2016af5652
Move creation of the minimum iteration check for the epilogue vector loop to VPlan. This is a first step towards breaking up and moving skeleton creation for epilogue vectorization to VPlan. It moves most logic out of EpilogueVectorizerEpilogueLoop: the minimum iteration check is created directly in VPlan, connecting the check blocks from the main vector loop is done as post-processing. Next steps are to move connecting and updating the branches from the check blocks to VPlan, as well as updating the incoming values for phis. Test changes are improvements due to folding of live-ins. PR: https://github.com/llvm/llvm-project/pull/157545
153 lines
11 KiB
LLVM
153 lines
11 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt < %s -passes=loop-vectorize -S | FileCheck %s
|
|
|
|
target datalayout = "e-m:e-i64:64-n32:64"
|
|
target triple = "powerpc64le-ibm-linux-gnu"
|
|
|
|
; We expect the loop with double reductions to be interleaved 8 times.
|
|
define void @test(ptr %arr, i32 %len) {
|
|
; CHECK-LABEL: define void @test(
|
|
; CHECK-SAME: ptr [[ARR:%.*]], i32 [[LEN:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*]]:
|
|
; CHECK-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[LEN]], 0
|
|
; CHECK-NEXT: br i1 [[CMP4]], label %[[ITER_CHECK:.*]], label %[[FOR_END:.*]]
|
|
; CHECK: [[ITER_CHECK]]:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[LEN]], -1
|
|
; CHECK-NEXT: [[TMP1:%.*]] = zext i32 [[LEN]] to i64
|
|
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP1]], 2
|
|
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH:.*]], label %[[VECTOR_MAIN_LOOP_ITER_CHECK:.*]]
|
|
; CHECK: [[VECTOR_MAIN_LOOP_ITER_CHECK]]:
|
|
; CHECK-NEXT: [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[TMP1]], 16
|
|
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK1]], label %[[VEC_EPILOG_PH:.*]], label %[[VECTOR_PH:.*]]
|
|
; CHECK: [[VECTOR_PH]]:
|
|
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP1]], 16
|
|
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP1]], [[N_MOD_VF]]
|
|
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
|
|
; CHECK: [[VECTOR_BODY]]:
|
|
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP12:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI2:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP13:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI3:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP14:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI4:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP15:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI5:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP16:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI6:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP17:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI7:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP18:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI8:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP19:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds double, ptr [[ARR]], i64 [[INDEX]]
|
|
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 2
|
|
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 4
|
|
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 6
|
|
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 8
|
|
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 10
|
|
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 12
|
|
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i32 14
|
|
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <2 x double>, ptr [[TMP5]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD10:%.*]] = load <2 x double>, ptr [[TMP6]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD11:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <2 x double>, ptr [[TMP8]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <2 x double>, ptr [[TMP9]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD14:%.*]] = load <2 x double>, ptr [[TMP10]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD15:%.*]] = load <2 x double>, ptr [[TMP11]], align 8
|
|
; CHECK-NEXT: [[TMP12]] = fadd fast <2 x double> [[WIDE_LOAD]], [[VEC_PHI]]
|
|
; CHECK-NEXT: [[TMP13]] = fadd fast <2 x double> [[WIDE_LOAD9]], [[VEC_PHI2]]
|
|
; CHECK-NEXT: [[TMP14]] = fadd fast <2 x double> [[WIDE_LOAD10]], [[VEC_PHI3]]
|
|
; CHECK-NEXT: [[TMP15]] = fadd fast <2 x double> [[WIDE_LOAD11]], [[VEC_PHI4]]
|
|
; CHECK-NEXT: [[TMP16]] = fadd fast <2 x double> [[WIDE_LOAD12]], [[VEC_PHI5]]
|
|
; CHECK-NEXT: [[TMP17]] = fadd fast <2 x double> [[WIDE_LOAD13]], [[VEC_PHI6]]
|
|
; CHECK-NEXT: [[TMP18]] = fadd fast <2 x double> [[WIDE_LOAD14]], [[VEC_PHI7]]
|
|
; CHECK-NEXT: [[TMP19]] = fadd fast <2 x double> [[WIDE_LOAD15]], [[VEC_PHI8]]
|
|
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
|
|
; CHECK-NEXT: [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
|
|
; CHECK-NEXT: br i1 [[TMP20]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; CHECK: [[MIDDLE_BLOCK]]:
|
|
; CHECK-NEXT: [[BIN_RDX:%.*]] = fadd fast <2 x double> [[TMP13]], [[TMP12]]
|
|
; CHECK-NEXT: [[BIN_RDX16:%.*]] = fadd fast <2 x double> [[TMP14]], [[BIN_RDX]]
|
|
; CHECK-NEXT: [[BIN_RDX17:%.*]] = fadd fast <2 x double> [[TMP15]], [[BIN_RDX16]]
|
|
; CHECK-NEXT: [[BIN_RDX18:%.*]] = fadd fast <2 x double> [[TMP16]], [[BIN_RDX17]]
|
|
; CHECK-NEXT: [[BIN_RDX19:%.*]] = fadd fast <2 x double> [[TMP17]], [[BIN_RDX18]]
|
|
; CHECK-NEXT: [[BIN_RDX20:%.*]] = fadd fast <2 x double> [[TMP18]], [[BIN_RDX19]]
|
|
; CHECK-NEXT: [[BIN_RDX21:%.*]] = fadd fast <2 x double> [[TMP19]], [[BIN_RDX20]]
|
|
; CHECK-NEXT: [[TMP21:%.*]] = call fast double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> [[BIN_RDX21]])
|
|
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC]]
|
|
; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END_LOOPEXIT:.*]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
|
|
; CHECK: [[VEC_EPILOG_ITER_CHECK]]:
|
|
; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_MOD_VF]], 2
|
|
; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]], !prof [[PROF3:![0-9]+]]
|
|
; CHECK: [[VEC_EPILOG_PH]]:
|
|
; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
|
|
; CHECK-NEXT: [[BC_MERGE_RDX:%.*]] = phi double [ [[TMP21]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0.000000e+00, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
|
|
; CHECK-NEXT: [[N_MOD_VF22:%.*]] = urem i64 [[TMP1]], 2
|
|
; CHECK-NEXT: [[N_VEC23:%.*]] = sub i64 [[TMP1]], [[N_MOD_VF22]]
|
|
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <2 x double> zeroinitializer, double [[BC_MERGE_RDX]], i32 0
|
|
; CHECK-NEXT: br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
|
|
; CHECK: [[VEC_EPILOG_VECTOR_BODY]]:
|
|
; CHECK-NEXT: [[TMP23:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT27:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_PHI25:%.*]] = phi <2 x double> [ [[TMP22]], %[[VEC_EPILOG_PH]] ], [ [[TMP26:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds double, ptr [[ARR]], i64 [[TMP23]]
|
|
; CHECK-NEXT: [[WIDE_LOAD26:%.*]] = load <2 x double>, ptr [[TMP24]], align 8
|
|
; CHECK-NEXT: [[TMP26]] = fadd fast <2 x double> [[WIDE_LOAD26]], [[VEC_PHI25]]
|
|
; CHECK-NEXT: [[INDEX_NEXT27]] = add nuw i64 [[TMP23]], 2
|
|
; CHECK-NEXT: [[TMP27:%.*]] = icmp eq i64 [[INDEX_NEXT27]], [[N_VEC23]]
|
|
; CHECK-NEXT: br i1 [[TMP27]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
|
|
; CHECK: [[VEC_EPILOG_MIDDLE_BLOCK]]:
|
|
; CHECK-NEXT: [[TMP28:%.*]] = call fast double @llvm.vector.reduce.fadd.v2f64(double 0.000000e+00, <2 x double> [[TMP26]])
|
|
; CHECK-NEXT: [[CMP_N28:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC23]]
|
|
; CHECK-NEXT: br i1 [[CMP_N28]], label %[[FOR_END_LOOPEXIT]], label %[[VEC_EPILOG_SCALAR_PH]]
|
|
; CHECK: [[VEC_EPILOG_SCALAR_PH]]:
|
|
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC23]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[ITER_CHECK]] ]
|
|
; CHECK-NEXT: [[BC_MERGE_RDX29:%.*]] = phi double [ [[TMP28]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[TMP21]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0.000000e+00, %[[ITER_CHECK]] ]
|
|
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
|
|
; CHECK: [[FOR_BODY]]:
|
|
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
|
|
; CHECK-NEXT: [[REDX_05:%.*]] = phi double [ [[BC_MERGE_RDX29]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
|
|
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, ptr [[ARR]], i64 [[INDVARS_IV]]
|
|
; CHECK-NEXT: [[TMP29:%.*]] = load double, ptr [[ARRAYIDX]], align 8
|
|
; CHECK-NEXT: [[ADD]] = fadd fast double [[TMP29]], [[REDX_05]]
|
|
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
|
|
; CHECK-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV]] to i32
|
|
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[TMP0]]
|
|
; CHECK-NEXT: br i1 [[EXITCOND]], label %[[FOR_END_LOOPEXIT]], label %[[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
|
|
; CHECK: [[FOR_END_LOOPEXIT]]:
|
|
; CHECK-NEXT: [[ADD_LCSSA:%.*]] = phi double [ [[ADD]], %[[FOR_BODY]] ], [ [[TMP21]], %[[MIDDLE_BLOCK]] ], [ [[TMP28]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ]
|
|
; CHECK-NEXT: br label %[[FOR_END]]
|
|
; CHECK: [[FOR_END]]:
|
|
; CHECK-NEXT: [[REDX_0_LCSSA:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[ADD_LCSSA]], %[[FOR_END_LOOPEXIT]] ]
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
%cmp4 = icmp sgt i32 %len, 0
|
|
br i1 %cmp4, label %for.body.lr.ph, label %for.end
|
|
|
|
for.body.lr.ph: ; preds = %entry
|
|
%0 = add i32 %len, -1
|
|
br label %for.body
|
|
|
|
for.body: ; preds = %for.body, %for.body.lr.ph
|
|
%indvars.iv = phi i64 [ 0, %for.body.lr.ph ], [ %indvars.iv.next, %for.body ]
|
|
%redx.05 = phi double [ 0.000000e+00, %for.body.lr.ph ], [ %add, %for.body ]
|
|
%arrayidx = getelementptr inbounds double, ptr %arr, i64 %indvars.iv
|
|
%1 = load double, ptr %arrayidx, align 8
|
|
%add = fadd fast double %1, %redx.05
|
|
%indvars.iv.next = add i64 %indvars.iv, 1
|
|
%lftr.wideiv = trunc i64 %indvars.iv to i32
|
|
%exitcond = icmp eq i32 %lftr.wideiv, %0
|
|
br i1 %exitcond, label %for.end.loopexit, label %for.body
|
|
|
|
for.end.loopexit: ; preds = %for.body
|
|
%add.lcssa = phi double [ %add, %for.body ]
|
|
br label %for.end
|
|
|
|
for.end: ; preds = %for.end.loopexit, %entry
|
|
%redx.0.lcssa = phi double [ 0.000000e+00, %entry ], [ %add.lcssa, %for.end.loopexit ]
|
|
ret void
|
|
}
|
|
;.
|
|
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
|
|
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
|
|
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
|
|
; CHECK: [[PROF3]] = !{!"branch_weights", i32 2, i32 14}
|
|
; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
|
|
; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META1]]}
|
|
;.
|