2ac5ebedea
SLP vectorizer emits extracts for externally used vectorized scalars and estimates the cost for each such extract. But in many cases these scalars are input for insertelement instructions, forming buildvector, and instead of extractelement/insertelement pair we can emit/cost estimate shuffle(s) cost and generate series of shuffles, which can be further optimized. Tested using test-suite (+SPEC2017), the tests passed, SLP was able to generate/vectorize more instructions in many cases and it allowed to reduce number of re-vectorization attempts (where we could try to vectorize buildector insertelements again and again). Differential Revision: https://reviews.llvm.org/D107966
135 lines
7.2 KiB
LLVM
135 lines
7.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt < %s -basic-aa -slp-vectorizer -dce -S -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 | FileCheck %s
|
|
|
|
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
|
|
target triple = "x86_64-apple-macosx10.8.0"
|
|
|
|
%class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113 = type { [4 x float] }
|
|
|
|
; Function Attrs: ssp uwtable
|
|
define void @_ZN11HullLibrary15CleanupVerticesEjPK9btVector3jRjPS0_fRS0_(%class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113* %vertices) #0 align 2 {
|
|
; CHECK-LABEL: @_ZN11HullLibrary15CleanupVerticesEjPK9btVector3jRjPS0_fRS0_(
|
|
; CHECK-NEXT: entry:
|
|
; CHECK-NEXT: br i1 undef, label [[RETURN:%.*]], label [[IF_END:%.*]]
|
|
; CHECK: if.end:
|
|
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
|
|
; CHECK: for.body:
|
|
; CHECK-NEXT: br i1 undef, label [[IF_THEN17_1:%.*]], label [[IF_END22_1:%.*]]
|
|
; CHECK: for.end36:
|
|
; CHECK-NEXT: br label [[FOR_BODY144:%.*]]
|
|
; CHECK: for.body144:
|
|
; CHECK-NEXT: br i1 undef, label [[FOR_END227:%.*]], label [[FOR_BODY144]]
|
|
; CHECK: for.end227:
|
|
; CHECK-NEXT: br i1 undef, label [[FOR_END271:%.*]], label [[FOR_BODY233:%.*]]
|
|
; CHECK: for.body233:
|
|
; CHECK-NEXT: br i1 undef, label [[FOR_BODY233]], label [[FOR_END271]]
|
|
; CHECK: for.end271:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = phi <2 x float> [ <float 0x47EFFFFFE0000000, float 0x47EFFFFFE0000000>, [[FOR_END227]] ], [ undef, [[FOR_BODY233]] ]
|
|
; CHECK-NEXT: [[TMP1:%.*]] = fsub <2 x float> undef, [[TMP0]]
|
|
; CHECK-NEXT: br i1 undef, label [[IF_THEN291:%.*]], label [[RETURN]]
|
|
; CHECK: if.then291:
|
|
; CHECK-NEXT: [[TMP2:%.*]] = fmul <2 x float> [[TMP1]], <float 5.000000e-01, float 5.000000e-01>
|
|
; CHECK-NEXT: [[TMP3:%.*]] = fadd <2 x float> [[TMP0]], [[TMP2]]
|
|
; CHECK-NEXT: br i1 undef, label [[IF_END332:%.*]], label [[IF_ELSE319:%.*]]
|
|
; CHECK: if.else319:
|
|
; CHECK-NEXT: br i1 undef, label [[IF_THEN325:%.*]], label [[IF_END327:%.*]]
|
|
; CHECK: if.then325:
|
|
; CHECK-NEXT: br label [[IF_END327]]
|
|
; CHECK: if.end327:
|
|
; CHECK-NEXT: br i1 undef, label [[IF_THEN329:%.*]], label [[IF_END332]]
|
|
; CHECK: if.then329:
|
|
; CHECK-NEXT: br label [[IF_END332]]
|
|
; CHECK: if.end332:
|
|
; CHECK-NEXT: [[TMP4:%.*]] = phi <2 x float> [ [[TMP1]], [[IF_THEN329]] ], [ [[TMP1]], [[IF_END327]] ], [ <float 0x3F847AE140000000, float 0x3F847AE140000000>, [[IF_THEN291]] ]
|
|
; CHECK-NEXT: [[TMP5:%.*]] = fsub <2 x float> [[TMP3]], [[TMP4]]
|
|
; CHECK-NEXT: [[ARRAYIDX_I_I606:%.*]] = getelementptr inbounds [[CLASS_BTVECTOR3_23_221_463_485_507_573_595_683_727_749_815_837_991_1585_1607_1629_1651_1849_2047_2069_2091_2113:%.*]], %class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113* [[VERTICES:%.*]], i64 0, i32 0, i64 0
|
|
; CHECK-NEXT: [[TMP6:%.*]] = bitcast float* [[ARRAYIDX_I_I606]] to <2 x float>*
|
|
; CHECK-NEXT: store <2 x float> [[TMP5]], <2 x float>* [[TMP6]], align 4
|
|
; CHECK-NEXT: br label [[RETURN]]
|
|
; CHECK: return:
|
|
; CHECK-NEXT: ret void
|
|
; CHECK: if.then17.1:
|
|
; CHECK-NEXT: br label [[IF_END22_1]]
|
|
; CHECK: if.end22.1:
|
|
; CHECK-NEXT: br i1 undef, label [[IF_THEN17_2:%.*]], label [[IF_END22_2:%.*]]
|
|
; CHECK: if.then17.2:
|
|
; CHECK-NEXT: br label [[IF_END22_2]]
|
|
; CHECK: if.end22.2:
|
|
; CHECK-NEXT: br i1 undef, label [[FOR_END36:%.*]], label [[FOR_BODY]]
|
|
;
|
|
entry:
|
|
br i1 undef, label %return, label %if.end
|
|
|
|
if.end: ; preds = %entry
|
|
br label %for.body
|
|
|
|
for.body: ; preds = %if.end22.2, %if.end
|
|
br i1 undef, label %if.then17.1, label %if.end22.1
|
|
|
|
for.end36: ; preds = %if.end22.2
|
|
br label %for.body144
|
|
|
|
for.body144: ; preds = %for.body144, %for.end36
|
|
br i1 undef, label %for.end227, label %for.body144
|
|
|
|
for.end227: ; preds = %for.body144
|
|
br i1 undef, label %for.end271, label %for.body233
|
|
|
|
for.body233: ; preds = %for.body233, %for.end227
|
|
br i1 undef, label %for.body233, label %for.end271
|
|
|
|
for.end271: ; preds = %for.body233, %for.end227
|
|
%0 = phi float [ 0x47EFFFFFE0000000, %for.end227 ], [ undef, %for.body233 ]
|
|
%1 = phi float [ 0x47EFFFFFE0000000, %for.end227 ], [ undef, %for.body233 ]
|
|
%sub275 = fsub float undef, %1
|
|
%sub279 = fsub float undef, %0
|
|
br i1 undef, label %if.then291, label %return
|
|
|
|
if.then291: ; preds = %for.end271
|
|
%mul292 = fmul float %sub275, 5.000000e-01
|
|
%add294 = fadd float %1, %mul292
|
|
%mul295 = fmul float %sub279, 5.000000e-01
|
|
%add297 = fadd float %0, %mul295
|
|
br i1 undef, label %if.end332, label %if.else319
|
|
|
|
if.else319: ; preds = %if.then291
|
|
br i1 undef, label %if.then325, label %if.end327
|
|
|
|
if.then325: ; preds = %if.else319
|
|
br label %if.end327
|
|
|
|
if.end327: ; preds = %if.then325, %if.else319
|
|
br i1 undef, label %if.then329, label %if.end332
|
|
|
|
if.then329: ; preds = %if.end327
|
|
br label %if.end332
|
|
|
|
if.end332: ; preds = %if.then329, %if.end327, %if.then291
|
|
%dx272.1 = phi float [ %sub275, %if.then329 ], [ %sub275, %if.end327 ], [ 0x3F847AE140000000, %if.then291 ]
|
|
%dy276.1 = phi float [ undef, %if.then329 ], [ undef, %if.end327 ], [ 0x3F847AE140000000, %if.then291 ]
|
|
%sub334 = fsub float %add294, %dx272.1
|
|
%sub338 = fsub float %add297, %dy276.1
|
|
%arrayidx.i.i606 = getelementptr inbounds %class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113, %class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113* %vertices, i64 0, i32 0, i64 0
|
|
store float %sub334, float* %arrayidx.i.i606, align 4
|
|
%arrayidx3.i607 = getelementptr inbounds %class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113, %class.btVector3.23.221.463.485.507.573.595.683.727.749.815.837.991.1585.1607.1629.1651.1849.2047.2069.2091.2113* %vertices, i64 0, i32 0, i64 1
|
|
store float %sub338, float* %arrayidx3.i607, align 4
|
|
br label %return
|
|
|
|
return: ; preds = %if.end332, %for.end271, %entry
|
|
ret void
|
|
|
|
if.then17.1: ; preds = %for.body
|
|
br label %if.end22.1
|
|
|
|
if.end22.1: ; preds = %if.then17.1, %for.body
|
|
br i1 undef, label %if.then17.2, label %if.end22.2
|
|
|
|
if.then17.2: ; preds = %if.end22.1
|
|
br label %if.end22.2
|
|
|
|
if.end22.2: ; preds = %if.then17.2, %if.end22.1
|
|
br i1 undef, label %for.end36, label %for.body
|
|
}
|
|
|
|
attributes #0 = { ssp uwtable "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "unsafe-fp-math"="false" "use-soft-float"="false" }
|