: [[TMP9:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00 +; VEC4_INTERL2-NEXT: store float [[TMP9]], ptr [[TMP8]], align 4 +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE4]] +; VEC4_INTERL2: pred.store.continue4: +; VEC4_INTERL2-NEXT: [[TMP10:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2 +; VEC4_INTERL2-NEXT: br i1 [[TMP10]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6:%.*]] +; VEC4_INTERL2: pred.store.if5: +; VEC4_INTERL2-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 2 +; VEC4_INTERL2-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP11]] +; VEC4_INTERL2-NEXT: [[TMP13:%.*]] = fadd fast float [[DOTCAST2]], 2.000000e+00 +; VEC4_INTERL2-NEXT: store float [[TMP13]], ptr [[TMP12]], align 4 +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE6]] +; VEC4_INTERL2: pred.store.continue6: +; VEC4_INTERL2-NEXT: [[TMP14:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3 +; VEC4_INTERL2-NEXT: br i1 [[TMP14]], label [[PRED_STORE_IF7:%.*]], label [[PRED_STORE_CONTINUE8:%.*]] +; VEC4_INTERL2: pred.store.if7: +; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = or disjoint i64 [[INDEX]], 3 +; VEC4_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP15]] +; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = fadd fast float [[DOTCAST2]], 3.000000e+00 +; VEC4_INTERL2-NEXT: store float [[TMP17]], ptr [[TMP16]], align 4 +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE8]] +; VEC4_INTERL2: pred.store.continue8: +; VEC4_INTERL2-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0 +; VEC4_INTERL2-NEXT: br i1 [[TMP18]], label [[PRED_STORE_IF9:%.*]], label [[PRED_STORE_CONTINUE10:%.*]] +; VEC4_INTERL2: pred.store.if9: +; VEC4_INTERL2-NEXT: [[TMP19:%.*]] = or disjoint i64 [[INDEX]], 4 +; VEC4_INTERL2-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP19]] ; VEC4_INTERL2-NEXT: [[TMP21:%.*]] = fadd fast float [[DOTCAST2]], 4.000000e+00 ; VEC4_INTERL2-NEXT: store float [[TMP21]], ptr [[TMP20]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE11]] -; VEC4_INTERL2: pred.store.continue11: +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE10]] +; VEC4_INTERL2: pred.store.continue10: ; VEC4_INTERL2-NEXT: [[TMP22:%.*]] = extractelement <4 x i1> [[TMP4]], i64 1 -; VEC4_INTERL2-NEXT: br i1 [[TMP22]], label [[PRED_STORE_IF12:%.*]], label [[PRED_STORE_CONTINUE13:%.*]] -; VEC4_INTERL2: pred.store.if12: +; VEC4_INTERL2-NEXT: br i1 [[TMP22]], label [[PRED_STORE_IF11:%.*]], label [[PRED_STORE_CONTINUE12:%.*]] +; VEC4_INTERL2: pred.store.if11: ; VEC4_INTERL2-NEXT: [[TMP23:%.*]] = or disjoint i64 [[INDEX]], 5 ; VEC4_INTERL2-NEXT: [[TMP24:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP23]] ; VEC4_INTERL2-NEXT: [[TMP25:%.*]] = fadd fast float [[DOTCAST2]], 5.000000e+00 ; VEC4_INTERL2-NEXT: store float [[TMP25]], ptr [[TMP24]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE13]] -; VEC4_INTERL2: pred.store.continue13: +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE12]] +; VEC4_INTERL2: pred.store.continue12: ; VEC4_INTERL2-NEXT: [[TMP26:%.*]] = extractelement <4 x i1> [[TMP4]], i64 2 -; VEC4_INTERL2-NEXT: br i1 [[TMP26]], label [[PRED_STORE_IF14:%.*]], label [[PRED_STORE_CONTINUE15:%.*]] -; VEC4_INTERL2: pred.store.if14: +; VEC4_INTERL2-NEXT: br i1 [[TMP26]], label [[PRED_STORE_IF13:%.*]], label [[PRED_STORE_CONTINUE14:%.*]] +; VEC4_INTERL2: pred.store.if13: ; VEC4_INTERL2-NEXT: [[TMP27:%.*]] = or disjoint i64 [[INDEX]], 6 ; VEC4_INTERL2-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP27]] ; VEC4_INTERL2-NEXT: [[TMP29:%.*]] = fadd fast float [[DOTCAST2]], 6.000000e+00 ; VEC4_INTERL2-NEXT: store float [[TMP29]], ptr [[TMP28]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE15]] -; VEC4_INTERL2: pred.store.continue15: +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE14]] +; VEC4_INTERL2: pred.store.continue14: ; VEC4_INTERL2-NEXT: [[TMP30:%.*]] = extractelement <4 x i1> [[TMP4]], i64 3 -; VEC4_INTERL2-NEXT: br i1 [[TMP30]], label [[PRED_STORE_IF16:%.*]], label [[PRED_STORE_CONTINUE17]] -; VEC4_INTERL2: pred.store.if16: +; VEC4_INTERL2-NEXT: br i1 [[TMP30]], label [[PRED_STORE_IF15:%.*]], label [[PRED_STORE_CONTINUE16]] +; VEC4_INTERL2: pred.store.if15: ; VEC4_INTERL2-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 7 ; VEC4_INTERL2-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP31]] ; VEC4_INTERL2-NEXT: [[TMP33:%.*]] = fadd fast float [[DOTCAST2]], 7.000000e+00 ; VEC4_INTERL2-NEXT: store float [[TMP33]], ptr [[TMP32]], align 4 -; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE17]] -; VEC4_INTERL2: pred.store.continue17: +; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE16]] +; VEC4_INTERL2: pred.store.continue16: ; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8 ; VEC4_INTERL2-NEXT: [[TMP34:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; VEC4_INTERL2-NEXT: br i1 [[TMP34]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]] @@ -1488,7 +1488,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC4_INTERL2: for.body: ; VEC4_INTERL2-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC4_INTERL2-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC4_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]] +; VEC4_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I]] ; VEC4_INTERL2-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4 ; VEC4_INTERL2-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00 ; VEC4_INTERL2-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]] @@ -1513,7 +1513,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC1_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float ; VEC1_INTERL2-NEXT: br label [[VECTOR_BODY:%.*]] ; VEC1_INTERL2: vector.body: -; VEC1_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE4:%.*]] ] +; VEC1_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE3:%.*]] ] ; VEC1_INTERL2-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 1 ; VEC1_INTERL2-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float ; VEC1_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] @@ -1527,12 +1527,12 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC1_INTERL2-NEXT: store float [[DOTCAST2]], ptr [[TMP1]], align 4 ; VEC1_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE]] ; VEC1_INTERL2: pred.store.continue: -; VEC1_INTERL2-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4]] -; VEC1_INTERL2: pred.store.if3: +; VEC1_INTERL2-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF2:%.*]], label [[PRED_STORE_CONTINUE3]] +; VEC1_INTERL2: pred.store.if2: ; VEC1_INTERL2-NEXT: [[TMP7:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00 ; VEC1_INTERL2-NEXT: store float [[TMP7]], ptr [[TMP2]], align 4 -; VEC1_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE4]] -; VEC1_INTERL2: pred.store.continue4: +; VEC1_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE3]] +; VEC1_INTERL2: pred.store.continue3: ; VEC1_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2 ; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; VEC1_INTERL2-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]] @@ -1546,7 +1546,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC1_INTERL2: for.body: ; VEC1_INTERL2-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; VEC1_INTERL2-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ] -; VEC1_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]] +; VEC1_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I]] ; VEC1_INTERL2-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4 ; VEC1_INTERL2-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00 ; VEC1_INTERL2-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]] @@ -1571,7 +1571,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC2_INTERL1_PRED_STORE-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float ; VEC2_INTERL1_PRED_STORE-NEXT: br label [[VECTOR_BODY:%.*]] ; VEC2_INTERL1_PRED_STORE: vector.body: -; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE4:%.*]] ] +; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE3:%.*]] ] ; VEC2_INTERL1_PRED_STORE-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float ; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]] ; VEC2_INTERL1_PRED_STORE-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP0]], align 4 @@ -1584,14 +1584,14 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) { ; VEC2_INTERL1_PRED_STORE-NEXT: br label [[PRED_STORE_CONTINUE]] ; VEC2_INTERL1_PRED_STORE: pred.store.continue: ; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = extractelement <2 x i1> [[TMP1]], i64 1 -; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4]] -; VEC2_INTERL1_PRED_STORE: pred.store.if3: +; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF2:%.*]], label [[PRED_STORE_CONTINUE3]] +; VEC2_INTERL1_PRED_STORE: pred.store.if2: ; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP5:%.*]] = or disjoint i64 [[INDEX]], 1 ; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP5]] ; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP7:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00 ; VEC2_INTERL1_PRED_STORE-NEXT: store float [[TMP7]], ptr [[TMP6]], align 4 -; VEC2_INTERL1_PRED_STORE-NEXT: br label [[PRED_STORE_CONTINUE4]] -; VEC2_INTERL1_PRED_STORE: pred.store.continue4: +; VEC2_INTERL1_PRED_STORE-NEXT: br label [[PRED_STORE_CONTINUE3]] +; VEC2_INTERL1_PRED_STORE: pred.store.continue3: ; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2 ; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] ; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]] |