210 lines
11 KiB
LLVM
210 lines
11 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | 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"
|
|
|
|
; Basic depth-3 chain
|
|
define double @test1(double %A1, double %A2, double %B1, double %B2) {
|
|
; CHECK-LABEL: @test1(
|
|
; CHECK-NEXT: [[X1_V_I1_1:%.*]] = insertelement <2 x double> undef, double [[B1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I1_2:%.*]] = insertelement <2 x double> [[X1_V_I1_1]], double [[B2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1_V_I0_1:%.*]] = insertelement <2 x double> undef, double [[A1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I0_2:%.*]] = insertelement <2 x double> [[X1_V_I0_1]], double [[A2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1:%.*]] = fsub <2 x double> [[X1_V_I0_2]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = fmul <2 x double> [[X1]], [[X1_V_I0_2]]
|
|
; CHECK-NEXT: [[Z1:%.*]] = fadd <2 x double> [[Y1]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Z1_V_R1:%.*]] = extractelement <2 x double> [[Z1]], i32 0
|
|
; CHECK-NEXT: [[Z1_V_R2:%.*]] = extractelement <2 x double> [[Z1]], i32 1
|
|
; CHECK-NEXT: [[R:%.*]] = fmul double [[Z1_V_R1]], [[Z1_V_R2]]
|
|
; CHECK-NEXT: ret double [[R]]
|
|
;
|
|
%X1 = fsub double %A1, %B1
|
|
%X2 = fsub double %A2, %B2
|
|
%Y1 = fmul double %X1, %A1
|
|
%Y2 = fmul double %X2, %A2
|
|
%Z1 = fadd double %Y1, %B1
|
|
%Z2 = fadd double %Y2, %B2
|
|
%R = fmul double %Z1, %Z2
|
|
ret double %R
|
|
}
|
|
|
|
; Basic depth-3 chain (last pair permuted)
|
|
define double @test2(double %A1, double %A2, double %B1, double %B2) {
|
|
; CHECK-LABEL: @test2(
|
|
; CHECK-NEXT: [[X1_V_I1_1:%.*]] = insertelement <2 x double> undef, double [[B1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I1_2:%.*]] = insertelement <2 x double> [[X1_V_I1_1]], double [[B2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1_V_I0_1:%.*]] = insertelement <2 x double> undef, double [[A1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I0_2:%.*]] = insertelement <2 x double> [[X1_V_I0_1]], double [[A2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1:%.*]] = fsub <2 x double> [[X1_V_I0_2]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = fmul <2 x double> [[X1]], [[X1_V_I0_2]]
|
|
; CHECK-NEXT: [[Z1_V_I1_1:%.*]] = insertelement <2 x double> undef, double [[B2]], i32 0
|
|
; CHECK-NEXT: [[Z1_V_I1_2:%.*]] = insertelement <2 x double> [[Z1_V_I1_1]], double [[B1]], i32 1
|
|
; CHECK-NEXT: [[Z2:%.*]] = fadd <2 x double> [[Y1]], [[Z1_V_I1_2]]
|
|
; CHECK-NEXT: [[Z2_V_R1:%.*]] = extractelement <2 x double> [[Z2]], i32 0
|
|
; CHECK-NEXT: [[Z2_V_R2:%.*]] = extractelement <2 x double> [[Z2]], i32 1
|
|
; CHECK-NEXT: [[R:%.*]] = fmul double [[Z2_V_R2]], [[Z2_V_R1]]
|
|
; CHECK-NEXT: ret double [[R]]
|
|
;
|
|
%X1 = fsub double %A1, %B1
|
|
%X2 = fsub double %A2, %B2
|
|
%Y1 = fmul double %X1, %A1
|
|
%Y2 = fmul double %X2, %A2
|
|
%Z1 = fadd double %Y2, %B1
|
|
%Z2 = fadd double %Y1, %B2
|
|
%R = fmul double %Z1, %Z2
|
|
ret double %R
|
|
}
|
|
|
|
; Basic depth-3 chain (last pair first splat)
|
|
define double @test3(double %A1, double %A2, double %B1, double %B2) {
|
|
; CHECK-LABEL: @test3(
|
|
; CHECK-NEXT: [[X1_V_I1_1:%.*]] = insertelement <2 x double> undef, double [[B1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I1_2:%.*]] = insertelement <2 x double> [[X1_V_I1_1]], double [[B2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1_V_I0_1:%.*]] = insertelement <2 x double> undef, double [[A1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I0_2:%.*]] = insertelement <2 x double> [[X1_V_I0_1]], double [[A2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1:%.*]] = fsub <2 x double> [[X1_V_I0_2]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = fmul <2 x double> [[X1]], [[X1_V_I0_2]]
|
|
; CHECK-NEXT: [[Z1_V_I0:%.*]] = shufflevector <2 x double> [[Y1]], <2 x double> undef, <2 x i32> <i32 1, i32 1>
|
|
; CHECK-NEXT: [[Z1:%.*]] = fadd <2 x double> [[Z1_V_I0]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Z1_V_R1:%.*]] = extractelement <2 x double> [[Z1]], i32 0
|
|
; CHECK-NEXT: [[Z1_V_R2:%.*]] = extractelement <2 x double> [[Z1]], i32 1
|
|
; CHECK-NEXT: [[R:%.*]] = fmul double [[Z1_V_R1]], [[Z1_V_R2]]
|
|
; CHECK-NEXT: ret double [[R]]
|
|
;
|
|
%X1 = fsub double %A1, %B1
|
|
%X2 = fsub double %A2, %B2
|
|
%Y1 = fmul double %X1, %A1
|
|
%Y2 = fmul double %X2, %A2
|
|
%Z1 = fadd double %Y2, %B1
|
|
%Z2 = fadd double %Y2, %B2
|
|
%R = fmul double %Z1, %Z2
|
|
ret double %R
|
|
}
|
|
|
|
; Basic depth-3 chain (last pair second splat)
|
|
define double @test4(double %A1, double %A2, double %B1, double %B2) {
|
|
; CHECK-LABEL: @test4(
|
|
; CHECK-NEXT: [[X1_V_I1_1:%.*]] = insertelement <2 x double> undef, double [[B1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I1_2:%.*]] = insertelement <2 x double> [[X1_V_I1_1]], double [[B2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1_V_I0_1:%.*]] = insertelement <2 x double> undef, double [[A1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I0_2:%.*]] = insertelement <2 x double> [[X1_V_I0_1]], double [[A2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1:%.*]] = fsub <2 x double> [[X1_V_I0_2]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = fmul <2 x double> [[X1]], [[X1_V_I0_2]]
|
|
; CHECK-NEXT: [[Z1_V_I0:%.*]] = shufflevector <2 x double> [[Y1]], <2 x double> undef, <2 x i32> zeroinitializer
|
|
; CHECK-NEXT: [[Z1:%.*]] = fadd <2 x double> [[Z1_V_I0]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Z1_V_R1:%.*]] = extractelement <2 x double> [[Z1]], i32 0
|
|
; CHECK-NEXT: [[Z1_V_R2:%.*]] = extractelement <2 x double> [[Z1]], i32 1
|
|
; CHECK-NEXT: [[R:%.*]] = fmul double [[Z1_V_R1]], [[Z1_V_R2]]
|
|
; CHECK-NEXT: ret double [[R]]
|
|
;
|
|
%X1 = fsub double %A1, %B1
|
|
%X2 = fsub double %A2, %B2
|
|
%Y1 = fmul double %X1, %A1
|
|
%Y2 = fmul double %X2, %A2
|
|
%Z1 = fadd double %Y1, %B1
|
|
%Z2 = fadd double %Y1, %B2
|
|
%R = fmul double %Z1, %Z2
|
|
ret double %R
|
|
}
|
|
|
|
; Basic depth-3 chain
|
|
define <2 x float> @test5(<2 x float> %A1, <2 x float> %A2, <2 x float> %B1, <2 x float> %B2) {
|
|
; CHECK-LABEL: @test5(
|
|
; CHECK-NEXT: [[X1_V_I1:%.*]] = shufflevector <2 x float> [[B1:%.*]], <2 x float> [[B2:%.*]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
|
|
; CHECK-NEXT: [[X1_V_I0:%.*]] = shufflevector <2 x float> [[A1:%.*]], <2 x float> [[A2:%.*]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
|
|
; CHECK-NEXT: [[X1:%.*]] = fsub <4 x float> [[X1_V_I0]], [[X1_V_I1]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = fmul <4 x float> [[X1]], [[X1_V_I0]]
|
|
; CHECK-NEXT: [[Z1:%.*]] = fadd <4 x float> [[Y1]], [[X1_V_I1]]
|
|
; CHECK-NEXT: [[Z1_V_R1:%.*]] = shufflevector <4 x float> [[Z1]], <4 x float> undef, <2 x i32> <i32 0, i32 1>
|
|
; CHECK-NEXT: [[Z1_V_R2:%.*]] = shufflevector <4 x float> [[Z1]], <4 x float> undef, <2 x i32> <i32 2, i32 3>
|
|
; CHECK-NEXT: [[R:%.*]] = fmul <2 x float> [[Z1_V_R1]], [[Z1_V_R2]]
|
|
; CHECK-NEXT: ret <2 x float> [[R]]
|
|
;
|
|
%X1 = fsub <2 x float> %A1, %B1
|
|
%X2 = fsub <2 x float> %A2, %B2
|
|
%Y1 = fmul <2 x float> %X1, %A1
|
|
%Y2 = fmul <2 x float> %X2, %A2
|
|
%Z1 = fadd <2 x float> %Y1, %B1
|
|
%Z2 = fadd <2 x float> %Y2, %B2
|
|
%R = fmul <2 x float> %Z1, %Z2
|
|
ret <2 x float> %R
|
|
}
|
|
|
|
; Basic chain with shuffles
|
|
define <8 x i8> @test6(<8 x i8> %A1, <8 x i8> %A2, <8 x i8> %B1, <8 x i8> %B2) {
|
|
; CHECK-LABEL: @test6(
|
|
; CHECK-NEXT: [[X1_V_I1:%.*]] = shufflevector <8 x i8> [[B1:%.*]], <8 x i8> [[B2:%.*]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
|
|
; CHECK-NEXT: [[X1_V_I0:%.*]] = shufflevector <8 x i8> [[A1:%.*]], <8 x i8> [[A2:%.*]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
|
|
; CHECK-NEXT: [[X1:%.*]] = sub <16 x i8> [[X1_V_I0]], [[X1_V_I1]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = mul <16 x i8> [[X1]], [[X1_V_I0]]
|
|
; CHECK-NEXT: [[Z1:%.*]] = add <16 x i8> [[Y1]], [[X1_V_I1]]
|
|
; CHECK-NEXT: [[Q1_V_I1:%.*]] = shufflevector <16 x i8> [[Z1]], <16 x i8> undef, <16 x i32> <i32 8, i32 undef, i32 10, i32 undef, i32 undef, i32 13, i32 undef, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
|
|
; CHECK-NEXT: [[Q1:%.*]] = shufflevector <16 x i8> [[Z1]], <16 x i8> [[Q1_V_I1]], <16 x i32> <i32 23, i32 16, i32 6, i32 1, i32 21, i32 18, i32 4, i32 3, i32 14, i32 15, i32 8, i32 9, i32 10, i32 12, i32 12, i32 9>
|
|
; CHECK-NEXT: [[Q1_V_R1:%.*]] = shufflevector <16 x i8> [[Q1]], <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
|
|
; CHECK-NEXT: [[Q1_V_R2:%.*]] = shufflevector <16 x i8> [[Q1]], <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
|
|
; CHECK-NEXT: [[R:%.*]] = mul <8 x i8> [[Q1_V_R1]], [[Q1_V_R2]]
|
|
; CHECK-NEXT: ret <8 x i8> [[R]]
|
|
;
|
|
%X1 = sub <8 x i8> %A1, %B1
|
|
%X2 = sub <8 x i8> %A2, %B2
|
|
%Y1 = mul <8 x i8> %X1, %A1
|
|
%Y2 = mul <8 x i8> %X2, %A2
|
|
%Z1 = add <8 x i8> %Y1, %B1
|
|
%Z2 = add <8 x i8> %Y2, %B2
|
|
%Q1 = shufflevector <8 x i8> %Z1, <8 x i8> %Z2, <8 x i32> <i32 15, i32 8, i32 6, i32 1, i32 13, i32 10, i32 4, i32 3>
|
|
%Q2 = shufflevector <8 x i8> %Z2, <8 x i8> %Z2, <8 x i32> <i32 6, i32 7, i32 0, i32 1, i32 2, i32 4, i32 4, i32 1>
|
|
%R = mul <8 x i8> %Q1, %Q2
|
|
ret <8 x i8> %R
|
|
}
|
|
|
|
; Basic depth-3 chain (flipped order)
|
|
define double @test7(double %A1, double %A2, double %B1, double %B2) {
|
|
; CHECK-LABEL: @test7(
|
|
; CHECK-NEXT: [[X1_V_I1_1:%.*]] = insertelement <2 x double> undef, double [[B1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I1_2:%.*]] = insertelement <2 x double> [[X1_V_I1_1]], double [[B2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1_V_I0_1:%.*]] = insertelement <2 x double> undef, double [[A1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I0_2:%.*]] = insertelement <2 x double> [[X1_V_I0_1]], double [[A2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1:%.*]] = fsub <2 x double> [[X1_V_I0_2]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = fmul <2 x double> [[X1]], [[X1_V_I0_2]]
|
|
; CHECK-NEXT: [[Z1:%.*]] = fadd <2 x double> [[Y1]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Z1_V_R1:%.*]] = extractelement <2 x double> [[Z1]], i32 0
|
|
; CHECK-NEXT: [[Z1_V_R2:%.*]] = extractelement <2 x double> [[Z1]], i32 1
|
|
; CHECK-NEXT: [[R:%.*]] = fmul double [[Z1_V_R1]], [[Z1_V_R2]]
|
|
; CHECK-NEXT: ret double [[R]]
|
|
;
|
|
%X1 = fsub double %A1, %B1
|
|
%X2 = fsub double %A2, %B2
|
|
%Y1 = fmul double %X1, %A1
|
|
%Y2 = fmul double %X2, %A2
|
|
%Z2 = fadd double %Y2, %B2
|
|
%Z1 = fadd double %Y1, %B1
|
|
%R = fmul double %Z1, %Z2
|
|
ret double %R
|
|
}
|
|
|
|
; Basic depth-3 chain (subclass data)
|
|
define i64 @test8(i64 %A1, i64 %A2, i64 %B1, i64 %B2) {
|
|
; CHECK-LABEL: @test8(
|
|
; CHECK-NEXT: [[X1_V_I1_1:%.*]] = insertelement <2 x i64> undef, i64 [[B1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I1_2:%.*]] = insertelement <2 x i64> [[X1_V_I1_1]], i64 [[B2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1_V_I0_1:%.*]] = insertelement <2 x i64> undef, i64 [[A1:%.*]], i32 0
|
|
; CHECK-NEXT: [[X1_V_I0_2:%.*]] = insertelement <2 x i64> [[X1_V_I0_1]], i64 [[A2:%.*]], i32 1
|
|
; CHECK-NEXT: [[X1:%.*]] = sub <2 x i64> [[X1_V_I0_2]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Y1:%.*]] = mul <2 x i64> [[X1]], [[X1_V_I0_2]]
|
|
; CHECK-NEXT: [[Z1:%.*]] = add <2 x i64> [[Y1]], [[X1_V_I1_2]]
|
|
; CHECK-NEXT: [[Z1_V_R1:%.*]] = extractelement <2 x i64> [[Z1]], i32 0
|
|
; CHECK-NEXT: [[Z1_V_R2:%.*]] = extractelement <2 x i64> [[Z1]], i32 1
|
|
; CHECK-NEXT: [[R:%.*]] = mul i64 [[Z1_V_R1]], [[Z1_V_R2]]
|
|
; CHECK-NEXT: ret i64 [[R]]
|
|
;
|
|
%X1 = sub nsw i64 %A1, %B1
|
|
%X2 = sub i64 %A2, %B2
|
|
%Y1 = mul i64 %X1, %A1
|
|
%Y2 = mul i64 %X2, %A2
|
|
%Z1 = add i64 %Y1, %B1
|
|
%Z2 = add i64 %Y2, %B2
|
|
%R = mul i64 %Z1, %Z2
|
|
ret i64 %R
|
|
}
|
|
|