diff options
author | Tanya Lattner <tonic@nondot.org> | 2008-02-19 08:07:33 +0000 |
---|---|---|
committer | Tanya Lattner <tonic@nondot.org> | 2008-02-19 08:07:33 +0000 |
commit | 5ebaf3ba1b5f8f05c01b1d7898a35273e83c1750 (patch) | |
tree | fb23c878019caf961242d6922d065dfae51fb7ca | |
parent | 660a728f833d167b82e04950e84270d8909c81f1 (diff) | |
download | external_llvm-5ebaf3ba1b5f8f05c01b1d7898a35273e83c1750.zip external_llvm-5ebaf3ba1b5f8f05c01b1d7898a35273e83c1750.tar.gz external_llvm-5ebaf3ba1b5f8f05c01b1d7898a35273e83c1750.tar.bz2 |
Remove llvm-upgrade and update tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@47325 91177308-0d34-0410-b5e6-96231b3b80d8
51 files changed, 1302 insertions, 1306 deletions
diff --git a/test/CodeGen/PowerPC/constants.ll b/test/CodeGen/PowerPC/constants.ll index 0c0d4ed..b58f59a 100644 --- a/test/CodeGen/PowerPC/constants.ll +++ b/test/CodeGen/PowerPC/constants.ll @@ -1,54 +1,52 @@ ; All of these routines should be perform optimal load of constants. -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep lis | count 5 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep ori | count 3 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep {li } | count 4 -implementation ; Functions: - -int %f1() { +define i32 @f1() { entry: - ret int 1 + ret i32 1 } -int %f2() { +define i32 @f2() { entry: - ret int -1 + ret i32 -1 } -int %f3() { +define i32 @f3() { entry: - ret int 0 + ret i32 0 } -int %f4() { +define i32 @f4() { entry: - ret int 32767 + ret i32 32767 } -int %f5() { +define i32 @f5() { entry: - ret int 65535 + ret i32 65535 } -int %f6() { +define i32 @f6() { entry: - ret int 65536 + ret i32 65536 } -int %f7() { +define i32 @f7() { entry: - ret int 131071 + ret i32 131071 } -int %f8() { +define i32 @f8() { entry: - ret int 2147483647 + ret i32 2147483647 } -int %f9() { +define i32 @f9() { entry: - ret int -2147483648 + ret i32 -2147483648 } diff --git a/test/CodeGen/PowerPC/cttz.ll b/test/CodeGen/PowerPC/cttz.ll index 3751d66..2c51e8a 100644 --- a/test/CodeGen/PowerPC/cttz.ll +++ b/test/CodeGen/PowerPC/cttz.ll @@ -1,12 +1,11 @@ ; Make sure this testcase does not use ctpop -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep -i cntlzw +; RUN: llvm-as < %s | llc -march=ppc32 | grep -i cntlzw -declare uint %llvm.cttz.i32(uint) +declare i32 @llvm.cttz.i32(i32) -implementation ; Functions: - -uint %bar(uint %x) { +define i32 @bar(i32 %x) { entry: - %tmp.1 = call uint %llvm.cttz.i32( uint %x ) - ret uint %tmp.1 + %tmp.1 = call i32 @llvm.cttz.i32( i32 %x ) ; <i32> [#uses=1] + ret i32 %tmp.1 } + diff --git a/test/CodeGen/PowerPC/darwin-labels.ll b/test/CodeGen/PowerPC/darwin-labels.ll index c8bf47c..ceebc70 100644 --- a/test/CodeGen/PowerPC/darwin-labels.ll +++ b/test/CodeGen/PowerPC/darwin-labels.ll @@ -1,8 +1,6 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc | grep {foo bar":} +; RUN: llvm-as < %s | llc | grep {foo bar":} -target endian = big -target pointersize = 32 +target datalayout = "E-p:32:32" target triple = "powerpc-apple-darwin8.2.0" - -"foo bar" = global int 4 +@"foo bar" = global i32 4 ; <i32*> [#uses=0] diff --git a/test/CodeGen/PowerPC/div-2.ll b/test/CodeGen/PowerPC/div-2.ll index a3cd73c..26e6221 100644 --- a/test/CodeGen/PowerPC/div-2.ll +++ b/test/CodeGen/PowerPC/div-2.ll @@ -1,29 +1,30 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep srawi -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep blr +; RUN: llvm-as < %s | llc -march=ppc32 | not grep srawi +; RUN: llvm-as < %s | llc -march=ppc32 | grep blr -int %test1(int %X) { - %Y = and int %X, 15 - %Z = div int %Y, 4 - ret int %Z +define i32 @test1(i32 %X) { + %Y = and i32 %X, 15 ; <i32> [#uses=1] + %Z = sdiv i32 %Y, 4 ; <i32> [#uses=1] + ret i32 %Z } -int %test2(int %W) { - %X = and int %W, 15 - %Y = sub int 16, %X - %Z = div int %Y, 4 - ret int %Z +define i32 @test2(i32 %W) { + %X = and i32 %W, 15 ; <i32> [#uses=1] + %Y = sub i32 16, %X ; <i32> [#uses=1] + %Z = sdiv i32 %Y, 4 ; <i32> [#uses=1] + ret i32 %Z } -int %test3(int %W) { - %X = and int %W, 15 - %Y = sub int 15, %X - %Z = div int %Y, 4 - ret int %Z +define i32 @test3(i32 %W) { + %X = and i32 %W, 15 ; <i32> [#uses=1] + %Y = sub i32 15, %X ; <i32> [#uses=1] + %Z = sdiv i32 %Y, 4 ; <i32> [#uses=1] + ret i32 %Z } -int %test4(int %W) { - %X = and int %W, 2 - %Y = sub int 5, %X - %Z = div int %Y, 2 - ret int %Z +define i32 @test4(i32 %W) { + %X = and i32 %W, 2 ; <i32> [#uses=1] + %Y = sub i32 5, %X ; <i32> [#uses=1] + %Z = sdiv i32 %Y, 2 ; <i32> [#uses=1] + ret i32 %Z } + diff --git a/test/CodeGen/PowerPC/eqv-andc-orc-nor.ll b/test/CodeGen/PowerPC/eqv-andc-orc-nor.ll index 30bd0b4..7be8a34 100644 --- a/test/CodeGen/PowerPC/eqv-andc-orc-nor.ll +++ b/test/CodeGen/PowerPC/eqv-andc-orc-nor.ll @@ -1,94 +1,93 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep eqv | count 3 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | \ ; RUN: grep andc | count 3 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep orc | count 2 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | \ ; RUN: grep nor | count 3 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep nand | count 1 -int %EQV1(int %X, int %Y) { - %A = xor int %X, %Y - %B = xor int %A, -1 - ret int %B +define i32 @EQV1(i32 %X, i32 %Y) { + %A = xor i32 %X, %Y ; <i32> [#uses=1] + %B = xor i32 %A, -1 ; <i32> [#uses=1] + ret i32 %B } -int %EQV2(int %X, int %Y) { - %A = xor int %X, -1 - %B = xor int %A, %Y - ret int %B +define i32 @EQV2(i32 %X, i32 %Y) { + %A = xor i32 %X, -1 ; <i32> [#uses=1] + %B = xor i32 %A, %Y ; <i32> [#uses=1] + ret i32 %B } -int %EQV3(int %X, int %Y) { - %A = xor int %X, -1 - %B = xor int %Y, %A - ret int %B +define i32 @EQV3(i32 %X, i32 %Y) { + %A = xor i32 %X, -1 ; <i32> [#uses=1] + %B = xor i32 %Y, %A ; <i32> [#uses=1] + ret i32 %B } -int %ANDC1(int %X, int %Y) { - %A = xor int %Y, -1 - %B = and int %X, %A - ret int %B +define i32 @ANDC1(i32 %X, i32 %Y) { + %A = xor i32 %Y, -1 ; <i32> [#uses=1] + %B = and i32 %X, %A ; <i32> [#uses=1] + ret i32 %B } -int %ANDC2(int %X, int %Y) { - %A = xor int %X, -1 - %B = and int %A, %Y - ret int %B +define i32 @ANDC2(i32 %X, i32 %Y) { + %A = xor i32 %X, -1 ; <i32> [#uses=1] + %B = and i32 %A, %Y ; <i32> [#uses=1] + ret i32 %B } -int %ORC1(int %X, int %Y) { - %A = xor int %Y, -1 - %B = or int %X, %A - ret int %B +define i32 @ORC1(i32 %X, i32 %Y) { + %A = xor i32 %Y, -1 ; <i32> [#uses=1] + %B = or i32 %X, %A ; <i32> [#uses=1] + ret i32 %B } -int %ORC2(int %X, int %Y) { - %A = xor int %X, -1 - %B = or int %A, %Y - ret int %B +define i32 @ORC2(i32 %X, i32 %Y) { + %A = xor i32 %X, -1 ; <i32> [#uses=1] + %B = or i32 %A, %Y ; <i32> [#uses=1] + ret i32 %B } -int %NOR1(int %X) { - %Y = xor int %X, -1 - ret int %Y +define i32 @NOR1(i32 %X) { + %Y = xor i32 %X, -1 ; <i32> [#uses=1] + ret i32 %Y } -int %NOR2(int %X, int %Y) { - %Z = or int %X, %Y - %R = xor int %Z, -1 - ret int %R +define i32 @NOR2(i32 %X, i32 %Y) { + %Z = or i32 %X, %Y ; <i32> [#uses=1] + %R = xor i32 %Z, -1 ; <i32> [#uses=1] + ret i32 %R } -int %NAND1(int %X, int %Y) { - %Z = and int %X, %Y - %W = xor int %Z, -1 - ret int %W +define i32 @NAND1(i32 %X, i32 %Y) { + %Z = and i32 %X, %Y ; <i32> [#uses=1] + %W = xor i32 %Z, -1 ; <i32> [#uses=1] + ret i32 %W } -void %VNOR(<4 x float>* %P, <4 x float>* %Q) { - %tmp = load <4 x float>* %P - %tmp = cast <4 x float> %tmp to <4 x int> - %tmp2 = load <4 x float>* %Q - %tmp2 = cast <4 x float> %tmp2 to <4 x int> - %tmp3 = or <4 x int> %tmp, %tmp2 - %tmp4 = xor <4 x int> %tmp3, < int -1, int -1, int -1, int -1 > - %tmp4 = cast <4 x int> %tmp4 to <4 x float> - store <4 x float> %tmp4, <4 x float>* %P - ret void +define void @VNOR(<4 x float>* %P, <4 x float>* %Q) { + %tmp = load <4 x float>* %P ; <<4 x float>> [#uses=1] + %tmp.upgrd.1 = bitcast <4 x float> %tmp to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp2 = load <4 x float>* %Q ; <<4 x float>> [#uses=1] + %tmp2.upgrd.2 = bitcast <4 x float> %tmp2 to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp3 = or <4 x i32> %tmp.upgrd.1, %tmp2.upgrd.2 ; <<4 x i32>> [#uses=1] + %tmp4 = xor <4 x i32> %tmp3, < i32 -1, i32 -1, i32 -1, i32 -1 > ; <<4 x i32>> [#uses=1] + %tmp4.upgrd.3 = bitcast <4 x i32> %tmp4 to <4 x float> ; <<4 x float>> [#uses=1] + store <4 x float> %tmp4.upgrd.3, <4 x float>* %P + ret void } -void %VANDC(<4 x float>* %P, <4 x float>* %Q) { - %tmp = load <4 x float>* %P - %tmp = cast <4 x float> %tmp to <4 x int> - %tmp2 = load <4 x float>* %Q - %tmp2 = cast <4 x float> %tmp2 to <4 x int> - %tmp4 = xor <4 x int> %tmp2, < int -1, int -1, int -1, int -1 > - %tmp3 = and <4 x int> %tmp, %tmp4 - %tmp4 = cast <4 x int> %tmp3 to <4 x float> - store <4 x float> %tmp4, <4 x float>* %P - ret void +define void @VANDC(<4 x float>* %P, <4 x float>* %Q) { + %tmp = load <4 x float>* %P ; <<4 x float>> [#uses=1] + %tmp.upgrd.4 = bitcast <4 x float> %tmp to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp2 = load <4 x float>* %Q ; <<4 x float>> [#uses=1] + %tmp2.upgrd.5 = bitcast <4 x float> %tmp2 to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp4 = xor <4 x i32> %tmp2.upgrd.5, < i32 -1, i32 -1, i32 -1, i32 -1 > ; <<4 x i32>> [#uses=1] + %tmp3 = and <4 x i32> %tmp.upgrd.4, %tmp4 ; <<4 x i32>> [#uses=1] + %tmp4.upgrd.6 = bitcast <4 x i32> %tmp3 to <4 x float> ; <<4 x float>> [#uses=1] + store <4 x float> %tmp4.upgrd.6, <4 x float>* %P + ret void } - diff --git a/test/CodeGen/PowerPC/extsh.ll b/test/CodeGen/PowerPC/extsh.ll index 6b5023f..5eca8ce 100644 --- a/test/CodeGen/PowerPC/extsh.ll +++ b/test/CodeGen/PowerPC/extsh.ll @@ -1,7 +1,8 @@ ; This should turn into a single extsh -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep extsh | count 1 -int %test(int %X) { - %tmp.81 = shl int %X, ubyte 16 ; <int> [#uses=1] - %tmp.82 = shr int %tmp.81, ubyte 16 ; <int> [#uses=1] - ret int %tmp.82 +; RUN: llvm-as < %s | llc -march=ppc32 | grep extsh | count 1 +define i32 @test(i32 %X) { + %tmp.81 = shl i32 %X, 16 ; <i32> [#uses=1] + %tmp.82 = ashr i32 %tmp.81, 16 ; <i32> [#uses=1] + ret i32 %tmp.82 } + diff --git a/test/CodeGen/PowerPC/fma.ll b/test/CodeGen/PowerPC/fma.ll index da571f6..fd9bd74 100644 --- a/test/CodeGen/PowerPC/fma.ll +++ b/test/CodeGen/PowerPC/fma.ll @@ -1,47 +1,54 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: egrep {fn?madd|fn?msub} | count 8 -double %test_FMADD1(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = add double %D, %C +define double @test_FMADD1(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = add double %D, %C ; <double> [#uses=1] ret double %E } -double %test_FMADD2(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = add double %D, %C + +define double @test_FMADD2(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = add double %D, %C ; <double> [#uses=1] ret double %E } -double %test_FMSUB(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = sub double %D, %C + +define double @test_FMSUB(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = sub double %D, %C ; <double> [#uses=1] ret double %E } -double %test_FNMADD1(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = add double %D, %C - %F = sub double -0.0, %E + +define double @test_FNMADD1(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = add double %D, %C ; <double> [#uses=1] + %F = sub double -0.000000e+00, %E ; <double> [#uses=1] ret double %F } -double %test_FNMADD2(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = add double %C, %D - %F = sub double -0.0, %E + +define double @test_FNMADD2(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = add double %C, %D ; <double> [#uses=1] + %F = sub double -0.000000e+00, %E ; <double> [#uses=1] ret double %F } -double %test_FNMSUB1(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = sub double %C, %D + +define double @test_FNMSUB1(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = sub double %C, %D ; <double> [#uses=1] ret double %E } -double %test_FNMSUB2(double %A, double %B, double %C) { - %D = mul double %A, %B - %E = sub double %D, %C - %F = sub double -0.0, %E + +define double @test_FNMSUB2(double %A, double %B, double %C) { + %D = mul double %A, %B ; <double> [#uses=1] + %E = sub double %D, %C ; <double> [#uses=1] + %F = sub double -0.000000e+00, %E ; <double> [#uses=1] ret double %F } -float %test_FNMSUBS(float %A, float %B, float %C) { - %D = mul float %A, %B - %E = sub float %D, %C - %F = sub float -0.0, %E + +define float @test_FNMSUBS(float %A, float %B, float %C) { + %D = mul float %A, %B ; <float> [#uses=1] + %E = sub float %D, %C ; <float> [#uses=1] + %F = sub float -0.000000e+00, %E ; <float> [#uses=1] ret float %F } diff --git a/test/CodeGen/PowerPC/fnabs.ll b/test/CodeGen/PowerPC/fnabs.ll index 5d0ef5f..b9517de 100644 --- a/test/CodeGen/PowerPC/fnabs.ll +++ b/test/CodeGen/PowerPC/fnabs.ll @@ -1,11 +1,10 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep fnabs +; RUN: llvm-as < %s | llc -march=ppc32 | grep fnabs -declare double %fabs(double) +declare double @fabs(double) -implementation - -double %test(double %X) { - %Y = call double %fabs(double %X) - %Z = sub double -0.0, %Y - ret double %Z +define double @test(double %X) { + %Y = call double @fabs( double %X ) ; <double> [#uses=1] + %Z = sub double -0.000000e+00, %Y ; <double> [#uses=1] + ret double %Z } + diff --git a/test/CodeGen/PowerPC/fnegsel.ll b/test/CodeGen/PowerPC/fnegsel.ll index b1b0645..c223a49 100644 --- a/test/CodeGen/PowerPC/fnegsel.ll +++ b/test/CodeGen/PowerPC/fnegsel.ll @@ -1,8 +1,9 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep fneg +; RUN: llvm-as < %s | llc -march=ppc32 | not grep fneg -double %test_FNEG_sel(double %A, double %B, double %C) { - %D = sub double -0.0, %A - %Cond = setgt double %D, -0.0 - %E = select bool %Cond, double %B, double %C - ret double %E +define double @test_FNEG_sel(double %A, double %B, double %C) { + %D = sub double -0.000000e+00, %A ; <double> [#uses=1] + %Cond = fcmp ogt double %D, -0.000000e+00 ; <i1> [#uses=1] + %E = select i1 %Cond, double %B, double %C ; <double> [#uses=1] + ret double %E } + diff --git a/test/CodeGen/PowerPC/fold-li.ll b/test/CodeGen/PowerPC/fold-li.ll index 66a900f..e96bc45 100644 --- a/test/CodeGen/PowerPC/fold-li.ll +++ b/test/CodeGen/PowerPC/fold-li.ll @@ -1,14 +1,15 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | \ +; RUN: llvm-as < %s | llc -march=ppc32 | \ ; RUN: grep -v align | not grep li ;; Test that immediates are folded into these instructions correctly. -int %ADD(int %X) { - %Y = add int %X, 65537 - ret int %Y +define i32 @ADD(i32 %X) { + %Y = add i32 %X, 65537 ; <i32> [#uses=1] + ret i32 %Y } -int %SUB(int %X) { - %Y = sub int %X, 65537 - ret int %Y +define i32 @SUB(i32 %X) { + %Y = sub i32 %X, 65537 ; <i32> [#uses=1] + ret i32 %Y } + diff --git a/test/CodeGen/PowerPC/fp-branch.ll b/test/CodeGen/PowerPC/fp-branch.ll index 804cbf6..3db6ced 100644 --- a/test/CodeGen/PowerPC/fp-branch.ll +++ b/test/CodeGen/PowerPC/fp-branch.ll @@ -1,20 +1,21 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep fcmp | count 1 +; RUN: llvm-as < %s | llc -march=ppc32 | grep fcmp | count 1 -declare bool %llvm.isunordered.f64(double, double) +declare i1 @llvm.isunordered.f64(double, double) -bool %intcoord_cond_next55(double %tmp48.reload) { +define i1 @intcoord_cond_next55(double %tmp48.reload) { newFuncRoot: - br label %cond_next55 + br label %cond_next55 -bb72.exitStub: ; preds = %cond_next55 - ret bool true +bb72.exitStub: ; preds = %cond_next55 + ret i1 true -cond_next62.exitStub: ; preds = %cond_next55 - ret bool false +cond_next62.exitStub: ; preds = %cond_next55 + ret i1 false -cond_next55: ; preds = %newFuncRoot - %tmp57 = setge double %tmp48.reload, 1.000000e+00 ; <bool> [#uses=1] - %tmp58 = tail call bool %llvm.isunordered.f64( double %tmp48.reload, double 1.000000e+00 ) ; <bool> [#uses=1] - %tmp59 = or bool %tmp57, %tmp58 ; <bool> [#uses=1] - br bool %tmp59, label %bb72.exitStub, label %cond_next62.exitStub +cond_next55: ; preds = %newFuncRoot + %tmp57 = fcmp oge double %tmp48.reload, 1.000000e+00 ; <i1> [#uses=1] + %tmp58 = fcmp uno double %tmp48.reload, 1.000000e+00 ; <i1> [#uses=1] + %tmp59 = or i1 %tmp57, %tmp58 ; <i1> [#uses=1] + br i1 %tmp59, label %bb72.exitStub, label %cond_next62.exitStub } + diff --git a/test/CodeGen/PowerPC/fp-int-fp.ll b/test/CodeGen/PowerPC/fp-int-fp.ll index 63ebc49..1b78b01 100644 --- a/test/CodeGen/PowerPC/fp-int-fp.ll +++ b/test/CodeGen/PowerPC/fp-int-fp.ll @@ -1,26 +1,27 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | not grep r1 +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | not grep r1 -double %test1(double %X) { - %Y = cast double %X to long - %Z = cast long %Y to double +define double @test1(double %X) { + %Y = fptosi double %X to i64 ; <i64> [#uses=1] + %Z = sitofp i64 %Y to double ; <double> [#uses=1] ret double %Z } -float %test2(double %X) { - %Y = cast double %X to long - %Z = cast long %Y to float +define float @test2(double %X) { + %Y = fptosi double %X to i64 ; <i64> [#uses=1] + %Z = sitofp i64 %Y to float ; <float> [#uses=1] ret float %Z } -double %test3(float %X) { - %Y = cast float %X to long - %Z = cast long %Y to double +define double @test3(float %X) { + %Y = fptosi float %X to i64 ; <i64> [#uses=1] + %Z = sitofp i64 %Y to double ; <double> [#uses=1] ret double %Z } -float %test4(float %X) { - %Y = cast float %X to long - %Z = cast long %Y to float +define float @test4(float %X) { + %Y = fptosi float %X to i64 ; <i64> [#uses=1] + %Z = sitofp i64 %Y to float ; <float> [#uses=1] ret float %Z } + diff --git a/test/CodeGen/PowerPC/fp_to_uint.ll b/test/CodeGen/PowerPC/fp_to_uint.ll index 4a76617..43502bb 100644 --- a/test/CodeGen/PowerPC/fp_to_uint.ll +++ b/test/CodeGen/PowerPC/fp_to_uint.ll @@ -1,9 +1,8 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep fctiwz | count 1 +; RUN: llvm-as < %s | llc -march=ppc32 | grep fctiwz | count 1 -implementation - -ushort %foo(float %a) { +define i16 @foo(float %a) { entry: - %tmp.1 = cast float %a to ushort - ret ushort %tmp.1 + %tmp.1 = fptoui float %a to i16 ; <i16> [#uses=1] + ret i16 %tmp.1 } + diff --git a/test/CodeGen/PowerPC/fpcopy.ll b/test/CodeGen/PowerPC/fpcopy.ll index ce86da8..7d80596 100644 --- a/test/CodeGen/PowerPC/fpcopy.ll +++ b/test/CodeGen/PowerPC/fpcopy.ll @@ -1,6 +1,7 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep fmr +; RUN: llvm-as < %s | llc -march=ppc32 | not grep fmr -double %test(float %F) { - %F = cast float %F to double - ret double %F +define double @test(float %F) { + %F.upgrd.1 = fpext float %F to double ; <double> [#uses=1] + ret double %F.upgrd.1 } + diff --git a/test/CodeGen/PowerPC/fsqrt.ll b/test/CodeGen/PowerPC/fsqrt.ll index 809077b..1260c60 100644 --- a/test/CodeGen/PowerPC/fsqrt.ll +++ b/test/CodeGen/PowerPC/fsqrt.ll @@ -1,21 +1,23 @@ ; fsqrt should be generated when the fsqrt feature is enabled, but not ; otherwise. -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 -mattr=+fsqrt | \ ; RUN: grep {fsqrt f1, f1} -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 -mcpu=g5 | \ ; RUN: grep {fsqrt f1, f1} -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 -mattr=-fsqrt | \ ; RUN: not grep {fsqrt f1, f1} -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 -mcpu=g4 | \ ; RUN: not grep {fsqrt f1, f1} -declare double %llvm.sqrt.f64(double) -double %X(double %Y) { - %Z = call double %llvm.sqrt.f64(double %Y) - ret double %Z +declare double @llvm.sqrt.f64(double) + +define double @X(double %Y) { + %Z = call double @llvm.sqrt.f64( double %Y ) ; <double> [#uses=1] + ret double %Z } + diff --git a/test/CodeGen/PowerPC/i64_fp.ll b/test/CodeGen/PowerPC/i64_fp.ll index 8720327..5ff2684 100644 --- a/test/CodeGen/PowerPC/i64_fp.ll +++ b/test/CodeGen/PowerPC/i64_fp.ll @@ -1,25 +1,26 @@ ; fcfid and fctid should be generated when the 64bit feature is enabled, but not ; otherwise. -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mattr=+64bit | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mattr=+64bit | \ ; RUN: grep fcfid -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mattr=+64bit | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mattr=+64bit | \ ; RUN: grep fctidz -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | \ ; RUN: grep fcfid -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | \ ; RUN: grep fctidz -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mattr=-64bit | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mattr=-64bit | \ ; RUN: not grep fcfid -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mattr=-64bit | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mattr=-64bit | \ ; RUN: not grep fctidz -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g4 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g4 | \ ; RUN: not grep fcfid -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g4 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g4 | \ ; RUN: not grep fctidz -double %X(double %Y) { - %A = cast double %Y to long - %B = cast long %A to double - ret double %B +define double @X(double %Y) { + %A = fptosi double %Y to i64 ; <i64> [#uses=1] + %B = sitofp i64 %A to double ; <double> [#uses=1] + ret double %B } + diff --git a/test/CodeGen/PowerPC/inlineasm-copy.ll b/test/CodeGen/PowerPC/inlineasm-copy.ll index 34594d2..c0a3979 100644 --- a/test/CodeGen/PowerPC/inlineasm-copy.ll +++ b/test/CodeGen/PowerPC/inlineasm-copy.ll @@ -1,13 +1,14 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep mr +; RUN: llvm-as < %s | llc -march=ppc32 | not grep mr -int %test(int %Y, int %X) { +define i32 @test(i32 %Y, i32 %X) { entry: - %tmp = tail call int asm "foo $0", "=r"( ) ; <int> [#uses=1] - ret int %tmp + %tmp = tail call i32 asm "foo $0", "=r"( ) ; <i32> [#uses=1] + ret i32 %tmp } -int %test2(int %Y, int %X) { +define i32 @test2(i32 %Y, i32 %X) { entry: - %tmp1 = tail call int asm "foo $0, $1", "=r,r"( int %X ) ; <int> [#uses=1] - ret int %tmp1 + %tmp1 = tail call i32 asm "foo $0, $1", "=r,r"( i32 %X ) ; <i32> [#uses=1] + ret i32 %tmp1 } + diff --git a/test/CodeGen/PowerPC/inverted-bool-compares.ll b/test/CodeGen/PowerPC/inverted-bool-compares.ll index fbbf6a5..f8c5f11 100644 --- a/test/CodeGen/PowerPC/inverted-bool-compares.ll +++ b/test/CodeGen/PowerPC/inverted-bool-compares.ll @@ -1,10 +1,13 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep xori - -int %test(bool %B, int* %P) { - br bool %B, label %T, label %F -T: - store int 123, int* %P - ret int 0 -F: -ret int 17 +; RUN: llvm-as < %s | llc -march=ppc32 | not grep xori + +define i32 @test(i1 %B, i32* %P) { + br i1 %B, label %T, label %F + +T: ; preds = %0 + store i32 123, i32* %P + ret i32 0 + +F: ; preds = %0 + ret i32 17 } + diff --git a/test/CodeGen/PowerPC/lha.ll b/test/CodeGen/PowerPC/lha.ll index cc35e8a..e8f73ee 100644 --- a/test/CodeGen/PowerPC/lha.ll +++ b/test/CodeGen/PowerPC/lha.ll @@ -1,7 +1,8 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep lha +; RUN: llvm-as < %s | llc -march=ppc32 | grep lha -uint %test(short* %a) { - %tmp.1 = load short* %a - %tmp.2 = cast short %tmp.1 to uint - ret uint %tmp.2 +define i32 @test(i16* %a) { + %tmp.1 = load i16* %a ; <i16> [#uses=1] + %tmp.2 = sext i16 %tmp.1 to i32 ; <i32> [#uses=1] + ret i32 %tmp.2 } + diff --git a/test/CodeGen/PowerPC/load-constant-addr.ll b/test/CodeGen/PowerPC/load-constant-addr.ll index 65ec782..d2be04e 100644 --- a/test/CodeGen/PowerPC/load-constant-addr.ll +++ b/test/CodeGen/PowerPC/load-constant-addr.ll @@ -1,9 +1,9 @@ ; Should fold the ori into the lfs. -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep lfs -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep ori +; RUN: llvm-as < %s | llc -march=ppc32 | grep lfs +; RUN: llvm-as < %s | llc -march=ppc32 | not grep ori -float %test() { - %tmp.i = load float* cast (uint 186018016 to float*) - ret float %tmp.i +define float @test() { + %tmp.i = load float* inttoptr (i32 186018016 to float*) ; <float> [#uses=1] + ret float %tmp.i } diff --git a/test/CodeGen/PowerPC/mem-rr-addr-mode.ll b/test/CodeGen/PowerPC/mem-rr-addr-mode.ll index 8aa7aa2..d5484bd 100644 --- a/test/CodeGen/PowerPC/mem-rr-addr-mode.ll +++ b/test/CodeGen/PowerPC/mem-rr-addr-mode.ll @@ -1,17 +1,18 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | grep li.*16 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | not grep addi +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | grep li.*16 +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | not grep addi ; Codegen lvx (R+16) as t = li 16, lvx t,R ; This shares the 16 between the two loads. -void %func(<4 x float>* %a, <4 x float>* %b) { - %tmp1 = getelementptr <4 x float>* %b, int 1 - %tmp = load <4 x float>* %tmp1 - %tmp3 = getelementptr <4 x float>* %a, int 1 - %tmp4 = load <4 x float>* %tmp3 - %tmp5 = mul <4 x float> %tmp, %tmp4 - %tmp8 = load <4 x float>* %b - %tmp9 = add <4 x float> %tmp5, %tmp8 - store <4 x float> %tmp9, <4 x float>* %a - ret void +define void @func(<4 x float>* %a, <4 x float>* %b) { + %tmp1 = getelementptr <4 x float>* %b, i32 1 ; <<4 x float>*> [#uses=1] + %tmp = load <4 x float>* %tmp1 ; <<4 x float>> [#uses=1] + %tmp3 = getelementptr <4 x float>* %a, i32 1 ; <<4 x float>*> [#uses=1] + %tmp4 = load <4 x float>* %tmp3 ; <<4 x float>> [#uses=1] + %tmp5 = mul <4 x float> %tmp, %tmp4 ; <<4 x float>> [#uses=1] + %tmp8 = load <4 x float>* %b ; <<4 x float>> [#uses=1] + %tmp9 = add <4 x float> %tmp5, %tmp8 ; <<4 x float>> [#uses=1] + store <4 x float> %tmp9, <4 x float>* %a + ret void } + diff --git a/test/CodeGen/PowerPC/mem_update.ll b/test/CodeGen/PowerPC/mem_update.ll index 4d3ebe9..a152762 100644 --- a/test/CodeGen/PowerPC/mem_update.ll +++ b/test/CodeGen/PowerPC/mem_update.ll @@ -1,68 +1,68 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -enable-ppc-preinc | \ +; RUN: llvm-as < %s | llc -march=ppc32 -enable-ppc-preinc | \ ; RUN: not grep addi -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc64 -enable-ppc-preinc | \ +; RUN: llvm-as < %s | llc -march=ppc64 -enable-ppc-preinc | \ ; RUN: not grep addi -%Glob = global ulong 4 -int *%test0(int *%X, int *%dest) { - %Y = getelementptr int* %X, int 4 - %A = load int* %Y - store int %A, int* %dest - ret int* %Y -} +@Glob = global i64 4 ; <i64*> [#uses=2] -int *%test1(int *%X, int *%dest) { - %Y = getelementptr int* %X, int 4 - %A = load int* %Y - store int %A, int* %dest - ret int* %Y +define i32* @test0(i32* %X, i32* %dest) { + %Y = getelementptr i32* %X, i32 4 ; <i32*> [#uses=2] + %A = load i32* %Y ; <i32> [#uses=1] + store i32 %A, i32* %dest + ret i32* %Y } -short *%test2(short *%X, int *%dest) { - %Y = getelementptr short* %X, int 4 - %A = load short* %Y - %B = cast short %A to int - store int %B, int* %dest - ret short* %Y +define i32* @test1(i32* %X, i32* %dest) { + %Y = getelementptr i32* %X, i32 4 ; <i32*> [#uses=2] + %A = load i32* %Y ; <i32> [#uses=1] + store i32 %A, i32* %dest + ret i32* %Y } -ushort *%test3(ushort *%X, int *%dest) { - %Y = getelementptr ushort* %X, int 4 - %A = load ushort* %Y - %B = cast ushort %A to int - store int %B, int* %dest - ret ushort* %Y +define i16* @test2(i16* %X, i32* %dest) { + %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] + %A = load i16* %Y ; <i16> [#uses=1] + %B = sext i16 %A to i32 ; <i32> [#uses=1] + store i32 %B, i32* %dest + ret i16* %Y } -short *%test3a(short *%X, long *%dest) { - %Y = getelementptr short* %X, int 4 - %A = load short* %Y - %B = cast short %A to long - store long %B, long* %dest - ret short* %Y +define i16* @test3(i16* %X, i32* %dest) { + %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] + %A = load i16* %Y ; <i16> [#uses=1] + %B = zext i16 %A to i32 ; <i32> [#uses=1] + store i32 %B, i32* %dest + ret i16* %Y } -long *%test4(long *%X, long *%dest) { - %Y = getelementptr long* %X, int 4 - %A = load long* %Y - store long %A, long* %dest - ret long* %Y +define i16* @test3a(i16* %X, i64* %dest) { + %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] + %A = load i16* %Y ; <i16> [#uses=1] + %B = sext i16 %A to i64 ; <i64> [#uses=1] + store i64 %B, i64* %dest + ret i16* %Y } -ushort *%test5(ushort *%X) { - %Y = getelementptr ushort* %X, int 4 - store ushort 7, ushort* %Y - ret ushort* %Y +define i64* @test4(i64* %X, i64* %dest) { + %Y = getelementptr i64* %X, i32 4 ; <i64*> [#uses=2] + %A = load i64* %Y ; <i64> [#uses=1] + store i64 %A, i64* %dest + ret i64* %Y } -ulong *%test6(ulong *%X, ulong %A) { - %Y = getelementptr ulong* %X, int 4 - store ulong %A, ulong* %Y - ret ulong* %Y +define i16* @test5(i16* %X) { + %Y = getelementptr i16* %X, i32 4 ; <i16*> [#uses=2] + store i16 7, i16* %Y + ret i16* %Y } -ulong *%test7(ulong *%X, ulong %A) { - store ulong %A, ulong* %Glob - ret ulong *%Glob +define i64* @test6(i64* %X, i64 %A) { + %Y = getelementptr i64* %X, i32 4 ; <i64*> [#uses=2] + store i64 %A, i64* %Y + ret i64* %Y } +define i64* @test7(i64* %X, i64 %A) { + store i64 %A, i64* @Glob + ret i64* @Glob +} diff --git a/test/CodeGen/PowerPC/mul-neg-power-2.ll b/test/CodeGen/PowerPC/mul-neg-power-2.ll index b9be1cc..90446d7 100644 --- a/test/CodeGen/PowerPC/mul-neg-power-2.ll +++ b/test/CodeGen/PowerPC/mul-neg-power-2.ll @@ -1,8 +1,8 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep mul +; RUN: llvm-as < %s | llc -march=ppc32 | not grep mul -int %test1(int %a) { - %tmp.1 = mul int %a, -2 ; <int> [#uses=1] - %tmp.2 = add int %tmp.1, 63 ; <int> [#uses=1] - ret int %tmp.2 +define i32 @test1(i32 %a) { + %tmp.1 = mul i32 %a, -2 ; <i32> [#uses=1] + %tmp.2 = add i32 %tmp.1, 63 ; <i32> [#uses=1] + ret i32 %tmp.2 } diff --git a/test/CodeGen/PowerPC/mulhs.ll b/test/CodeGen/PowerPC/mulhs.ll index 7661e9c..3b0daad 100644 --- a/test/CodeGen/PowerPC/mulhs.ll +++ b/test/CodeGen/PowerPC/mulhs.ll @@ -1,18 +1,17 @@ ; All of these ands and shifts should be folded into rlwimi's -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -o %t -f +; RUN: llvm-as < %s | llc -march=ppc32 -o %t -f ; RUN: not grep mulhwu %t ; RUN: not grep srawi %t ; RUN: not grep add %t ; RUN: grep mulhw %t | count 1 -implementation ; Functions: - -int %mulhs(int %a, int %b) { +define i32 @mulhs(i32 %a, i32 %b) { entry: - %tmp.1 = cast int %a to ulong ; <ulong> [#uses=1] - %tmp.3 = cast int %b to ulong ; <ulong> [#uses=1] - %tmp.4 = mul ulong %tmp.3, %tmp.1 ; <ulong> [#uses=1] - %tmp.6 = shr ulong %tmp.4, ubyte 32 ; <ulong> [#uses=1] - %tmp.7 = cast ulong %tmp.6 to int ; <int> [#uses=1] - ret int %tmp.7 + %tmp.1 = sext i32 %a to i64 ; <i64> [#uses=1] + %tmp.3 = sext i32 %b to i64 ; <i64> [#uses=1] + %tmp.4 = mul i64 %tmp.3, %tmp.1 ; <i64> [#uses=1] + %tmp.6 = lshr i64 %tmp.4, 32 ; <i64> [#uses=1] + %tmp.7 = trunc i64 %tmp.6 to i32 ; <i32> [#uses=1] + ret i32 %tmp.7 } + diff --git a/test/CodeGen/PowerPC/neg.ll b/test/CodeGen/PowerPC/neg.ll index 7119f6c..c135599 100644 --- a/test/CodeGen/PowerPC/neg.ll +++ b/test/CodeGen/PowerPC/neg.ll @@ -1,6 +1,7 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep neg +; RUN: llvm-as < %s | llc -march=ppc32 | grep neg -int %test(int %X) { - %Y = sub int 0, %X - ret int %Y +define i32 @test(i32 %X) { + %Y = sub i32 0, %X ; <i32> [#uses=1] + ret i32 %Y } + diff --git a/test/CodeGen/PowerPC/or-addressing-mode.ll b/test/CodeGen/PowerPC/or-addressing-mode.ll index e448140..9b6e955 100644 --- a/test/CodeGen/PowerPC/or-addressing-mode.ll +++ b/test/CodeGen/PowerPC/or-addressing-mode.ll @@ -1,22 +1,22 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -mtriple=powerpc-apple-darwin8 | not grep ori -; RUN: llvm-upgrade < %s | llvm-as | llc -mtriple=powerpc-apple-darwin8 | not grep rlwimi +; RUN: llvm-as < %s | llc -mtriple=powerpc-apple-darwin8 | not grep ori +; RUN: llvm-as < %s | llc -mtriple=powerpc-apple-darwin8 | not grep rlwimi -int %test1(sbyte* %P) { ;; or -> lwzx - %tmp.2.i = cast sbyte* %P to uint - %tmp.4.i = and uint %tmp.2.i, 4294901760 - %tmp.10.i = shr uint %tmp.2.i, ubyte 5 - %tmp.11.i = and uint %tmp.10.i, 2040 - %tmp.13.i = or uint %tmp.11.i, %tmp.4.i - %tmp.14.i = cast uint %tmp.13.i to int* - %tmp.3 = load int* %tmp.14.i - ret int %tmp.3 +define i32 @test1(i8* %P) { + %tmp.2.i = ptrtoint i8* %P to i32 ; <i32> [#uses=2] + %tmp.4.i = and i32 %tmp.2.i, -65536 ; <i32> [#uses=1] + %tmp.10.i = lshr i32 %tmp.2.i, 5 ; <i32> [#uses=1] + %tmp.11.i = and i32 %tmp.10.i, 2040 ; <i32> [#uses=1] + %tmp.13.i = or i32 %tmp.11.i, %tmp.4.i ; <i32> [#uses=1] + %tmp.14.i = inttoptr i32 %tmp.13.i to i32* ; <i32*> [#uses=1] + %tmp.3 = load i32* %tmp.14.i ; <i32> [#uses=1] + ret i32 %tmp.3 } -int %test2(int %P) { ;; or -> lwz - %tmp.2 = shl int %P, ubyte 4 - %tmp.3 = or int %tmp.2, 2 - %tmp.4 = cast int %tmp.3 to int* - %tmp.5 = load int* %tmp.4 - ret int %tmp.5 +define i32 @test2(i32 %P) { + %tmp.2 = shl i32 %P, 4 ; <i32> [#uses=1] + %tmp.3 = or i32 %tmp.2, 2 ; <i32> [#uses=1] + %tmp.4 = inttoptr i32 %tmp.3 to i32* ; <i32*> [#uses=1] + %tmp.5 = load i32* %tmp.4 ; <i32> [#uses=1] + ret i32 %tmp.5 } diff --git a/test/CodeGen/PowerPC/reg-coalesce-simple.ll b/test/CodeGen/PowerPC/reg-coalesce-simple.ll index 60e9458..dccf21e 100644 --- a/test/CodeGen/PowerPC/reg-coalesce-simple.ll +++ b/test/CodeGen/PowerPC/reg-coalesce-simple.ll @@ -1,11 +1,12 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep or +; RUN: llvm-as < %s | llc -march=ppc32 | not grep or -%struct.foo = type { int, int, [0 x ubyte] } -int %test(%struct.foo* %X) { - %tmp1 = getelementptr %struct.foo* %X, int 0, uint 2, int 100 - %tmp = load ubyte* %tmp1 ; <ubyte> [#uses=1] - %tmp2 = cast ubyte %tmp to int ; <int> [#uses=1] - ret int %tmp2} +%struct.foo = type { i32, i32, [0 x i8] } +define i32 @test(%struct.foo* %X) { + %tmp1 = getelementptr %struct.foo* %X, i32 0, i32 2, i32 100 ; <i8*> [#uses=1] + %tmp = load i8* %tmp1 ; <i8> [#uses=1] + %tmp2 = zext i8 %tmp to i32 ; <i32> [#uses=1] + ret i32 %tmp2 +} diff --git a/test/CodeGen/PowerPC/rlwimi-commute.ll b/test/CodeGen/PowerPC/rlwimi-commute.ll index 8e6b1d6..ed2e7b0 100644 --- a/test/CodeGen/PowerPC/rlwimi-commute.ll +++ b/test/CodeGen/PowerPC/rlwimi-commute.ll @@ -1,26 +1,26 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep rlwimi -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep {or } +; RUN: llvm-as < %s | llc -march=ppc32 | grep rlwimi +; RUN: llvm-as < %s | llc -march=ppc32 | not grep {or } ; Make sure there is no register-register copies here. -void %test1(int *%A, int *%B, int *%D, int* %E) { - %A = load int* %A - %B = load int* %B - %X = and int %A, 15 - %Y = and int %B, -16 - %Z = or int %X, %Y - store int %Z, int* %D - store int %A, int* %E +define void @test1(i32* %A, i32* %B, i32* %D, i32* %E) { + %A.upgrd.1 = load i32* %A ; <i32> [#uses=2] + %B.upgrd.2 = load i32* %B ; <i32> [#uses=1] + %X = and i32 %A.upgrd.1, 15 ; <i32> [#uses=1] + %Y = and i32 %B.upgrd.2, -16 ; <i32> [#uses=1] + %Z = or i32 %X, %Y ; <i32> [#uses=1] + store i32 %Z, i32* %D + store i32 %A.upgrd.1, i32* %E ret void } -void %test2(int *%A, int *%B, int *%D, int* %E) { - %A = load int* %A - %B = load int* %B - %X = and int %A, 15 - %Y = and int %B, -16 - %Z = or int %X, %Y - store int %Z, int* %D - store int %B, int* %E +define void @test2(i32* %A, i32* %B, i32* %D, i32* %E) { + %A.upgrd.3 = load i32* %A ; <i32> [#uses=1] + %B.upgrd.4 = load i32* %B ; <i32> [#uses=2] + %X = and i32 %A.upgrd.3, 15 ; <i32> [#uses=1] + %Y = and i32 %B.upgrd.4, -16 ; <i32> [#uses=1] + %Z = or i32 %X, %Y ; <i32> [#uses=1] + store i32 %Z, i32* %D + store i32 %B.upgrd.4, i32* %E ret void } diff --git a/test/CodeGen/PowerPC/rlwimi.ll b/test/CodeGen/PowerPC/rlwimi.ll index 25ac0e9..5e310bb 100644 --- a/test/CodeGen/PowerPC/rlwimi.ll +++ b/test/CodeGen/PowerPC/rlwimi.ll @@ -1,72 +1,70 @@ ; All of these ands and shifts should be folded into rlwimi's -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep and -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep rlwimi | count 8 +; RUN: llvm-as < %s | llc -march=ppc32 | not grep and +; RUN: llvm-as < %s | llc -march=ppc32 | grep rlwimi | count 8 -implementation ; Functions: - -int %test1(int %x, int %y) { +define i32 @test1(i32 %x, i32 %y) { entry: - %tmp.3 = shl int %x, ubyte 16 ; <int> [#uses=1] - %tmp.7 = and int %y, 65535 ; <int> [#uses=1] - %tmp.9 = or int %tmp.7, %tmp.3 ; <int> [#uses=1] - ret int %tmp.9 + %tmp.3 = shl i32 %x, 16 ; <i32> [#uses=1] + %tmp.7 = and i32 %y, 65535 ; <i32> [#uses=1] + %tmp.9 = or i32 %tmp.7, %tmp.3 ; <i32> [#uses=1] + ret i32 %tmp.9 } -int %test2(int %x, int %y) { +define i32 @test2(i32 %x, i32 %y) { entry: - %tmp.7 = and int %x, 65535 ; <int> [#uses=1] - %tmp.3 = shl int %y, ubyte 16 ; <int> [#uses=1] - %tmp.9 = or int %tmp.7, %tmp.3 ; <int> [#uses=1] - ret int %tmp.9 + %tmp.7 = and i32 %x, 65535 ; <i32> [#uses=1] + %tmp.3 = shl i32 %y, 16 ; <i32> [#uses=1] + %tmp.9 = or i32 %tmp.7, %tmp.3 ; <i32> [#uses=1] + ret i32 %tmp.9 } -uint %test3(uint %x, uint %y) { +define i32 @test3(i32 %x, i32 %y) { entry: - %tmp.3 = shr uint %x, ubyte 16 ; <uint> [#uses=1] - %tmp.6 = and uint %y, 4294901760 ; <uint> [#uses=1] - %tmp.7 = or uint %tmp.6, %tmp.3 ; <uint> [#uses=1] - ret uint %tmp.7 + %tmp.3 = lshr i32 %x, 16 ; <i32> [#uses=1] + %tmp.6 = and i32 %y, -65536 ; <i32> [#uses=1] + %tmp.7 = or i32 %tmp.6, %tmp.3 ; <i32> [#uses=1] + ret i32 %tmp.7 } -uint %test4(uint %x, uint %y) { +define i32 @test4(i32 %x, i32 %y) { entry: - %tmp.6 = and uint %x, 4294901760 ; <uint> [#uses=1] - %tmp.3 = shr uint %y, ubyte 16 ; <uint> [#uses=1] - %tmp.7 = or uint %tmp.6, %tmp.3 ; <uint> [#uses=1] - ret uint %tmp.7 + %tmp.6 = and i32 %x, -65536 ; <i32> [#uses=1] + %tmp.3 = lshr i32 %y, 16 ; <i32> [#uses=1] + %tmp.7 = or i32 %tmp.6, %tmp.3 ; <i32> [#uses=1] + ret i32 %tmp.7 } -int %test5(int %x, int %y) { +define i32 @test5(i32 %x, i32 %y) { entry: - %tmp.3 = shl int %x, ubyte 1 ; <int> [#uses=1] - %tmp.4 = and int %tmp.3, -65536 ; <int> [#uses=1] - %tmp.7 = and int %y, 65535 ; <int> [#uses=1] - %tmp.9 = or int %tmp.4, %tmp.7 ; <int> [#uses=1] - ret int %tmp.9 + %tmp.3 = shl i32 %x, 1 ; <i32> [#uses=1] + %tmp.4 = and i32 %tmp.3, -65536 ; <i32> [#uses=1] + %tmp.7 = and i32 %y, 65535 ; <i32> [#uses=1] + %tmp.9 = or i32 %tmp.4, %tmp.7 ; <i32> [#uses=1] + ret i32 %tmp.9 } -int %test6(int %x, int %y) { +define i32 @test6(i32 %x, i32 %y) { entry: - %tmp.7 = and int %x, 65535 ; <int> [#uses=1] - %tmp.3 = shl int %y, ubyte 1 ; <int> [#uses=1] - %tmp.4 = and int %tmp.3, -65536 ; <int> [#uses=1] - %tmp.9 = or int %tmp.4, %tmp.7 ; <int> [#uses=1] - ret int %tmp.9 + %tmp.7 = and i32 %x, 65535 ; <i32> [#uses=1] + %tmp.3 = shl i32 %y, 1 ; <i32> [#uses=1] + %tmp.4 = and i32 %tmp.3, -65536 ; <i32> [#uses=1] + %tmp.9 = or i32 %tmp.4, %tmp.7 ; <i32> [#uses=1] + ret i32 %tmp.9 } -int %test7(int %x, int %y) { +define i32 @test7(i32 %x, i32 %y) { entry: - %tmp.2 = and int %x, -65536 ; <int> [#uses=1] - %tmp.5 = and int %y, 65535 ; <int> [#uses=1] - %tmp.7 = or int %tmp.5, %tmp.2 ; <int> [#uses=1] - ret int %tmp.7 + %tmp.2 = and i32 %x, -65536 ; <i32> [#uses=1] + %tmp.5 = and i32 %y, 65535 ; <i32> [#uses=1] + %tmp.7 = or i32 %tmp.5, %tmp.2 ; <i32> [#uses=1] + ret i32 %tmp.7 } -uint %test8(uint %bar) { +define i32 @test8(i32 %bar) { entry: - %tmp.3 = shl uint %bar, ubyte 1 ; <uint> [#uses=1] - %tmp.4 = and uint %tmp.3, 2 ; <uint> [#uses=1] - %tmp.6 = and uint %bar, 4294967293 ; <uint> [#uses=1] - %tmp.7 = or uint %tmp.4, %tmp.6 ; <uint> [#uses=1] - ret uint %tmp.7 + %tmp.3 = shl i32 %bar, 1 ; <i32> [#uses=1] + %tmp.4 = and i32 %tmp.3, 2 ; <i32> [#uses=1] + %tmp.6 = and i32 %bar, -3 ; <i32> [#uses=1] + %tmp.7 = or i32 %tmp.4, %tmp.6 ; <i32> [#uses=1] + ret i32 %tmp.7 } diff --git a/test/CodeGen/PowerPC/rlwimi2.ll b/test/CodeGen/PowerPC/rlwimi2.ll index 501daf0..33eaacf 100644 --- a/test/CodeGen/PowerPC/rlwimi2.ll +++ b/test/CodeGen/PowerPC/rlwimi2.ll @@ -1,31 +1,29 @@ ; All of these ands and shifts should be folded into rlwimi's -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -o %t -f +; RUN: llvm-as < %s | llc -march=ppc32 -o %t -f ; RUN: grep rlwimi %t | count 3 ; RUN: grep srwi %t | count 1 ; RUN: not grep slwi %t -implementation ; Functions: - -ushort %test1(uint %srcA, uint %srcB, uint %alpha) { +define i16 @test1(i32 %srcA, i32 %srcB, i32 %alpha) { entry: - %tmp.1 = shl uint %srcA, ubyte 15 ; <uint> [#uses=1] - %tmp.4 = and uint %tmp.1, 32505856 ; <uint> [#uses=1] - %tmp.6 = and uint %srcA, 31775 ; <uint> [#uses=1] - %tmp.7 = or uint %tmp.4, %tmp.6 ; <uint> [#uses=1] - %tmp.9 = shl uint %srcB, ubyte 15 ; <uint> [#uses=1] - %tmp.12 = and uint %tmp.9, 32505856 ; <uint> [#uses=1] - %tmp.14 = and uint %srcB, 31775 ; <uint> [#uses=1] - %tmp.15 = or uint %tmp.12, %tmp.14 ; <uint> [#uses=1] - %tmp.18 = mul uint %tmp.7, %alpha ; <uint> [#uses=1] - %tmp.20 = sub uint 32, %alpha ; <uint> [#uses=1] - %tmp.22 = mul uint %tmp.15, %tmp.20 ; <uint> [#uses=1] - %tmp.23 = add uint %tmp.22, %tmp.18 ; <uint> [#uses=2] - %tmp.27 = shr uint %tmp.23, ubyte 5 ; <uint> [#uses=1] - %tmp.28 = cast uint %tmp.27 to ushort ; <ushort> [#uses=1] - %tmp.29 = and ushort %tmp.28, 31775 ; <ushort> [#uses=1] - %tmp.33 = shr uint %tmp.23, ubyte 20 ; <uint> [#uses=1] - %tmp.34 = cast uint %tmp.33 to ushort ; <ushort> [#uses=1] - %tmp.35 = and ushort %tmp.34, 992 ; <ushort> [#uses=1] - %tmp.36 = or ushort %tmp.29, %tmp.35 ; <ushort> [#uses=1] - ret ushort %tmp.36 + %tmp.1 = shl i32 %srcA, 15 ; <i32> [#uses=1] + %tmp.4 = and i32 %tmp.1, 32505856 ; <i32> [#uses=1] + %tmp.6 = and i32 %srcA, 31775 ; <i32> [#uses=1] + %tmp.7 = or i32 %tmp.4, %tmp.6 ; <i32> [#uses=1] + %tmp.9 = shl i32 %srcB, 15 ; <i32> [#uses=1] + %tmp.12 = and i32 %tmp.9, 32505856 ; <i32> [#uses=1] + %tmp.14 = and i32 %srcB, 31775 ; <i32> [#uses=1] + %tmp.15 = or i32 %tmp.12, %tmp.14 ; <i32> [#uses=1] + %tmp.18 = mul i32 %tmp.7, %alpha ; <i32> [#uses=1] + %tmp.20 = sub i32 32, %alpha ; <i32> [#uses=1] + %tmp.22 = mul i32 %tmp.15, %tmp.20 ; <i32> [#uses=1] + %tmp.23 = add i32 %tmp.22, %tmp.18 ; <i32> [#uses=2] + %tmp.27 = lshr i32 %tmp.23, 5 ; <i32> [#uses=1] + %tmp.28 = trunc i32 %tmp.27 to i16 ; <i16> [#uses=1] + %tmp.29 = and i16 %tmp.28, 31775 ; <i16> [#uses=1] + %tmp.33 = lshr i32 %tmp.23, 20 ; <i32> [#uses=1] + %tmp.34 = trunc i32 %tmp.33 to i16 ; <i16> [#uses=1] + %tmp.35 = and i16 %tmp.34, 992 ; <i16> [#uses=1] + %tmp.36 = or i16 %tmp.29, %tmp.35 ; <i16> [#uses=1] + ret i16 %tmp.36 } diff --git a/test/CodeGen/PowerPC/rlwimi3.ll b/test/CodeGen/PowerPC/rlwimi3.ll index b313ef9..fedcfbf 100644 --- a/test/CodeGen/PowerPC/rlwimi3.ll +++ b/test/CodeGen/PowerPC/rlwimi3.ll @@ -1,26 +1,25 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -stats |& \ +; RUN: llvm-as < %s | llc -march=ppc32 -stats |& \ ; RUN: grep {Number of machine instrs printed} | grep 12 -ushort %Trans16Bit(uint %srcA, uint %srcB, uint %alpha) { - %tmp1 = shl uint %srcA, ubyte 15 ; <uint> [#uses=1] - %tmp2 = and uint %tmp1, 32505856 ; <uint> [#uses=1] - %tmp4 = and uint %srcA, 31775 ; <uint> [#uses=1] - %tmp5 = or uint %tmp2, %tmp4 ; <uint> [#uses=1] - %tmp7 = shl uint %srcB, ubyte 15 ; <uint> [#uses=1] - %tmp8 = and uint %tmp7, 32505856 ; <uint> [#uses=1] - %tmp10 = and uint %srcB, 31775 ; <uint> [#uses=1] - %tmp11 = or uint %tmp8, %tmp10 ; <uint> [#uses=1] - %tmp14 = mul uint %tmp5, %alpha ; <uint> [#uses=1] - %tmp16 = sub uint 32, %alpha ; <uint> [#uses=1] - %tmp18 = mul uint %tmp11, %tmp16 ; <uint> [#uses=1] - %tmp19 = add uint %tmp18, %tmp14 ; <uint> [#uses=2] - %tmp21 = shr uint %tmp19, ubyte 5 ; <uint> [#uses=1] - %tmp21 = cast uint %tmp21 to ushort ; <ushort> [#uses=1] - %tmp = and ushort %tmp21, 31775 ; <ushort> [#uses=1] - %tmp23 = shr uint %tmp19, ubyte 20 ; <uint> [#uses=1] - %tmp23 = cast uint %tmp23 to ushort ; <ushort> [#uses=1] - %tmp24 = and ushort %tmp23, 992 ; <ushort> [#uses=1] - %tmp25 = or ushort %tmp, %tmp24 ; <ushort> [#uses=1] - ret ushort %tmp25 +define i16 @Trans16Bit(i32 %srcA, i32 %srcB, i32 %alpha) { + %tmp1 = shl i32 %srcA, 15 ; <i32> [#uses=1] + %tmp2 = and i32 %tmp1, 32505856 ; <i32> [#uses=1] + %tmp4 = and i32 %srcA, 31775 ; <i32> [#uses=1] + %tmp5 = or i32 %tmp2, %tmp4 ; <i32> [#uses=1] + %tmp7 = shl i32 %srcB, 15 ; <i32> [#uses=1] + %tmp8 = and i32 %tmp7, 32505856 ; <i32> [#uses=1] + %tmp10 = and i32 %srcB, 31775 ; <i32> [#uses=1] + %tmp11 = or i32 %tmp8, %tmp10 ; <i32> [#uses=1] + %tmp14 = mul i32 %tmp5, %alpha ; <i32> [#uses=1] + %tmp16 = sub i32 32, %alpha ; <i32> [#uses=1] + %tmp18 = mul i32 %tmp11, %tmp16 ; <i32> [#uses=1] + %tmp19 = add i32 %tmp18, %tmp14 ; <i32> [#uses=2] + %tmp21 = lshr i32 %tmp19, 5 ; <i32> [#uses=1] + %tmp21.upgrd.1 = trunc i32 %tmp21 to i16 ; <i16> [#uses=1] + %tmp = and i16 %tmp21.upgrd.1, 31775 ; <i16> [#uses=1] + %tmp23 = lshr i32 %tmp19, 20 ; <i32> [#uses=1] + %tmp23.upgrd.2 = trunc i32 %tmp23 to i16 ; <i16> [#uses=1] + %tmp24 = and i16 %tmp23.upgrd.2, 992 ; <i16> [#uses=1] + %tmp25 = or i16 %tmp, %tmp24 ; <i16> [#uses=1] + ret i16 %tmp25 } - diff --git a/test/CodeGen/PowerPC/rlwinm.ll b/test/CodeGen/PowerPC/rlwinm.ll index a80371b..9d34865 100644 --- a/test/CodeGen/PowerPC/rlwinm.ll +++ b/test/CodeGen/PowerPC/rlwinm.ll @@ -1,64 +1,61 @@ ; All of these ands and shifts should be folded into rlwimi's -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -o %t -f -; RUN: not grep and %t -; RUN: not grep srawi %t -; RUN: not grep srwi %t -; RUN: not grep slwi %t +; RUN: llvm-as < %s | llc -march=ppc32 -o %t -f +; RUN: not grep and %t +; RUN: not grep srawi %t +; RUN: not grep srwi %t +; RUN: not grep slwi %t ; RUN: grep rlwinm %t | count 8 -implementation ; Functions: - -int %test1(int %a) { +define i32 @test1(i32 %a) { entry: - %tmp.1 = and int %a, 268431360 ; <int> [#uses=1] - ret int %tmp.1 + %tmp.1 = and i32 %a, 268431360 ; <i32> [#uses=1] + ret i32 %tmp.1 } -int %test2(int %a) { +define i32 @test2(i32 %a) { entry: - %tmp.1 = and int %a, -268435441 ; <int> [#uses=1] - ret int %tmp.1 + %tmp.1 = and i32 %a, -268435441 ; <i32> [#uses=1] + ret i32 %tmp.1 } -int %test3(int %a) { +define i32 @test3(i32 %a) { entry: - %tmp.2 = shr int %a, ubyte 8 ; <int> [#uses=1] - %tmp.3 = and int %tmp.2, 255 ; <int> [#uses=1] - ret int %tmp.3 + %tmp.2 = ashr i32 %a, 8 ; <i32> [#uses=1] + %tmp.3 = and i32 %tmp.2, 255 ; <i32> [#uses=1] + ret i32 %tmp.3 } -uint %test4(uint %a) { +define i32 @test4(i32 %a) { entry: - %tmp.3 = shr uint %a, ubyte 8 ; <uint> [#uses=1] - %tmp.4 = and uint %tmp.3, 255 ; <uint> [#uses=1] - ret uint %tmp.4 + %tmp.3 = lshr i32 %a, 8 ; <i32> [#uses=1] + %tmp.4 = and i32 %tmp.3, 255 ; <i32> [#uses=1] + ret i32 %tmp.4 } -int %test5(int %a) { +define i32 @test5(i32 %a) { entry: - %tmp.2 = shl int %a, ubyte 8 ; <int> [#uses=1] - %tmp.3 = and int %tmp.2, -8388608 ; <int> [#uses=1] - ret int %tmp.3 + %tmp.2 = shl i32 %a, 8 ; <i32> [#uses=1] + %tmp.3 = and i32 %tmp.2, -8388608 ; <i32> [#uses=1] + ret i32 %tmp.3 } -int %test6(int %a) { +define i32 @test6(i32 %a) { entry: - %tmp.1 = and int %a, 65280 ; <int> [#uses=1] - %tmp.2 = shr int %tmp.1, ubyte 8 ; <uint> [#uses=1] - ret int %tmp.2 + %tmp.1 = and i32 %a, 65280 ; <i32> [#uses=1] + %tmp.2 = ashr i32 %tmp.1, 8 ; <i32> [#uses=1] + ret i32 %tmp.2 } -uint %test7(uint %a) { +define i32 @test7(i32 %a) { entry: - %tmp.1 = and uint %a, 65280 ; <uint> [#uses=1] - %tmp.2 = shr uint %tmp.1, ubyte 8 ; <uint> [#uses=1] - ret uint %tmp.2 + %tmp.1 = and i32 %a, 65280 ; <i32> [#uses=1] + %tmp.2 = lshr i32 %tmp.1, 8 ; <i32> [#uses=1] + ret i32 %tmp.2 } -int %test8(int %a) { +define i32 @test8(i32 %a) { entry: - %tmp.1 = and int %a, 16711680 ; <int> [#uses=1] - %tmp.2 = shl int %tmp.1, ubyte 8 ; <int> [#uses=1] - ret int %tmp.2 + %tmp.1 = and i32 %a, 16711680 ; <i32> [#uses=1] + %tmp.2 = shl i32 %tmp.1, 8 ; <i32> [#uses=1] + ret i32 %tmp.2 } - diff --git a/test/CodeGen/PowerPC/select_lt0.ll b/test/CodeGen/PowerPC/select_lt0.ll index bb5213f..86eb201 100644 --- a/test/CodeGen/PowerPC/select_lt0.ll +++ b/test/CodeGen/PowerPC/select_lt0.ll @@ -1,51 +1,50 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep cmp +; RUN: llvm-as < %s | llc -march=ppc32 | not grep cmp -int %seli32_1(int %a) { +define i32 @seli32_1(i32 %a) { entry: - %tmp.1 = setlt int %a, 0 - %retval = select bool %tmp.1, int 5, int 0 - ret int %retval + %tmp.1 = icmp slt i32 %a, 0 ; <i1> [#uses=1] + %retval = select i1 %tmp.1, i32 5, i32 0 ; <i32> [#uses=1] + ret i32 %retval } -int %seli32_2(int %a, int %b) { +define i32 @seli32_2(i32 %a, i32 %b) { entry: - %tmp.1 = setlt int %a, 0 - %retval = select bool %tmp.1, int %b, int 0 - ret int %retval + %tmp.1 = icmp slt i32 %a, 0 ; <i1> [#uses=1] + %retval = select i1 %tmp.1, i32 %b, i32 0 ; <i32> [#uses=1] + ret i32 %retval } -int %seli32_3(int %a, short %b) { +define i32 @seli32_3(i32 %a, i16 %b) { entry: - %tmp.2 = cast short %b to int - %tmp.1 = setlt int %a, 0 - %retval = select bool %tmp.1, int %tmp.2, int 0 - ret int %retval + %tmp.2 = sext i16 %b to i32 ; <i32> [#uses=1] + %tmp.1 = icmp slt i32 %a, 0 ; <i1> [#uses=1] + %retval = select i1 %tmp.1, i32 %tmp.2, i32 0 ; <i32> [#uses=1] + ret i32 %retval } -int %seli32_4(int %a, ushort %b) { +define i32 @seli32_4(i32 %a, i16 %b) { entry: - %tmp.2 = cast ushort %b to int - %tmp.1 = setlt int %a, 0 - %retval = select bool %tmp.1, int %tmp.2, int 0 - ret int %retval + %tmp.2 = zext i16 %b to i32 ; <i32> [#uses=1] + %tmp.1 = icmp slt i32 %a, 0 ; <i1> [#uses=1] + %retval = select i1 %tmp.1, i32 %tmp.2, i32 0 ; <i32> [#uses=1] + ret i32 %retval } -short %seli16_1(short %a) { +define i16 @seli16_1(i16 %a) { entry: - %tmp.1 = setlt short %a, 0 - %retval = select bool %tmp.1, short 7, short 0 - ret short %retval + %tmp.1 = icmp slt i16 %a, 0 ; <i1> [#uses=1] + %retval = select i1 %tmp.1, i16 7, i16 0 ; <i16> [#uses=1] + ret i16 %retval } -short %seli16_2(int %a, short %b) { - %tmp.1 = setlt int %a, 0 - %retval = select bool %tmp.1, short %b, short 0 - ret short %retval +define i16 @seli16_2(i32 %a, i16 %b) { + %tmp.1 = icmp slt i32 %a, 0 ; <i1> [#uses=1] + %retval = select i1 %tmp.1, i16 %b, i16 0 ; <i16> [#uses=1] + ret i16 %retval } -int %seli32_a_a(int %a) { - %tmp = setlt int %a, 1 - %min = select bool %tmp, int %a, int 0 - ret int %min +define i32 @seli32_a_a(i32 %a) { + %tmp = icmp slt i32 %a, 1 ; <i1> [#uses=1] + %min = select i1 %tmp, i32 %a, i32 0 ; <i32> [#uses=1] + ret i32 %min } - diff --git a/test/CodeGen/PowerPC/setcc_no_zext.ll b/test/CodeGen/PowerPC/setcc_no_zext.ll index 00e9bf0..c31f35c 100644 --- a/test/CodeGen/PowerPC/setcc_no_zext.ll +++ b/test/CodeGen/PowerPC/setcc_no_zext.ll @@ -1,8 +1,9 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep rlwinm +; RUN: llvm-as < %s | llc -march=ppc32 | not grep rlwinm -int %setcc_one_or_zero(int* %a) { +define i32 @setcc_one_or_zero(i32* %a) { entry: - %tmp.1 = setne int* %a, null - %inc.1 = cast bool %tmp.1 to int - ret int %inc.1 + %tmp.1 = icmp ne i32* %a, null ; <i1> [#uses=1] + %inc.1 = zext i1 %tmp.1 to i32 ; <i32> [#uses=1] + ret i32 %inc.1 } + diff --git a/test/CodeGen/PowerPC/seteq-0.ll b/test/CodeGen/PowerPC/seteq-0.ll index a574100..0f0afe9 100644 --- a/test/CodeGen/PowerPC/seteq-0.ll +++ b/test/CodeGen/PowerPC/seteq-0.ll @@ -1,9 +1,10 @@ -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 | \ ; RUN: grep {srwi r., r., 5} -int %eq0(int %a) { - %tmp.1 = seteq int %a, 0 ; <bool> [#uses=1] - %tmp.2 = cast bool %tmp.1 to int ; <int> [#uses=1] - ret int %tmp.2 +define i32 @eq0(i32 %a) { + %tmp.1 = icmp eq i32 %a, 0 ; <i1> [#uses=1] + %tmp.2 = zext i1 %tmp.1 to i32 ; <i32> [#uses=1] + ret i32 %tmp.2 } + diff --git a/test/CodeGen/PowerPC/shl_sext.ll b/test/CodeGen/PowerPC/shl_sext.ll index af18338..61e5cdb 100644 --- a/test/CodeGen/PowerPC/shl_sext.ll +++ b/test/CodeGen/PowerPC/shl_sext.ll @@ -1,17 +1,18 @@ ; This test should not contain a sign extend -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep extsb +; RUN: llvm-as < %s | llc -march=ppc32 | not grep extsb -int %test(uint %mode.0.i.0) { - %tmp.79 = cast uint %mode.0.i.0 to sbyte ; <sbyte> [#uses=1] - %tmp.80 = cast sbyte %tmp.79 to int ; <int> [#uses=1] - %tmp.81 = shl int %tmp.80, ubyte 24 ; <int> [#uses=1] - ret int %tmp.81 +define i32 @test(i32 %mode.0.i.0) { + %tmp.79 = trunc i32 %mode.0.i.0 to i8 ; <i8> [#uses=1] + %tmp.80 = sext i8 %tmp.79 to i32 ; <i32> [#uses=1] + %tmp.81 = shl i32 %tmp.80, 24 ; <i32> [#uses=1] + ret i32 %tmp.81 } -int %test2(uint %mode.0.i.0) { - %tmp.79 = cast uint %mode.0.i.0 to sbyte ; <sbyte> [#uses=1] - %tmp.80 = cast sbyte %tmp.79 to int ; <int> [#uses=1] - %tmp.81 = shl int %tmp.80, ubyte 16 ; <int> [#uses=1] - %tmp.82 = and int %tmp.81, 16711680 - ret int %tmp.82 +define i32 @test2(i32 %mode.0.i.0) { + %tmp.79 = trunc i32 %mode.0.i.0 to i8 ; <i8> [#uses=1] + %tmp.80 = sext i8 %tmp.79 to i32 ; <i32> [#uses=1] + %tmp.81 = shl i32 %tmp.80, 16 ; <i32> [#uses=1] + %tmp.82 = and i32 %tmp.81, 16711680 ; <i32> [#uses=1] + ret i32 %tmp.82 } + diff --git a/test/CodeGen/PowerPC/stfiwx.ll b/test/CodeGen/PowerPC/stfiwx.ll index 2eebc07..c4afb63 100644 --- a/test/CodeGen/PowerPC/stfiwx.ll +++ b/test/CodeGen/PowerPC/stfiwx.ll @@ -1,26 +1,26 @@ -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 -mattr=stfiwx -o %t1 -f ; RUN: grep stfiwx %t1 ; RUN: not grep r1 %t1 -; RUN: llvm-upgrade < %s | llvm-as | \ +; RUN: llvm-as < %s | \ ; RUN: llc -march=ppc32 -mtriple=powerpc-apple-darwin8 -mattr=-stfiwx \ ; RUN: -o %t2 -f ; RUN: not grep stfiwx %t2 ; RUN: grep r1 %t2 -void %test(float %a, int* %b) { - %tmp.2 = cast float %a to int - store int %tmp.2, int* %b +define void @test(float %a, i32* %b) { + %tmp.2 = fptosi float %a to i32 ; <i32> [#uses=1] + store i32 %tmp.2, i32* %b ret void } -void %test2(float %a, int* %b, int %i) { - %tmp.2 = getelementptr int* %b, int 1 - %tmp.5 = getelementptr int* %b, int %i - %tmp.7 = cast float %a to int - store int %tmp.7, int* %tmp.5 - store int %tmp.7, int* %tmp.2 - store int %tmp.7, int* %b +define void @test2(float %a, i32* %b, i32 %i) { + %tmp.2 = getelementptr i32* %b, i32 1 ; <i32*> [#uses=1] + %tmp.5 = getelementptr i32* %b, i32 %i ; <i32*> [#uses=1] + %tmp.7 = fptosi float %a to i32 ; <i32> [#uses=3] + store i32 %tmp.7, i32* %tmp.5 + store i32 %tmp.7, i32* %tmp.2 + store i32 %tmp.7, i32* %b ret void } diff --git a/test/CodeGen/PowerPC/store-load-fwd.ll b/test/CodeGen/PowerPC/store-load-fwd.ll index 761fb5a..5cc4784 100644 --- a/test/CodeGen/PowerPC/store-load-fwd.ll +++ b/test/CodeGen/PowerPC/store-load-fwd.ll @@ -1,6 +1,8 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | not grep lwz -int %test(int* %P) { - store int 1, int* %P - %V = load int* %P - ret int %V +; RUN: llvm-as < %s | llc -march=ppc32 | not grep lwz + +define i32 @test(i32* %P) { + store i32 1, i32* %P + %V = load i32* %P ; <i32> [#uses=1] + ret i32 %V } + diff --git a/test/CodeGen/PowerPC/subc.ll b/test/CodeGen/PowerPC/subc.ll index 36eddb1..4ac9596 100644 --- a/test/CodeGen/PowerPC/subc.ll +++ b/test/CodeGen/PowerPC/subc.ll @@ -1,26 +1,25 @@ ; All of these should be codegen'd without loading immediates -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -o %t -f +; RUN: llvm-as < %s | llc -march=ppc32 -o %t -f ; RUN: grep subfc %t | count 1 ; RUN: grep subfe %t | count 1 ; RUN: grep subfze %t | count 1 ; RUN: grep subfme %t | count 1 ; RUN: grep subfic %t | count 2 -implementation ; Functions: -long %sub_ll(long %a, long %b) { +define i64 @sub_ll(i64 %a, i64 %b) { entry: - %tmp.2 = sub long %a, %b ; <long> [#uses=1] - ret long %tmp.2 + %tmp.2 = sub i64 %a, %b ; <i64> [#uses=1] + ret i64 %tmp.2 } -long %sub_l_5(long %a) { +define i64 @sub_l_5(i64 %a) { entry: - %tmp.1 = sub long 5, %a ; <long> [#uses=1] - ret long %tmp.1 + %tmp.1 = sub i64 5, %a ; <i64> [#uses=1] + ret i64 %tmp.1 } -long %sub_l_m5(long %a) { +define i64 @sub_l_m5(i64 %a) { entry: - %tmp.1 = sub long -5, %a ; <long> [#uses=1] - ret long %tmp.1 + %tmp.1 = sub i64 -5, %a ; <i64> [#uses=1] + ret i64 %tmp.1 } diff --git a/test/CodeGen/PowerPC/vcmp-fold.ll b/test/CodeGen/PowerPC/vcmp-fold.ll index efb727c..815bb0a 100644 --- a/test/CodeGen/PowerPC/vcmp-fold.ll +++ b/test/CodeGen/PowerPC/vcmp-fold.ll @@ -1,21 +1,22 @@ -; This should fold the "vcmpbfp." and "vcmpbfp" instructions into a single +; This should fold the "vcmpbfp." and "vcmpbfp" instructions into a single ; "vcmpbfp.". -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | grep vcmpbfp | count 1 +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | grep vcmpbfp | count 1 -void %test(<4 x float>* %x, <4 x float>* %y, int* %P) { + +define void @test(<4 x float>* %x, <4 x float>* %y, i32* %P) { entry: %tmp = load <4 x float>* %x ; <<4 x float>> [#uses=1] %tmp2 = load <4 x float>* %y ; <<4 x float>> [#uses=1] - %tmp = call int %llvm.ppc.altivec.vcmpbfp.p( int 1, <4 x float> %tmp, <4 x float> %tmp2 ) ; <int> [#uses=1] + %tmp.upgrd.1 = call i32 @llvm.ppc.altivec.vcmpbfp.p( i32 1, <4 x float> %tmp, <4 x float> %tmp2 ) ; <i32> [#uses=1] %tmp4 = load <4 x float>* %x ; <<4 x float>> [#uses=1] %tmp6 = load <4 x float>* %y ; <<4 x float>> [#uses=1] - %tmp = call <4 x int> %llvm.ppc.altivec.vcmpbfp( <4 x float> %tmp4, <4 x float> %tmp6 ) ; <<4 x int>> [#uses=1] - %tmp7 = cast <4 x int> %tmp to <4 x float> ; <<4 x float>> [#uses=1] + %tmp.upgrd.2 = call <4 x i32> @llvm.ppc.altivec.vcmpbfp( <4 x float> %tmp4, <4 x float> %tmp6 ) ; <<4 x i32>> [#uses=1] + %tmp7 = bitcast <4 x i32> %tmp.upgrd.2 to <4 x float> ; <<4 x float>> [#uses=1] store <4 x float> %tmp7, <4 x float>* %x - store int %tmp, int* %P + store i32 %tmp.upgrd.1, i32* %P ret void } -declare int %llvm.ppc.altivec.vcmpbfp.p(int, <4 x float>, <4 x float>) +declare i32 @llvm.ppc.altivec.vcmpbfp.p(i32, <4 x float>, <4 x float>) -declare <4 x int> %llvm.ppc.altivec.vcmpbfp(<4 x float>, <4 x float>) +declare <4 x i32> @llvm.ppc.altivec.vcmpbfp(<4 x float>, <4 x float>) diff --git a/test/CodeGen/PowerPC/vec_br_cmp.ll b/test/CodeGen/PowerPC/vec_br_cmp.ll index bc60bae..6d79967 100644 --- a/test/CodeGen/PowerPC/vec_br_cmp.ll +++ b/test/CodeGen/PowerPC/vec_br_cmp.ll @@ -1,23 +1,22 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 -o %t -f +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 -o %t -f ; RUN: grep vcmpeqfp. %t ; RUN: not grep mfcr %t ; A predicate compare used immediately by a branch should not generate an mfcr. -void %test(<4 x float>* %A, <4 x float>* %B) { - %tmp = load <4 x float>* %A - %tmp3 = load <4 x float>* %B - %tmp = tail call int %llvm.ppc.altivec.vcmpeqfp.p( int 1, <4 x float> %tmp, <4 x float> %tmp3 ) - %tmp = seteq int %tmp, 0 - br bool %tmp, label %cond_true, label %UnifiedReturnBlock +define void @test(<4 x float>* %A, <4 x float>* %B) { + %tmp = load <4 x float>* %A ; <<4 x float>> [#uses=1] + %tmp3 = load <4 x float>* %B ; <<4 x float>> [#uses=1] + %tmp.upgrd.1 = tail call i32 @llvm.ppc.altivec.vcmpeqfp.p( i32 1, <4 x float> %tmp, <4 x float> %tmp3 ) ; <i32> [#uses=1] + %tmp.upgrd.2 = icmp eq i32 %tmp.upgrd.1, 0 ; <i1> [#uses=1] + br i1 %tmp.upgrd.2, label %cond_true, label %UnifiedReturnBlock -cond_true: - store <4 x float> zeroinitializer, <4 x float>* %B - ret void +cond_true: ; preds = %0 + store <4 x float> zeroinitializer, <4 x float>* %B + ret void -UnifiedReturnBlock: - ret void +UnifiedReturnBlock: ; preds = %0 + ret void } -declare int %llvm.ppc.altivec.vcmpeqfp.p(int, <4 x float>, <4 x float>) - +declare i32 @llvm.ppc.altivec.vcmpeqfp.p(i32, <4 x float>, <4 x float>) diff --git a/test/CodeGen/PowerPC/vec_call.ll b/test/CodeGen/PowerPC/vec_call.ll index b2b91fe..8e7a08e 100644 --- a/test/CodeGen/PowerPC/vec_call.ll +++ b/test/CodeGen/PowerPC/vec_call.ll @@ -1,11 +1,11 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 -<4 x int> %test_arg(<4 x int> %A, <4 x int> %B) { - %C = add <4 x int> %A, %B - ret <4 x int> %C +define <4 x i32> @test_arg(<4 x i32> %A, <4 x i32> %B) { + %C = add <4 x i32> %A, %B ; <<4 x i32>> [#uses=1] + ret <4 x i32> %C } -<4 x int> %foo() { - %X = call <4 x int> %test_arg(<4 x int> zeroinitializer, <4 x int> zeroinitializer) - ret <4 x int> %X +define <4 x i32> @foo() { + %X = call <4 x i32> @test_arg( <4 x i32> zeroinitializer, <4 x i32> zeroinitializer ) ; <<4 x i32>> [#uses=1] + ret <4 x i32> %X } diff --git a/test/CodeGen/PowerPC/vec_constants.ll b/test/CodeGen/PowerPC/vec_constants.ll index 507d2d9..c4b42b9 100644 --- a/test/CodeGen/PowerPC/vec_constants.ll +++ b/test/CodeGen/PowerPC/vec_constants.ll @@ -1,47 +1,40 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | not grep CPI - - -; Tests spltw(0x80000000) and spltw(0x7FFFFFFF). -void %test1(<4 x int>* %P1, <4 x int>* %P2, <4 x float>* %P3) { - %tmp = load <4 x int>* %P1 - %tmp4 = and <4 x int> %tmp, < int -2147483648, int -2147483648, int -2147483648, int -2147483648 > - store <4 x int> %tmp4, <4 x int>* %P1 - %tmp7 = load <4 x int>* %P2 - %tmp9 = and <4 x int> %tmp7, < int 2147483647, int 2147483647, int 2147483647, int 2147483647 > - store <4 x int> %tmp9, <4 x int>* %P2 - %tmp = load <4 x float>* %P3 - %tmp11 = cast <4 x float> %tmp to <4 x int> - %tmp12 = and <4 x int> %tmp11, < int 2147483647, int 2147483647, int 2147483647, int 2147483647 > - %tmp13 = cast <4 x int> %tmp12 to <4 x float> - store <4 x float> %tmp13, <4 x float>* %P3 - ret void +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | not grep CPI + +define void @test1(<4 x i32>* %P1, <4 x i32>* %P2, <4 x float>* %P3) { + %tmp = load <4 x i32>* %P1 ; <<4 x i32>> [#uses=1] + %tmp4 = and <4 x i32> %tmp, < i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648 > ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp4, <4 x i32>* %P1 + %tmp7 = load <4 x i32>* %P2 ; <<4 x i32>> [#uses=1] + %tmp9 = and <4 x i32> %tmp7, < i32 2147483647, i32 2147483647, i32 2147483647, i32 2147483647 > ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp9, <4 x i32>* %P2 + %tmp.upgrd.1 = load <4 x float>* %P3 ; <<4 x float>> [#uses=1] + %tmp11 = bitcast <4 x float> %tmp.upgrd.1 to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp12 = and <4 x i32> %tmp11, < i32 2147483647, i32 2147483647, i32 2147483647, i32 2147483647 > ; <<4 x i32>> [#uses=1] + %tmp13 = bitcast <4 x i32> %tmp12 to <4 x float> ; <<4 x float>> [#uses=1] + store <4 x float> %tmp13, <4 x float>* %P3 + ret void } -<4 x int> %test_30() { - ret <4 x int> <int 30, int 30, int 30, int 30> +define <4 x i32> @test_30() { + ret <4 x i32> < i32 30, i32 30, i32 30, i32 30 > } -<4 x int> %test_29() { - ret <4 x int> <int 29, int 29, int 29, int 29> +define <4 x i32> @test_29() { + ret <4 x i32> < i32 29, i32 29, i32 29, i32 29 > } -<8 x short> %test_n30() { - ret <8 x short> <short -30, short -30, short -30, short -30, - short -30, short -30, short -30, short -30> +define <8 x i16> @test_n30() { + ret <8 x i16> < i16 -30, i16 -30, i16 -30, i16 -30, i16 -30, i16 -30, i16 -30, i16 -30 > } -<16 x sbyte> %test_n104() { - ret <16 x sbyte> <sbyte -104, sbyte -104, sbyte -104, sbyte -104, - sbyte -104, sbyte -104, sbyte -104, sbyte -104, - sbyte -104, sbyte -104, sbyte -104, sbyte -104, - sbyte -104, sbyte -104, sbyte -104, sbyte -104> +define <16 x i8> @test_n104() { + ret <16 x i8> < i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104 > } -<4 x int> %test_vsldoi() { - ret <4 x int> <int 512, int 512, int 512, int 512> +define <4 x i32> @test_vsldoi() { + ret <4 x i32> < i32 512, i32 512, i32 512, i32 512 > } -<4 x int> %test_rol() { - ret <4 x int> <int -11534337, int -11534337, int -11534337, int -11534337> +define <4 x i32> @test_rol() { + ret <4 x i32> < i32 -11534337, i32 -11534337, i32 -11534337, i32 -11534337 > } - diff --git a/test/CodeGen/PowerPC/vec_mul.ll b/test/CodeGen/PowerPC/vec_mul.ll index eea1def..b061fa9 100644 --- a/test/CodeGen/PowerPC/vec_mul.ll +++ b/test/CodeGen/PowerPC/vec_mul.ll @@ -1,24 +1,23 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | not grep mullw -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | grep vmsumuhm +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | not grep mullw +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | grep vmsumuhm -<4 x int> %test_v4i32(<4 x int>* %X, <4 x int>* %Y) { - %tmp = load <4 x int>* %X - %tmp2 = load <4 x int>* %Y - %tmp3 = mul <4 x int> %tmp, %tmp2 - ret <4 x int> %tmp3 +define <4 x i32> @test_v4i32(<4 x i32>* %X, <4 x i32>* %Y) { + %tmp = load <4 x i32>* %X ; <<4 x i32>> [#uses=1] + %tmp2 = load <4 x i32>* %Y ; <<4 x i32>> [#uses=1] + %tmp3 = mul <4 x i32> %tmp, %tmp2 ; <<4 x i32>> [#uses=1] + ret <4 x i32> %tmp3 } -<8 x short> %test_v8i16(<8 x short>* %X, <8 x short>* %Y) { - %tmp = load <8 x short>* %X - %tmp2 = load <8 x short>* %Y - %tmp3 = mul <8 x short> %tmp, %tmp2 - ret <8 x short> %tmp3 +define <8 x i16> @test_v8i16(<8 x i16>* %X, <8 x i16>* %Y) { + %tmp = load <8 x i16>* %X ; <<8 x i16>> [#uses=1] + %tmp2 = load <8 x i16>* %Y ; <<8 x i16>> [#uses=1] + %tmp3 = mul <8 x i16> %tmp, %tmp2 ; <<8 x i16>> [#uses=1] + ret <8 x i16> %tmp3 } -<16 x sbyte> %test_v16i8(<16 x sbyte>* %X, <16 x sbyte>* %Y) { - %tmp = load <16 x sbyte>* %X - %tmp2 = load <16 x sbyte>* %Y - %tmp3 = mul <16 x sbyte> %tmp, %tmp2 - ret <16 x sbyte> %tmp3 +define <16 x i8> @test_v16i8(<16 x i8>* %X, <16 x i8>* %Y) { + %tmp = load <16 x i8>* %X ; <<16 x i8>> [#uses=1] + %tmp2 = load <16 x i8>* %Y ; <<16 x i8>> [#uses=1] + %tmp3 = mul <16 x i8> %tmp, %tmp2 ; <<16 x i8>> [#uses=1] + ret <16 x i8> %tmp3 } - diff --git a/test/CodeGen/PowerPC/vec_perf_shuffle.ll b/test/CodeGen/PowerPC/vec_perf_shuffle.ll index 6177b5f..5bb1b60 100644 --- a/test/CodeGen/PowerPC/vec_perf_shuffle.ll +++ b/test/CodeGen/PowerPC/vec_perf_shuffle.ll @@ -1,42 +1,36 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | not grep vperm +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | not grep vperm -<4 x float> %test_uu72(<4 x float> *%P1, <4 x float> *%P2) { - %V1 = load <4 x float> *%P1 - %V2 = load <4 x float> *%P2 - ; vmrglw + vsldoi - %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, - <4 x uint> <uint undef, uint undef, uint 7, uint 2> +define <4 x float> @test_uu72(<4 x float>* %P1, <4 x float>* %P2) { + %V1 = load <4 x float>* %P1 ; <<4 x float>> [#uses=1] + %V2 = load <4 x float>* %P2 ; <<4 x float>> [#uses=1] + %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, <4 x i32> < i32 undef, i32 undef, i32 7, i32 2 > ; <<4 x float>> [#uses=1] ret <4 x float> %V3 } -<4 x float> %test_30u5(<4 x float> *%P1, <4 x float> *%P2) { - %V1 = load <4 x float> *%P1 - %V2 = load <4 x float> *%P2 - %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, - <4 x uint> <uint 3, uint 0, uint undef, uint 5> +define <4 x float> @test_30u5(<4 x float>* %P1, <4 x float>* %P2) { + %V1 = load <4 x float>* %P1 ; <<4 x float>> [#uses=1] + %V2 = load <4 x float>* %P2 ; <<4 x float>> [#uses=1] + %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, <4 x i32> < i32 3, i32 0, i32 undef, i32 5 > ; <<4 x float>> [#uses=1] ret <4 x float> %V3 } -<4 x float> %test_3u73(<4 x float> *%P1, <4 x float> *%P2) { - %V1 = load <4 x float> *%P1 - %V2 = load <4 x float> *%P2 - %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, - <4 x uint> <uint 3, uint undef, uint 7, uint 3> +define <4 x float> @test_3u73(<4 x float>* %P1, <4 x float>* %P2) { + %V1 = load <4 x float>* %P1 ; <<4 x float>> [#uses=1] + %V2 = load <4 x float>* %P2 ; <<4 x float>> [#uses=1] + %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, <4 x i32> < i32 3, i32 undef, i32 7, i32 3 > ; <<4 x float>> [#uses=1] ret <4 x float> %V3 } -<4 x float> %test_3774(<4 x float> *%P1, <4 x float> *%P2) { - %V1 = load <4 x float> *%P1 - %V2 = load <4 x float> *%P2 - %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, - <4 x uint> <uint 3, uint 7, uint 7, uint 4> +define <4 x float> @test_3774(<4 x float>* %P1, <4 x float>* %P2) { + %V1 = load <4 x float>* %P1 ; <<4 x float>> [#uses=1] + %V2 = load <4 x float>* %P2 ; <<4 x float>> [#uses=1] + %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, <4 x i32> < i32 3, i32 7, i32 7, i32 4 > ; <<4 x float>> [#uses=1] ret <4 x float> %V3 } -<4 x float> %test_4450(<4 x float> *%P1, <4 x float> *%P2) { - %V1 = load <4 x float> *%P1 - %V2 = load <4 x float> *%P2 - %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, - <4 x uint> <uint 4, uint 4, uint 5, uint 0> +define <4 x float> @test_4450(<4 x float>* %P1, <4 x float>* %P2) { + %V1 = load <4 x float>* %P1 ; <<4 x float>> [#uses=1] + %V2 = load <4 x float>* %P2 ; <<4 x float>> [#uses=1] + %V3 = shufflevector <4 x float> %V1, <4 x float> %V2, <4 x i32> < i32 4, i32 4, i32 5, i32 0 > ; <<4 x float>> [#uses=1] ret <4 x float> %V3 } diff --git a/test/CodeGen/PowerPC/vec_shuffle.ll b/test/CodeGen/PowerPC/vec_shuffle.ll index e26f6c1..1289dca 100644 --- a/test/CodeGen/PowerPC/vec_shuffle.ll +++ b/test/CodeGen/PowerPC/vec_shuffle.ll @@ -1,506 +1,504 @@ -; RUN: llvm-upgrade < %s | llvm-as | opt -instcombine | \ +; RUN: llvm-as < %s | opt -instcombine | \ ; RUN: llc -march=ppc32 -mcpu=g5 | not grep vperm -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 > %t +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 > %t ; RUN: grep vsldoi %t | count 2 ; RUN: grep vmrgh %t | count 7 ; RUN: grep vmrgl %t | count 6 ; RUN: grep vpkuhum %t | count 1 ; RUN: grep vpkuwum %t | count 1 -void %VSLDOI_xy(<8 x short>* %A, <8 x short>* %B) { +define void @VSLDOI_xy(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=1] - %tmp2 = load <8 x short>* %B ; <<8 x short>> [#uses=1] - %tmp = cast <8 x short> %tmp to <16 x sbyte> ; <<16 x sbyte>> [#uses=11] - %tmp2 = cast <8 x short> %tmp2 to <16 x sbyte> ; <<16 x sbyte>> [#uses=5] - %tmp = extractelement <16 x sbyte> %tmp, uint 5 ; <sbyte> [#uses=1] - %tmp3 = extractelement <16 x sbyte> %tmp, uint 6 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 7 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp, uint 8 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 9 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp, uint 10 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 11 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp, uint 12 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 13 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp, uint 14 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp, uint 15 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp2, uint 0 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp2, uint 1 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp2, uint 2 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp2, uint 3 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp2, uint 4 ; <sbyte> [#uses=1] - %tmp18 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp19 = insertelement <16 x sbyte> %tmp18, sbyte %tmp3, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 15 ; <<16 x sbyte>> [#uses=1] - %tmp33 = cast <16 x sbyte> %tmp33 to <8 x short> ; <<8 x short>> [#uses=1] - store <8 x short> %tmp33, <8 x short>* %A + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=1] + %tmp2 = load <8 x i16>* %B ; <<8 x i16>> [#uses=1] + %tmp.upgrd.1 = bitcast <8 x i16> %tmp to <16 x i8> ; <<16 x i8>> [#uses=11] + %tmp2.upgrd.2 = bitcast <8 x i16> %tmp2 to <16 x i8> ; <<16 x i8>> [#uses=5] + %tmp.upgrd.3 = extractelement <16 x i8> %tmp.upgrd.1, i32 5 ; <i8> [#uses=1] + %tmp3 = extractelement <16 x i8> %tmp.upgrd.1, i32 6 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp.upgrd.1, i32 7 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp.upgrd.1, i32 8 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp.upgrd.1, i32 9 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp.upgrd.1, i32 10 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp.upgrd.1, i32 11 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp.upgrd.1, i32 12 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp.upgrd.1, i32 13 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp.upgrd.1, i32 14 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp.upgrd.1, i32 15 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp2.upgrd.2, i32 0 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp2.upgrd.2, i32 1 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp2.upgrd.2, i32 2 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp2.upgrd.2, i32 3 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp2.upgrd.2, i32 4 ; <i8> [#uses=1] + %tmp18 = insertelement <16 x i8> undef, i8 %tmp.upgrd.3, i32 0 ; <<16 x i8>> [#uses=1] + %tmp19 = insertelement <16 x i8> %tmp18, i8 %tmp3, i32 1 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 2 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 3 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 4 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 5 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 6 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 7 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 8 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 9 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 10 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 11 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 12 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 13 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 14 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 15 ; <<16 x i8>> [#uses=1] + %tmp33.upgrd.4 = bitcast <16 x i8> %tmp33 to <8 x i16> ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp33.upgrd.4, <8 x i16>* %A ret void } -void %VSLDOI_xx(<8 x short>* %A, <8 x short>* %B) { - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=1] - %tmp2 = load <8 x short>* %A ; <<8 x short>> [#uses=1] - %tmp = cast <8 x short> %tmp to <16 x sbyte> ; <<16 x sbyte>> [#uses=11] - %tmp2 = cast <8 x short> %tmp2 to <16 x sbyte> ; <<16 x sbyte>> [#uses=5] - %tmp = extractelement <16 x sbyte> %tmp, uint 5 ; <sbyte> [#uses=1] - %tmp3 = extractelement <16 x sbyte> %tmp, uint 6 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 7 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp, uint 8 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 9 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp, uint 10 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 11 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp, uint 12 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 13 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp, uint 14 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp, uint 15 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp2, uint 0 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp2, uint 1 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp2, uint 2 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp2, uint 3 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp2, uint 4 ; <sbyte> [#uses=1] - %tmp18 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp19 = insertelement <16 x sbyte> %tmp18, sbyte %tmp3, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 15 ; <<16 x sbyte>> [#uses=1] - %tmp33 = cast <16 x sbyte> %tmp33 to <8 x short> ; <<8 x short>> [#uses=1] - store <8 x short> %tmp33, <8 x short>* %A +define void @VSLDOI_xx(<8 x i16>* %A, <8 x i16>* %B) { + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=1] + %tmp2 = load <8 x i16>* %A ; <<8 x i16>> [#uses=1] + %tmp.upgrd.5 = bitcast <8 x i16> %tmp to <16 x i8> ; <<16 x i8>> [#uses=11] + %tmp2.upgrd.6 = bitcast <8 x i16> %tmp2 to <16 x i8> ; <<16 x i8>> [#uses=5] + %tmp.upgrd.7 = extractelement <16 x i8> %tmp.upgrd.5, i32 5 ; <i8> [#uses=1] + %tmp3 = extractelement <16 x i8> %tmp.upgrd.5, i32 6 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp.upgrd.5, i32 7 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp.upgrd.5, i32 8 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp.upgrd.5, i32 9 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp.upgrd.5, i32 10 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp.upgrd.5, i32 11 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp.upgrd.5, i32 12 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp.upgrd.5, i32 13 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp.upgrd.5, i32 14 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp.upgrd.5, i32 15 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp2.upgrd.6, i32 0 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp2.upgrd.6, i32 1 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp2.upgrd.6, i32 2 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp2.upgrd.6, i32 3 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp2.upgrd.6, i32 4 ; <i8> [#uses=1] + %tmp18 = insertelement <16 x i8> undef, i8 %tmp.upgrd.7, i32 0 ; <<16 x i8>> [#uses=1] + %tmp19 = insertelement <16 x i8> %tmp18, i8 %tmp3, i32 1 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 2 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 3 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 4 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 5 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 6 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 7 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 8 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 9 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 10 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 11 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 12 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 13 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 14 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 15 ; <<16 x i8>> [#uses=1] + %tmp33.upgrd.8 = bitcast <16 x i8> %tmp33 to <8 x i16> ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp33.upgrd.8, <8 x i16>* %A ret void } -void %VPERM_promote(<8 x short>* %A, <8 x short>* %B) { +define void @VPERM_promote(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=1] - %tmp = cast <8 x short> %tmp to <4 x int> ; <<4 x int>> [#uses=1] - %tmp2 = load <8 x short>* %B ; <<8 x short>> [#uses=1] - %tmp2 = cast <8 x short> %tmp2 to <4 x int> ; <<4 x int>> [#uses=1] - %tmp3 = call <4 x int> %llvm.ppc.altivec.vperm( <4 x int> %tmp, <4 x int> %tmp2, <16 x sbyte> < sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14, sbyte 14 > ) ; <<4 x int>> [#uses=1] - %tmp3 = cast <4 x int> %tmp3 to <8 x short> ; <<8 x short>> [#uses=1] - store <8 x short> %tmp3, <8 x short>* %A - ret void + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=1] + %tmp.upgrd.9 = bitcast <8 x i16> %tmp to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp2 = load <8 x i16>* %B ; <<8 x i16>> [#uses=1] + %tmp2.upgrd.10 = bitcast <8 x i16> %tmp2 to <4 x i32> ; <<4 x i32>> [#uses=1] + %tmp3 = call <4 x i32> @llvm.ppc.altivec.vperm( <4 x i32> %tmp.upgrd.9, <4 x i32> %tmp2.upgrd.10, <16 x i8> < i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14, i8 14 > ) ; <<4 x i32>> [#uses=1] + %tmp3.upgrd.11 = bitcast <4 x i32> %tmp3 to <8 x i16> ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp3.upgrd.11, <8 x i16>* %A + ret void } -declare <4 x int> %llvm.ppc.altivec.vperm(<4 x int>, <4 x int>, <16 x sbyte>) - +declare <4 x i32> @llvm.ppc.altivec.vperm(<4 x i32>, <4 x i32>, <16 x i8>) -void %tb_l(<16 x sbyte>* %A, <16 x sbyte>* %B) { +define void @tb_l(<16 x i8>* %A, <16 x i8>* %B) { entry: - %tmp = load <16 x sbyte>* %A ; <<16 x sbyte>> [#uses=8] - %tmp2 = load <16 x sbyte>* %B ; <<16 x sbyte>> [#uses=8] - %tmp = extractelement <16 x sbyte> %tmp, uint 8 ; <sbyte> [#uses=1] - %tmp3 = extractelement <16 x sbyte> %tmp2, uint 8 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 9 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp2, uint 9 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 10 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp2, uint 10 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 11 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp2, uint 11 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 12 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp2, uint 12 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp, uint 13 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp2, uint 13 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp, uint 14 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp2, uint 14 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp, uint 15 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp2, uint 15 ; <sbyte> [#uses=1] - %tmp18 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp19 = insertelement <16 x sbyte> %tmp18, sbyte %tmp3, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 15 ; <<16 x sbyte>> [#uses=1] - store <16 x sbyte> %tmp33, <16 x sbyte>* %A + %tmp = load <16 x i8>* %A ; <<16 x i8>> [#uses=8] + %tmp2 = load <16 x i8>* %B ; <<16 x i8>> [#uses=8] + %tmp.upgrd.12 = extractelement <16 x i8> %tmp, i32 8 ; <i8> [#uses=1] + %tmp3 = extractelement <16 x i8> %tmp2, i32 8 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp, i32 9 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp2, i32 9 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp, i32 10 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp2, i32 10 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp, i32 11 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp2, i32 11 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp, i32 12 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp2, i32 12 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp, i32 13 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp2, i32 13 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp, i32 14 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp2, i32 14 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp, i32 15 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp2, i32 15 ; <i8> [#uses=1] + %tmp18 = insertelement <16 x i8> undef, i8 %tmp.upgrd.12, i32 0 ; <<16 x i8>> [#uses=1] + %tmp19 = insertelement <16 x i8> %tmp18, i8 %tmp3, i32 1 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 2 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 3 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 4 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 5 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 6 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 7 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 8 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 9 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 10 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 11 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 12 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 13 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 14 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 15 ; <<16 x i8>> [#uses=1] + store <16 x i8> %tmp33, <16 x i8>* %A ret void } -void %th_l(<8 x short>* %A, <8 x short>* %B) { +define void @th_l(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=4] - %tmp2 = load <8 x short>* %B ; <<8 x short>> [#uses=4] - %tmp = extractelement <8 x short> %tmp, uint 4 ; <short> [#uses=1] - %tmp3 = extractelement <8 x short> %tmp2, uint 4 ; <short> [#uses=1] - %tmp4 = extractelement <8 x short> %tmp, uint 5 ; <short> [#uses=1] - %tmp5 = extractelement <8 x short> %tmp2, uint 5 ; <short> [#uses=1] - %tmp6 = extractelement <8 x short> %tmp, uint 6 ; <short> [#uses=1] - %tmp7 = extractelement <8 x short> %tmp2, uint 6 ; <short> [#uses=1] - %tmp8 = extractelement <8 x short> %tmp, uint 7 ; <short> [#uses=1] - %tmp9 = extractelement <8 x short> %tmp2, uint 7 ; <short> [#uses=1] - %tmp10 = insertelement <8 x short> undef, short %tmp, uint 0 ; <<8 x short>> [#uses=1] - %tmp11 = insertelement <8 x short> %tmp10, short %tmp3, uint 1 ; <<8 x short>> [#uses=1] - %tmp12 = insertelement <8 x short> %tmp11, short %tmp4, uint 2 ; <<8 x short>> [#uses=1] - %tmp13 = insertelement <8 x short> %tmp12, short %tmp5, uint 3 ; <<8 x short>> [#uses=1] - %tmp14 = insertelement <8 x short> %tmp13, short %tmp6, uint 4 ; <<8 x short>> [#uses=1] - %tmp15 = insertelement <8 x short> %tmp14, short %tmp7, uint 5 ; <<8 x short>> [#uses=1] - %tmp16 = insertelement <8 x short> %tmp15, short %tmp8, uint 6 ; <<8 x short>> [#uses=1] - %tmp17 = insertelement <8 x short> %tmp16, short %tmp9, uint 7 ; <<8 x short>> [#uses=1] - store <8 x short> %tmp17, <8 x short>* %A + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=4] + %tmp2 = load <8 x i16>* %B ; <<8 x i16>> [#uses=4] + %tmp.upgrd.13 = extractelement <8 x i16> %tmp, i32 4 ; <i16> [#uses=1] + %tmp3 = extractelement <8 x i16> %tmp2, i32 4 ; <i16> [#uses=1] + %tmp4 = extractelement <8 x i16> %tmp, i32 5 ; <i16> [#uses=1] + %tmp5 = extractelement <8 x i16> %tmp2, i32 5 ; <i16> [#uses=1] + %tmp6 = extractelement <8 x i16> %tmp, i32 6 ; <i16> [#uses=1] + %tmp7 = extractelement <8 x i16> %tmp2, i32 6 ; <i16> [#uses=1] + %tmp8 = extractelement <8 x i16> %tmp, i32 7 ; <i16> [#uses=1] + %tmp9 = extractelement <8 x i16> %tmp2, i32 7 ; <i16> [#uses=1] + %tmp10 = insertelement <8 x i16> undef, i16 %tmp.upgrd.13, i32 0 ; <<8 x i16>> [#uses=1] + %tmp11 = insertelement <8 x i16> %tmp10, i16 %tmp3, i32 1 ; <<8 x i16>> [#uses=1] + %tmp12 = insertelement <8 x i16> %tmp11, i16 %tmp4, i32 2 ; <<8 x i16>> [#uses=1] + %tmp13 = insertelement <8 x i16> %tmp12, i16 %tmp5, i32 3 ; <<8 x i16>> [#uses=1] + %tmp14 = insertelement <8 x i16> %tmp13, i16 %tmp6, i32 4 ; <<8 x i16>> [#uses=1] + %tmp15 = insertelement <8 x i16> %tmp14, i16 %tmp7, i32 5 ; <<8 x i16>> [#uses=1] + %tmp16 = insertelement <8 x i16> %tmp15, i16 %tmp8, i32 6 ; <<8 x i16>> [#uses=1] + %tmp17 = insertelement <8 x i16> %tmp16, i16 %tmp9, i32 7 ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp17, <8 x i16>* %A ret void } -void %tw_l(<4 x int>* %A, <4 x int>* %B) { +define void @tw_l(<4 x i32>* %A, <4 x i32>* %B) { entry: - %tmp = load <4 x int>* %A ; <<4 x int>> [#uses=2] - %tmp2 = load <4 x int>* %B ; <<4 x int>> [#uses=2] - %tmp = extractelement <4 x int> %tmp, uint 2 ; <int> [#uses=1] - %tmp3 = extractelement <4 x int> %tmp2, uint 2 ; <int> [#uses=1] - %tmp4 = extractelement <4 x int> %tmp, uint 3 ; <int> [#uses=1] - %tmp5 = extractelement <4 x int> %tmp2, uint 3 ; <int> [#uses=1] - %tmp6 = insertelement <4 x int> undef, int %tmp, uint 0 ; <<4 x int>> [#uses=1] - %tmp7 = insertelement <4 x int> %tmp6, int %tmp3, uint 1 ; <<4 x int>> [#uses=1] - %tmp8 = insertelement <4 x int> %tmp7, int %tmp4, uint 2 ; <<4 x int>> [#uses=1] - %tmp9 = insertelement <4 x int> %tmp8, int %tmp5, uint 3 ; <<4 x int>> [#uses=1] - store <4 x int> %tmp9, <4 x int>* %A + %tmp = load <4 x i32>* %A ; <<4 x i32>> [#uses=2] + %tmp2 = load <4 x i32>* %B ; <<4 x i32>> [#uses=2] + %tmp.upgrd.14 = extractelement <4 x i32> %tmp, i32 2 ; <i32> [#uses=1] + %tmp3 = extractelement <4 x i32> %tmp2, i32 2 ; <i32> [#uses=1] + %tmp4 = extractelement <4 x i32> %tmp, i32 3 ; <i32> [#uses=1] + %tmp5 = extractelement <4 x i32> %tmp2, i32 3 ; <i32> [#uses=1] + %tmp6 = insertelement <4 x i32> undef, i32 %tmp.upgrd.14, i32 0 ; <<4 x i32>> [#uses=1] + %tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 1 ; <<4 x i32>> [#uses=1] + %tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 2 ; <<4 x i32>> [#uses=1] + %tmp9 = insertelement <4 x i32> %tmp8, i32 %tmp5, i32 3 ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp9, <4 x i32>* %A ret void } -void %tb_h(<16 x sbyte>* %A, <16 x sbyte>* %B) { +define void @tb_h(<16 x i8>* %A, <16 x i8>* %B) { entry: - %tmp = load <16 x sbyte>* %A ; <<16 x sbyte>> [#uses=8] - %tmp2 = load <16 x sbyte>* %B ; <<16 x sbyte>> [#uses=8] - %tmp = extractelement <16 x sbyte> %tmp, uint 0 ; <sbyte> [#uses=1] - %tmp3 = extractelement <16 x sbyte> %tmp2, uint 0 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 1 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp2, uint 1 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 2 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp2, uint 2 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 3 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp2, uint 3 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 4 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp2, uint 4 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp, uint 5 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp2, uint 5 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp, uint 6 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp2, uint 6 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp, uint 7 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp2, uint 7 ; <sbyte> [#uses=1] - %tmp18 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp19 = insertelement <16 x sbyte> %tmp18, sbyte %tmp3, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 15 ; <<16 x sbyte>> [#uses=1] - store <16 x sbyte> %tmp33, <16 x sbyte>* %A + %tmp = load <16 x i8>* %A ; <<16 x i8>> [#uses=8] + %tmp2 = load <16 x i8>* %B ; <<16 x i8>> [#uses=8] + %tmp.upgrd.15 = extractelement <16 x i8> %tmp, i32 0 ; <i8> [#uses=1] + %tmp3 = extractelement <16 x i8> %tmp2, i32 0 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp, i32 1 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp2, i32 1 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp, i32 2 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp2, i32 2 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp, i32 3 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp2, i32 3 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp, i32 4 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp2, i32 4 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp, i32 5 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp2, i32 5 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp, i32 6 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp2, i32 6 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp, i32 7 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp2, i32 7 ; <i8> [#uses=1] + %tmp18 = insertelement <16 x i8> undef, i8 %tmp.upgrd.15, i32 0 ; <<16 x i8>> [#uses=1] + %tmp19 = insertelement <16 x i8> %tmp18, i8 %tmp3, i32 1 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 2 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 3 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 4 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 5 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 6 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 7 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 8 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 9 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 10 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 11 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 12 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 13 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 14 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 15 ; <<16 x i8>> [#uses=1] + store <16 x i8> %tmp33, <16 x i8>* %A ret void } -void %th_h(<8 x short>* %A, <8 x short>* %B) { +define void @th_h(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=4] - %tmp2 = load <8 x short>* %B ; <<8 x short>> [#uses=4] - %tmp = extractelement <8 x short> %tmp, uint 0 ; <short> [#uses=1] - %tmp3 = extractelement <8 x short> %tmp2, uint 0 ; <short> [#uses=1] - %tmp4 = extractelement <8 x short> %tmp, uint 1 ; <short> [#uses=1] - %tmp5 = extractelement <8 x short> %tmp2, uint 1 ; <short> [#uses=1] - %tmp6 = extractelement <8 x short> %tmp, uint 2 ; <short> [#uses=1] - %tmp7 = extractelement <8 x short> %tmp2, uint 2 ; <short> [#uses=1] - %tmp8 = extractelement <8 x short> %tmp, uint 3 ; <short> [#uses=1] - %tmp9 = extractelement <8 x short> %tmp2, uint 3 ; <short> [#uses=1] - %tmp10 = insertelement <8 x short> undef, short %tmp, uint 0 ; <<8 x short>> [#uses=1] - %tmp11 = insertelement <8 x short> %tmp10, short %tmp3, uint 1 ; <<8 x short>> [#uses=1] - %tmp12 = insertelement <8 x short> %tmp11, short %tmp4, uint 2 ; <<8 x short>> [#uses=1] - %tmp13 = insertelement <8 x short> %tmp12, short %tmp5, uint 3 ; <<8 x short>> [#uses=1] - %tmp14 = insertelement <8 x short> %tmp13, short %tmp6, uint 4 ; <<8 x short>> [#uses=1] - %tmp15 = insertelement <8 x short> %tmp14, short %tmp7, uint 5 ; <<8 x short>> [#uses=1] - %tmp16 = insertelement <8 x short> %tmp15, short %tmp8, uint 6 ; <<8 x short>> [#uses=1] - %tmp17 = insertelement <8 x short> %tmp16, short %tmp9, uint 7 ; <<8 x short>> [#uses=1] - store <8 x short> %tmp17, <8 x short>* %A + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=4] + %tmp2 = load <8 x i16>* %B ; <<8 x i16>> [#uses=4] + %tmp.upgrd.16 = extractelement <8 x i16> %tmp, i32 0 ; <i16> [#uses=1] + %tmp3 = extractelement <8 x i16> %tmp2, i32 0 ; <i16> [#uses=1] + %tmp4 = extractelement <8 x i16> %tmp, i32 1 ; <i16> [#uses=1] + %tmp5 = extractelement <8 x i16> %tmp2, i32 1 ; <i16> [#uses=1] + %tmp6 = extractelement <8 x i16> %tmp, i32 2 ; <i16> [#uses=1] + %tmp7 = extractelement <8 x i16> %tmp2, i32 2 ; <i16> [#uses=1] + %tmp8 = extractelement <8 x i16> %tmp, i32 3 ; <i16> [#uses=1] + %tmp9 = extractelement <8 x i16> %tmp2, i32 3 ; <i16> [#uses=1] + %tmp10 = insertelement <8 x i16> undef, i16 %tmp.upgrd.16, i32 0 ; <<8 x i16>> [#uses=1] + %tmp11 = insertelement <8 x i16> %tmp10, i16 %tmp3, i32 1 ; <<8 x i16>> [#uses=1] + %tmp12 = insertelement <8 x i16> %tmp11, i16 %tmp4, i32 2 ; <<8 x i16>> [#uses=1] + %tmp13 = insertelement <8 x i16> %tmp12, i16 %tmp5, i32 3 ; <<8 x i16>> [#uses=1] + %tmp14 = insertelement <8 x i16> %tmp13, i16 %tmp6, i32 4 ; <<8 x i16>> [#uses=1] + %tmp15 = insertelement <8 x i16> %tmp14, i16 %tmp7, i32 5 ; <<8 x i16>> [#uses=1] + %tmp16 = insertelement <8 x i16> %tmp15, i16 %tmp8, i32 6 ; <<8 x i16>> [#uses=1] + %tmp17 = insertelement <8 x i16> %tmp16, i16 %tmp9, i32 7 ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp17, <8 x i16>* %A ret void } -void %tw_h(<4 x int>* %A, <4 x int>* %B) { +define void @tw_h(<4 x i32>* %A, <4 x i32>* %B) { entry: - %tmp = load <4 x int>* %A ; <<4 x int>> [#uses=2] - %tmp2 = load <4 x int>* %B ; <<4 x int>> [#uses=2] - %tmp = extractelement <4 x int> %tmp2, uint 0 ; <int> [#uses=1] - %tmp3 = extractelement <4 x int> %tmp, uint 0 ; <int> [#uses=1] - %tmp4 = extractelement <4 x int> %tmp2, uint 1 ; <int> [#uses=1] - %tmp5 = extractelement <4 x int> %tmp, uint 1 ; <int> [#uses=1] - %tmp6 = insertelement <4 x int> undef, int %tmp, uint 0 ; <<4 x int>> [#uses=1] - %tmp7 = insertelement <4 x int> %tmp6, int %tmp3, uint 1 ; <<4 x int>> [#uses=1] - %tmp8 = insertelement <4 x int> %tmp7, int %tmp4, uint 2 ; <<4 x int>> [#uses=1] - %tmp9 = insertelement <4 x int> %tmp8, int %tmp5, uint 3 ; <<4 x int>> [#uses=1] - store <4 x int> %tmp9, <4 x int>* %A + %tmp = load <4 x i32>* %A ; <<4 x i32>> [#uses=2] + %tmp2 = load <4 x i32>* %B ; <<4 x i32>> [#uses=2] + %tmp.upgrd.17 = extractelement <4 x i32> %tmp2, i32 0 ; <i32> [#uses=1] + %tmp3 = extractelement <4 x i32> %tmp, i32 0 ; <i32> [#uses=1] + %tmp4 = extractelement <4 x i32> %tmp2, i32 1 ; <i32> [#uses=1] + %tmp5 = extractelement <4 x i32> %tmp, i32 1 ; <i32> [#uses=1] + %tmp6 = insertelement <4 x i32> undef, i32 %tmp.upgrd.17, i32 0 ; <<4 x i32>> [#uses=1] + %tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 1 ; <<4 x i32>> [#uses=1] + %tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 2 ; <<4 x i32>> [#uses=1] + %tmp9 = insertelement <4 x i32> %tmp8, i32 %tmp5, i32 3 ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp9, <4 x i32>* %A ret void } -void %tw_h_flop(<4 x int>* %A, <4 x int>* %B) { - %tmp = load <4 x int>* %A ; <<4 x int>> [#uses=2] - %tmp2 = load <4 x int>* %B ; <<4 x int>> [#uses=2] - %tmp = extractelement <4 x int> %tmp, uint 0 ; <int> [#uses=1] - %tmp3 = extractelement <4 x int> %tmp2, uint 0 ; <int> [#uses=1] - %tmp4 = extractelement <4 x int> %tmp, uint 1 ; <int> [#uses=1] - %tmp5 = extractelement <4 x int> %tmp2, uint 1 ; <int> [#uses=1] - %tmp6 = insertelement <4 x int> undef, int %tmp, uint 0 ; <<4 x int>> [#uses=1] - %tmp7 = insertelement <4 x int> %tmp6, int %tmp3, uint 1 ; <<4 x int>> [#uses=1] - %tmp8 = insertelement <4 x int> %tmp7, int %tmp4, uint 2 ; <<4 x int>> [#uses=1] - %tmp9 = insertelement <4 x int> %tmp8, int %tmp5, uint 3 ; <<4 x int>> [#uses=1] - store <4 x int> %tmp9, <4 x int>* %A +define void @tw_h_flop(<4 x i32>* %A, <4 x i32>* %B) { + %tmp = load <4 x i32>* %A ; <<4 x i32>> [#uses=2] + %tmp2 = load <4 x i32>* %B ; <<4 x i32>> [#uses=2] + %tmp.upgrd.18 = extractelement <4 x i32> %tmp, i32 0 ; <i32> [#uses=1] + %tmp3 = extractelement <4 x i32> %tmp2, i32 0 ; <i32> [#uses=1] + %tmp4 = extractelement <4 x i32> %tmp, i32 1 ; <i32> [#uses=1] + %tmp5 = extractelement <4 x i32> %tmp2, i32 1 ; <i32> [#uses=1] + %tmp6 = insertelement <4 x i32> undef, i32 %tmp.upgrd.18, i32 0 ; <<4 x i32>> [#uses=1] + %tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 1 ; <<4 x i32>> [#uses=1] + %tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 2 ; <<4 x i32>> [#uses=1] + %tmp9 = insertelement <4 x i32> %tmp8, i32 %tmp5, i32 3 ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp9, <4 x i32>* %A ret void } - -void %VMRG_UNARY_tb_l(<16 x sbyte>* %A, <16 x sbyte>* %B) { +define void @VMRG_UNARY_tb_l(<16 x i8>* %A, <16 x i8>* %B) { entry: - %tmp = load <16 x sbyte>* %A ; <<16 x sbyte>> [#uses=16] - %tmp = extractelement <16 x sbyte> %tmp, uint 8 ; <sbyte> [#uses=1] - %tmp3 = extractelement <16 x sbyte> %tmp, uint 8 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 9 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp, uint 9 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 10 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp, uint 10 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 11 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp, uint 11 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 12 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp, uint 12 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp, uint 13 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp, uint 13 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp, uint 14 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp, uint 14 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp, uint 15 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp, uint 15 ; <sbyte> [#uses=1] - %tmp18 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp19 = insertelement <16 x sbyte> %tmp18, sbyte %tmp3, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 15 ; <<16 x sbyte>> [#uses=1] - store <16 x sbyte> %tmp33, <16 x sbyte>* %A + %tmp = load <16 x i8>* %A ; <<16 x i8>> [#uses=16] + %tmp.upgrd.19 = extractelement <16 x i8> %tmp, i32 8 ; <i8> [#uses=1] + %tmp3 = extractelement <16 x i8> %tmp, i32 8 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp, i32 9 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp, i32 9 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp, i32 10 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp, i32 10 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp, i32 11 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp, i32 11 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp, i32 12 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp, i32 12 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp, i32 13 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp, i32 13 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp, i32 14 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp, i32 14 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp, i32 15 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp, i32 15 ; <i8> [#uses=1] + %tmp18 = insertelement <16 x i8> undef, i8 %tmp.upgrd.19, i32 0 ; <<16 x i8>> [#uses=1] + %tmp19 = insertelement <16 x i8> %tmp18, i8 %tmp3, i32 1 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 2 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 3 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 4 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 5 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 6 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 7 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 8 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 9 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 10 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 11 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 12 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 13 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 14 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 15 ; <<16 x i8>> [#uses=1] + store <16 x i8> %tmp33, <16 x i8>* %A ret void } -void %VMRG_UNARY_th_l(<8 x short>* %A, <8 x short>* %B) { +define void @VMRG_UNARY_th_l(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=8] - %tmp = extractelement <8 x short> %tmp, uint 4 ; <short> [#uses=1] - %tmp3 = extractelement <8 x short> %tmp, uint 4 ; <short> [#uses=1] - %tmp4 = extractelement <8 x short> %tmp, uint 5 ; <short> [#uses=1] - %tmp5 = extractelement <8 x short> %tmp, uint 5 ; <short> [#uses=1] - %tmp6 = extractelement <8 x short> %tmp, uint 6 ; <short> [#uses=1] - %tmp7 = extractelement <8 x short> %tmp, uint 6 ; <short> [#uses=1] - %tmp8 = extractelement <8 x short> %tmp, uint 7 ; <short> [#uses=1] - %tmp9 = extractelement <8 x short> %tmp, uint 7 ; <short> [#uses=1] - %tmp10 = insertelement <8 x short> undef, short %tmp, uint 0 ; <<8 x short>> [#uses=1] - %tmp11 = insertelement <8 x short> %tmp10, short %tmp3, uint 1 ; <<8 x short>> [#uses=1] - %tmp12 = insertelement <8 x short> %tmp11, short %tmp4, uint 2 ; <<8 x short>> [#uses=1] - %tmp13 = insertelement <8 x short> %tmp12, short %tmp5, uint 3 ; <<8 x short>> [#uses=1] - %tmp14 = insertelement <8 x short> %tmp13, short %tmp6, uint 4 ; <<8 x short>> [#uses=1] - %tmp15 = insertelement <8 x short> %tmp14, short %tmp7, uint 5 ; <<8 x short>> [#uses=1] - %tmp16 = insertelement <8 x short> %tmp15, short %tmp8, uint 6 ; <<8 x short>> [#uses=1] - %tmp17 = insertelement <8 x short> %tmp16, short %tmp9, uint 7 ; <<8 x short>> [#uses=1] - store <8 x short> %tmp17, <8 x short>* %A + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=8] + %tmp.upgrd.20 = extractelement <8 x i16> %tmp, i32 4 ; <i16> [#uses=1] + %tmp3 = extractelement <8 x i16> %tmp, i32 4 ; <i16> [#uses=1] + %tmp4 = extractelement <8 x i16> %tmp, i32 5 ; <i16> [#uses=1] + %tmp5 = extractelement <8 x i16> %tmp, i32 5 ; <i16> [#uses=1] + %tmp6 = extractelement <8 x i16> %tmp, i32 6 ; <i16> [#uses=1] + %tmp7 = extractelement <8 x i16> %tmp, i32 6 ; <i16> [#uses=1] + %tmp8 = extractelement <8 x i16> %tmp, i32 7 ; <i16> [#uses=1] + %tmp9 = extractelement <8 x i16> %tmp, i32 7 ; <i16> [#uses=1] + %tmp10 = insertelement <8 x i16> undef, i16 %tmp.upgrd.20, i32 0 ; <<8 x i16>> [#uses=1] + %tmp11 = insertelement <8 x i16> %tmp10, i16 %tmp3, i32 1 ; <<8 x i16>> [#uses=1] + %tmp12 = insertelement <8 x i16> %tmp11, i16 %tmp4, i32 2 ; <<8 x i16>> [#uses=1] + %tmp13 = insertelement <8 x i16> %tmp12, i16 %tmp5, i32 3 ; <<8 x i16>> [#uses=1] + %tmp14 = insertelement <8 x i16> %tmp13, i16 %tmp6, i32 4 ; <<8 x i16>> [#uses=1] + %tmp15 = insertelement <8 x i16> %tmp14, i16 %tmp7, i32 5 ; <<8 x i16>> [#uses=1] + %tmp16 = insertelement <8 x i16> %tmp15, i16 %tmp8, i32 6 ; <<8 x i16>> [#uses=1] + %tmp17 = insertelement <8 x i16> %tmp16, i16 %tmp9, i32 7 ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp17, <8 x i16>* %A ret void } -void %VMRG_UNARY_tw_l(<4 x int>* %A, <4 x int>* %B) { +define void @VMRG_UNARY_tw_l(<4 x i32>* %A, <4 x i32>* %B) { entry: - %tmp = load <4 x int>* %A ; <<4 x int>> [#uses=4] - %tmp = extractelement <4 x int> %tmp, uint 2 ; <int> [#uses=1] - %tmp3 = extractelement <4 x int> %tmp, uint 2 ; <int> [#uses=1] - %tmp4 = extractelement <4 x int> %tmp, uint 3 ; <int> [#uses=1] - %tmp5 = extractelement <4 x int> %tmp, uint 3 ; <int> [#uses=1] - %tmp6 = insertelement <4 x int> undef, int %tmp, uint 0 ; <<4 x int>> [#uses=1] - %tmp7 = insertelement <4 x int> %tmp6, int %tmp3, uint 1 ; <<4 x int>> [#uses=1] - %tmp8 = insertelement <4 x int> %tmp7, int %tmp4, uint 2 ; <<4 x int>> [#uses=1] - %tmp9 = insertelement <4 x int> %tmp8, int %tmp5, uint 3 ; <<4 x int>> [#uses=1] - store <4 x int> %tmp9, <4 x int>* %A + %tmp = load <4 x i32>* %A ; <<4 x i32>> [#uses=4] + %tmp.upgrd.21 = extractelement <4 x i32> %tmp, i32 2 ; <i32> [#uses=1] + %tmp3 = extractelement <4 x i32> %tmp, i32 2 ; <i32> [#uses=1] + %tmp4 = extractelement <4 x i32> %tmp, i32 3 ; <i32> [#uses=1] + %tmp5 = extractelement <4 x i32> %tmp, i32 3 ; <i32> [#uses=1] + %tmp6 = insertelement <4 x i32> undef, i32 %tmp.upgrd.21, i32 0 ; <<4 x i32>> [#uses=1] + %tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 1 ; <<4 x i32>> [#uses=1] + %tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 2 ; <<4 x i32>> [#uses=1] + %tmp9 = insertelement <4 x i32> %tmp8, i32 %tmp5, i32 3 ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp9, <4 x i32>* %A ret void } -void %VMRG_UNARY_tb_h(<16 x sbyte>* %A, <16 x sbyte>* %B) { +define void @VMRG_UNARY_tb_h(<16 x i8>* %A, <16 x i8>* %B) { entry: - %tmp = load <16 x sbyte>* %A ; <<16 x sbyte>> [#uses=16] - %tmp = extractelement <16 x sbyte> %tmp, uint 0 ; <sbyte> [#uses=1] - %tmp3 = extractelement <16 x sbyte> %tmp, uint 0 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 1 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp, uint 1 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 2 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp, uint 2 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 3 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp, uint 3 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 4 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp, uint 4 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp, uint 5 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp, uint 5 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp, uint 6 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp, uint 6 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp, uint 7 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp, uint 7 ; <sbyte> [#uses=1] - %tmp18 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp19 = insertelement <16 x sbyte> %tmp18, sbyte %tmp3, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 15 ; <<16 x sbyte>> [#uses=1] - store <16 x sbyte> %tmp33, <16 x sbyte>* %A + %tmp = load <16 x i8>* %A ; <<16 x i8>> [#uses=16] + %tmp.upgrd.22 = extractelement <16 x i8> %tmp, i32 0 ; <i8> [#uses=1] + %tmp3 = extractelement <16 x i8> %tmp, i32 0 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp, i32 1 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp, i32 1 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp, i32 2 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp, i32 2 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp, i32 3 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp, i32 3 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp, i32 4 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp, i32 4 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp, i32 5 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp, i32 5 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp, i32 6 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp, i32 6 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp, i32 7 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp, i32 7 ; <i8> [#uses=1] + %tmp18 = insertelement <16 x i8> undef, i8 %tmp.upgrd.22, i32 0 ; <<16 x i8>> [#uses=1] + %tmp19 = insertelement <16 x i8> %tmp18, i8 %tmp3, i32 1 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 2 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 3 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 4 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 5 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 6 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 7 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 8 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 9 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 10 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 11 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 12 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 13 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 14 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 15 ; <<16 x i8>> [#uses=1] + store <16 x i8> %tmp33, <16 x i8>* %A ret void } -void %VMRG_UNARY_th_h(<8 x short>* %A, <8 x short>* %B) { +define void @VMRG_UNARY_th_h(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=8] - %tmp = extractelement <8 x short> %tmp, uint 0 ; <short> [#uses=1] - %tmp3 = extractelement <8 x short> %tmp, uint 0 ; <short> [#uses=1] - %tmp4 = extractelement <8 x short> %tmp, uint 1 ; <short> [#uses=1] - %tmp5 = extractelement <8 x short> %tmp, uint 1 ; <short> [#uses=1] - %tmp6 = extractelement <8 x short> %tmp, uint 2 ; <short> [#uses=1] - %tmp7 = extractelement <8 x short> %tmp, uint 2 ; <short> [#uses=1] - %tmp8 = extractelement <8 x short> %tmp, uint 3 ; <short> [#uses=1] - %tmp9 = extractelement <8 x short> %tmp, uint 3 ; <short> [#uses=1] - %tmp10 = insertelement <8 x short> undef, short %tmp, uint 0 ; <<8 x short>> [#uses=1] - %tmp11 = insertelement <8 x short> %tmp10, short %tmp3, uint 1 ; <<8 x short>> [#uses=1] - %tmp12 = insertelement <8 x short> %tmp11, short %tmp4, uint 2 ; <<8 x short>> [#uses=1] - %tmp13 = insertelement <8 x short> %tmp12, short %tmp5, uint 3 ; <<8 x short>> [#uses=1] - %tmp14 = insertelement <8 x short> %tmp13, short %tmp6, uint 4 ; <<8 x short>> [#uses=1] - %tmp15 = insertelement <8 x short> %tmp14, short %tmp7, uint 5 ; <<8 x short>> [#uses=1] - %tmp16 = insertelement <8 x short> %tmp15, short %tmp8, uint 6 ; <<8 x short>> [#uses=1] - %tmp17 = insertelement <8 x short> %tmp16, short %tmp9, uint 7 ; <<8 x short>> [#uses=1] - store <8 x short> %tmp17, <8 x short>* %A + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=8] + %tmp.upgrd.23 = extractelement <8 x i16> %tmp, i32 0 ; <i16> [#uses=1] + %tmp3 = extractelement <8 x i16> %tmp, i32 0 ; <i16> [#uses=1] + %tmp4 = extractelement <8 x i16> %tmp, i32 1 ; <i16> [#uses=1] + %tmp5 = extractelement <8 x i16> %tmp, i32 1 ; <i16> [#uses=1] + %tmp6 = extractelement <8 x i16> %tmp, i32 2 ; <i16> [#uses=1] + %tmp7 = extractelement <8 x i16> %tmp, i32 2 ; <i16> [#uses=1] + %tmp8 = extractelement <8 x i16> %tmp, i32 3 ; <i16> [#uses=1] + %tmp9 = extractelement <8 x i16> %tmp, i32 3 ; <i16> [#uses=1] + %tmp10 = insertelement <8 x i16> undef, i16 %tmp.upgrd.23, i32 0 ; <<8 x i16>> [#uses=1] + %tmp11 = insertelement <8 x i16> %tmp10, i16 %tmp3, i32 1 ; <<8 x i16>> [#uses=1] + %tmp12 = insertelement <8 x i16> %tmp11, i16 %tmp4, i32 2 ; <<8 x i16>> [#uses=1] + %tmp13 = insertelement <8 x i16> %tmp12, i16 %tmp5, i32 3 ; <<8 x i16>> [#uses=1] + %tmp14 = insertelement <8 x i16> %tmp13, i16 %tmp6, i32 4 ; <<8 x i16>> [#uses=1] + %tmp15 = insertelement <8 x i16> %tmp14, i16 %tmp7, i32 5 ; <<8 x i16>> [#uses=1] + %tmp16 = insertelement <8 x i16> %tmp15, i16 %tmp8, i32 6 ; <<8 x i16>> [#uses=1] + %tmp17 = insertelement <8 x i16> %tmp16, i16 %tmp9, i32 7 ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp17, <8 x i16>* %A ret void } -void %VMRG_UNARY_tw_h(<4 x int>* %A, <4 x int>* %B) { +define void @VMRG_UNARY_tw_h(<4 x i32>* %A, <4 x i32>* %B) { entry: - %tmp = load <4 x int>* %A ; <<4 x int>> [#uses=4] - %tmp = extractelement <4 x int> %tmp, uint 0 ; <int> [#uses=1] - %tmp3 = extractelement <4 x int> %tmp, uint 0 ; <int> [#uses=1] - %tmp4 = extractelement <4 x int> %tmp, uint 1 ; <int> [#uses=1] - %tmp5 = extractelement <4 x int> %tmp, uint 1 ; <int> [#uses=1] - %tmp6 = insertelement <4 x int> undef, int %tmp, uint 0 ; <<4 x int>> [#uses=1] - %tmp7 = insertelement <4 x int> %tmp6, int %tmp3, uint 1 ; <<4 x int>> [#uses=1] - %tmp8 = insertelement <4 x int> %tmp7, int %tmp4, uint 2 ; <<4 x int>> [#uses=1] - %tmp9 = insertelement <4 x int> %tmp8, int %tmp5, uint 3 ; <<4 x int>> [#uses=1] - store <4 x int> %tmp9, <4 x int>* %A + %tmp = load <4 x i32>* %A ; <<4 x i32>> [#uses=4] + %tmp.upgrd.24 = extractelement <4 x i32> %tmp, i32 0 ; <i32> [#uses=1] + %tmp3 = extractelement <4 x i32> %tmp, i32 0 ; <i32> [#uses=1] + %tmp4 = extractelement <4 x i32> %tmp, i32 1 ; <i32> [#uses=1] + %tmp5 = extractelement <4 x i32> %tmp, i32 1 ; <i32> [#uses=1] + %tmp6 = insertelement <4 x i32> undef, i32 %tmp.upgrd.24, i32 0 ; <<4 x i32>> [#uses=1] + %tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 1 ; <<4 x i32>> [#uses=1] + %tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 2 ; <<4 x i32>> [#uses=1] + %tmp9 = insertelement <4 x i32> %tmp8, i32 %tmp5, i32 3 ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp9, <4 x i32>* %A ret void } -void %VPCKUHUM_unary(<8 x short>* %A, <8 x short>* %B) { +define void @VPCKUHUM_unary(<8 x i16>* %A, <8 x i16>* %B) { entry: - %tmp = load <8 x short>* %A ; <<8 x short>> [#uses=2] - %tmp = cast <8 x short> %tmp to <16 x sbyte> ; <<16 x sbyte>> [#uses=8] - %tmp3 = cast <8 x short> %tmp to <16 x sbyte> ; <<16 x sbyte>> [#uses=8] - %tmp = extractelement <16 x sbyte> %tmp, uint 1 ; <sbyte> [#uses=1] - %tmp4 = extractelement <16 x sbyte> %tmp, uint 3 ; <sbyte> [#uses=1] - %tmp5 = extractelement <16 x sbyte> %tmp, uint 5 ; <sbyte> [#uses=1] - %tmp6 = extractelement <16 x sbyte> %tmp, uint 7 ; <sbyte> [#uses=1] - %tmp7 = extractelement <16 x sbyte> %tmp, uint 9 ; <sbyte> [#uses=1] - %tmp8 = extractelement <16 x sbyte> %tmp, uint 11 ; <sbyte> [#uses=1] - %tmp9 = extractelement <16 x sbyte> %tmp, uint 13 ; <sbyte> [#uses=1] - %tmp10 = extractelement <16 x sbyte> %tmp, uint 15 ; <sbyte> [#uses=1] - %tmp11 = extractelement <16 x sbyte> %tmp3, uint 1 ; <sbyte> [#uses=1] - %tmp12 = extractelement <16 x sbyte> %tmp3, uint 3 ; <sbyte> [#uses=1] - %tmp13 = extractelement <16 x sbyte> %tmp3, uint 5 ; <sbyte> [#uses=1] - %tmp14 = extractelement <16 x sbyte> %tmp3, uint 7 ; <sbyte> [#uses=1] - %tmp15 = extractelement <16 x sbyte> %tmp3, uint 9 ; <sbyte> [#uses=1] - %tmp16 = extractelement <16 x sbyte> %tmp3, uint 11 ; <sbyte> [#uses=1] - %tmp17 = extractelement <16 x sbyte> %tmp3, uint 13 ; <sbyte> [#uses=1] - %tmp18 = extractelement <16 x sbyte> %tmp3, uint 15 ; <sbyte> [#uses=1] - %tmp19 = insertelement <16 x sbyte> undef, sbyte %tmp, uint 0 ; <<16 x sbyte>> [#uses=1] - %tmp20 = insertelement <16 x sbyte> %tmp19, sbyte %tmp4, uint 1 ; <<16 x sbyte>> [#uses=1] - %tmp21 = insertelement <16 x sbyte> %tmp20, sbyte %tmp5, uint 2 ; <<16 x sbyte>> [#uses=1] - %tmp22 = insertelement <16 x sbyte> %tmp21, sbyte %tmp6, uint 3 ; <<16 x sbyte>> [#uses=1] - %tmp23 = insertelement <16 x sbyte> %tmp22, sbyte %tmp7, uint 4 ; <<16 x sbyte>> [#uses=1] - %tmp24 = insertelement <16 x sbyte> %tmp23, sbyte %tmp8, uint 5 ; <<16 x sbyte>> [#uses=1] - %tmp25 = insertelement <16 x sbyte> %tmp24, sbyte %tmp9, uint 6 ; <<16 x sbyte>> [#uses=1] - %tmp26 = insertelement <16 x sbyte> %tmp25, sbyte %tmp10, uint 7 ; <<16 x sbyte>> [#uses=1] - %tmp27 = insertelement <16 x sbyte> %tmp26, sbyte %tmp11, uint 8 ; <<16 x sbyte>> [#uses=1] - %tmp28 = insertelement <16 x sbyte> %tmp27, sbyte %tmp12, uint 9 ; <<16 x sbyte>> [#uses=1] - %tmp29 = insertelement <16 x sbyte> %tmp28, sbyte %tmp13, uint 10 ; <<16 x sbyte>> [#uses=1] - %tmp30 = insertelement <16 x sbyte> %tmp29, sbyte %tmp14, uint 11 ; <<16 x sbyte>> [#uses=1] - %tmp31 = insertelement <16 x sbyte> %tmp30, sbyte %tmp15, uint 12 ; <<16 x sbyte>> [#uses=1] - %tmp32 = insertelement <16 x sbyte> %tmp31, sbyte %tmp16, uint 13 ; <<16 x sbyte>> [#uses=1] - %tmp33 = insertelement <16 x sbyte> %tmp32, sbyte %tmp17, uint 14 ; <<16 x sbyte>> [#uses=1] - %tmp34 = insertelement <16 x sbyte> %tmp33, sbyte %tmp18, uint 15 ; <<16 x sbyte>> [#uses=1] - %tmp34 = cast <16 x sbyte> %tmp34 to <8 x short> ; <<8 x short>> [#uses=1] - store <8 x short> %tmp34, <8 x short>* %A + %tmp = load <8 x i16>* %A ; <<8 x i16>> [#uses=2] + %tmp.upgrd.25 = bitcast <8 x i16> %tmp to <16 x i8> ; <<16 x i8>> [#uses=8] + %tmp3 = bitcast <8 x i16> %tmp to <16 x i8> ; <<16 x i8>> [#uses=8] + %tmp.upgrd.26 = extractelement <16 x i8> %tmp.upgrd.25, i32 1 ; <i8> [#uses=1] + %tmp4 = extractelement <16 x i8> %tmp.upgrd.25, i32 3 ; <i8> [#uses=1] + %tmp5 = extractelement <16 x i8> %tmp.upgrd.25, i32 5 ; <i8> [#uses=1] + %tmp6 = extractelement <16 x i8> %tmp.upgrd.25, i32 7 ; <i8> [#uses=1] + %tmp7 = extractelement <16 x i8> %tmp.upgrd.25, i32 9 ; <i8> [#uses=1] + %tmp8 = extractelement <16 x i8> %tmp.upgrd.25, i32 11 ; <i8> [#uses=1] + %tmp9 = extractelement <16 x i8> %tmp.upgrd.25, i32 13 ; <i8> [#uses=1] + %tmp10 = extractelement <16 x i8> %tmp.upgrd.25, i32 15 ; <i8> [#uses=1] + %tmp11 = extractelement <16 x i8> %tmp3, i32 1 ; <i8> [#uses=1] + %tmp12 = extractelement <16 x i8> %tmp3, i32 3 ; <i8> [#uses=1] + %tmp13 = extractelement <16 x i8> %tmp3, i32 5 ; <i8> [#uses=1] + %tmp14 = extractelement <16 x i8> %tmp3, i32 7 ; <i8> [#uses=1] + %tmp15 = extractelement <16 x i8> %tmp3, i32 9 ; <i8> [#uses=1] + %tmp16 = extractelement <16 x i8> %tmp3, i32 11 ; <i8> [#uses=1] + %tmp17 = extractelement <16 x i8> %tmp3, i32 13 ; <i8> [#uses=1] + %tmp18 = extractelement <16 x i8> %tmp3, i32 15 ; <i8> [#uses=1] + %tmp19 = insertelement <16 x i8> undef, i8 %tmp.upgrd.26, i32 0 ; <<16 x i8>> [#uses=1] + %tmp20 = insertelement <16 x i8> %tmp19, i8 %tmp4, i32 1 ; <<16 x i8>> [#uses=1] + %tmp21 = insertelement <16 x i8> %tmp20, i8 %tmp5, i32 2 ; <<16 x i8>> [#uses=1] + %tmp22 = insertelement <16 x i8> %tmp21, i8 %tmp6, i32 3 ; <<16 x i8>> [#uses=1] + %tmp23 = insertelement <16 x i8> %tmp22, i8 %tmp7, i32 4 ; <<16 x i8>> [#uses=1] + %tmp24 = insertelement <16 x i8> %tmp23, i8 %tmp8, i32 5 ; <<16 x i8>> [#uses=1] + %tmp25 = insertelement <16 x i8> %tmp24, i8 %tmp9, i32 6 ; <<16 x i8>> [#uses=1] + %tmp26 = insertelement <16 x i8> %tmp25, i8 %tmp10, i32 7 ; <<16 x i8>> [#uses=1] + %tmp27 = insertelement <16 x i8> %tmp26, i8 %tmp11, i32 8 ; <<16 x i8>> [#uses=1] + %tmp28 = insertelement <16 x i8> %tmp27, i8 %tmp12, i32 9 ; <<16 x i8>> [#uses=1] + %tmp29 = insertelement <16 x i8> %tmp28, i8 %tmp13, i32 10 ; <<16 x i8>> [#uses=1] + %tmp30 = insertelement <16 x i8> %tmp29, i8 %tmp14, i32 11 ; <<16 x i8>> [#uses=1] + %tmp31 = insertelement <16 x i8> %tmp30, i8 %tmp15, i32 12 ; <<16 x i8>> [#uses=1] + %tmp32 = insertelement <16 x i8> %tmp31, i8 %tmp16, i32 13 ; <<16 x i8>> [#uses=1] + %tmp33 = insertelement <16 x i8> %tmp32, i8 %tmp17, i32 14 ; <<16 x i8>> [#uses=1] + %tmp34 = insertelement <16 x i8> %tmp33, i8 %tmp18, i32 15 ; <<16 x i8>> [#uses=1] + %tmp34.upgrd.27 = bitcast <16 x i8> %tmp34 to <8 x i16> ; <<8 x i16>> [#uses=1] + store <8 x i16> %tmp34.upgrd.27, <8 x i16>* %A ret void } -void %VPCKUWUM_unary(<4 x int>* %A, <4 x int>* %B) { +define void @VPCKUWUM_unary(<4 x i32>* %A, <4 x i32>* %B) { entry: - %tmp = load <4 x int>* %A ; <<4 x int>> [#uses=2] - %tmp = cast <4 x int> %tmp to <8 x short> ; <<8 x short>> [#uses=4] - %tmp3 = cast <4 x int> %tmp to <8 x short> ; <<8 x short>> [#uses=4] - %tmp = extractelement <8 x short> %tmp, uint 1 ; <short> [#uses=1] - %tmp4 = extractelement <8 x short> %tmp, uint 3 ; <short> [#uses=1] - %tmp5 = extractelement <8 x short> %tmp, uint 5 ; <short> [#uses=1] - %tmp6 = extractelement <8 x short> %tmp, uint 7 ; <short> [#uses=1] - %tmp7 = extractelement <8 x short> %tmp3, uint 1 ; <short> [#uses=1] - %tmp8 = extractelement <8 x short> %tmp3, uint 3 ; <short> [#uses=1] - %tmp9 = extractelement <8 x short> %tmp3, uint 5 ; <short> [#uses=1] - %tmp10 = extractelement <8 x short> %tmp3, uint 7 ; <short> [#uses=1] - %tmp11 = insertelement <8 x short> undef, short %tmp, uint 0 ; <<8 x short>> [#uses=1] - %tmp12 = insertelement <8 x short> %tmp11, short %tmp4, uint 1 ; <<8 x short>> [#uses=1] - %tmp13 = insertelement <8 x short> %tmp12, short %tmp5, uint 2 ; <<8 x short>> [#uses=1] - %tmp14 = insertelement <8 x short> %tmp13, short %tmp6, uint 3 ; <<8 x short>> [#uses=1] - %tmp15 = insertelement <8 x short> %tmp14, short %tmp7, uint 4 ; <<8 x short>> [#uses=1] - %tmp16 = insertelement <8 x short> %tmp15, short %tmp8, uint 5 ; <<8 x short>> [#uses=1] - %tmp17 = insertelement <8 x short> %tmp16, short %tmp9, uint 6 ; <<8 x short>> [#uses=1] - %tmp18 = insertelement <8 x short> %tmp17, short %tmp10, uint 7 ; <<8 x short>> [#uses=1] - %tmp18 = cast <8 x short> %tmp18 to <4 x int> ; <<4 x int>> [#uses=1] - store <4 x int> %tmp18, <4 x int>* %A + %tmp = load <4 x i32>* %A ; <<4 x i32>> [#uses=2] + %tmp.upgrd.28 = bitcast <4 x i32> %tmp to <8 x i16> ; <<8 x i16>> [#uses=4] + %tmp3 = bitcast <4 x i32> %tmp to <8 x i16> ; <<8 x i16>> [#uses=4] + %tmp.upgrd.29 = extractelement <8 x i16> %tmp.upgrd.28, i32 1 ; <i16> [#uses=1] + %tmp4 = extractelement <8 x i16> %tmp.upgrd.28, i32 3 ; <i16> [#uses=1] + %tmp5 = extractelement <8 x i16> %tmp.upgrd.28, i32 5 ; <i16> [#uses=1] + %tmp6 = extractelement <8 x i16> %tmp.upgrd.28, i32 7 ; <i16> [#uses=1] + %tmp7 = extractelement <8 x i16> %tmp3, i32 1 ; <i16> [#uses=1] + %tmp8 = extractelement <8 x i16> %tmp3, i32 3 ; <i16> [#uses=1] + %tmp9 = extractelement <8 x i16> %tmp3, i32 5 ; <i16> [#uses=1] + %tmp10 = extractelement <8 x i16> %tmp3, i32 7 ; <i16> [#uses=1] + %tmp11 = insertelement <8 x i16> undef, i16 %tmp.upgrd.29, i32 0 ; <<8 x i16>> [#uses=1] + %tmp12 = insertelement <8 x i16> %tmp11, i16 %tmp4, i32 1 ; <<8 x i16>> [#uses=1] + %tmp13 = insertelement <8 x i16> %tmp12, i16 %tmp5, i32 2 ; <<8 x i16>> [#uses=1] + %tmp14 = insertelement <8 x i16> %tmp13, i16 %tmp6, i32 3 ; <<8 x i16>> [#uses=1] + %tmp15 = insertelement <8 x i16> %tmp14, i16 %tmp7, i32 4 ; <<8 x i16>> [#uses=1] + %tmp16 = insertelement <8 x i16> %tmp15, i16 %tmp8, i32 5 ; <<8 x i16>> [#uses=1] + %tmp17 = insertelement <8 x i16> %tmp16, i16 %tmp9, i32 6 ; <<8 x i16>> [#uses=1] + %tmp18 = insertelement <8 x i16> %tmp17, i16 %tmp10, i32 7 ; <<8 x i16>> [#uses=1] + %tmp18.upgrd.30 = bitcast <8 x i16> %tmp18 to <4 x i32> ; <<4 x i32>> [#uses=1] + store <4 x i32> %tmp18.upgrd.30, <4 x i32>* %A ret void } diff --git a/test/CodeGen/PowerPC/vec_spat.ll b/test/CodeGen/PowerPC/vec_spat.ll index a8acee0..7a35d87 100644 --- a/test/CodeGen/PowerPC/vec_spat.ll +++ b/test/CodeGen/PowerPC/vec_spat.ll @@ -1,73 +1,71 @@ ; Test that vectors are scalarized/lowered correctly. -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g3 | \ +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g3 | \ ; RUN: grep stfs | count 4 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 -o %t -f +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 -o %t -f ; RUN: grep vspltw %t | count 2 ; RUN: grep vsplti %t | count 3 ; RUN: grep vsplth %t | count 1 -%f4 = type <4 x float> -%i4 = type <4 x int> + %f4 = type <4 x float> + %i4 = type <4 x i32> -implementation - -void %splat(%f4* %P, %f4* %Q, float %X) { - %tmp = insertelement %f4 undef, float %X, uint 0 - %tmp2 = insertelement %f4 %tmp, float %X, uint 1 - %tmp4 = insertelement %f4 %tmp2, float %X, uint 2 - %tmp6 = insertelement %f4 %tmp4, float %X, uint 3 - %q = load %f4* %Q - %R = add %f4 %q, %tmp6 +define void @splat(%f4* %P, %f4* %Q, float %X) { + %tmp = insertelement %f4 undef, float %X, i32 0 ; <%f4> [#uses=1] + %tmp2 = insertelement %f4 %tmp, float %X, i32 1 ; <%f4> [#uses=1] + %tmp4 = insertelement %f4 %tmp2, float %X, i32 2 ; <%f4> [#uses=1] + %tmp6 = insertelement %f4 %tmp4, float %X, i32 3 ; <%f4> [#uses=1] + %q = load %f4* %Q ; <%f4> [#uses=1] + %R = add %f4 %q, %tmp6 ; <%f4> [#uses=1] store %f4 %R, %f4* %P ret void } -void %splat_i4(%i4* %P, %i4* %Q, int %X) { - %tmp = insertelement %i4 undef, int %X, uint 0 - %tmp2 = insertelement %i4 %tmp, int %X, uint 1 - %tmp4 = insertelement %i4 %tmp2, int %X, uint 2 - %tmp6 = insertelement %i4 %tmp4, int %X, uint 3 - %q = load %i4* %Q - %R = add %i4 %q, %tmp6 +define void @splat_i4(%i4* %P, %i4* %Q, i32 %X) { + %tmp = insertelement %i4 undef, i32 %X, i32 0 ; <%i4> [#uses=1] + %tmp2 = insertelement %i4 %tmp, i32 %X, i32 1 ; <%i4> [#uses=1] + %tmp4 = insertelement %i4 %tmp2, i32 %X, i32 2 ; <%i4> [#uses=1] + %tmp6 = insertelement %i4 %tmp4, i32 %X, i32 3 ; <%i4> [#uses=1] + %q = load %i4* %Q ; <%i4> [#uses=1] + %R = add %i4 %q, %tmp6 ; <%i4> [#uses=1] store %i4 %R, %i4* %P ret void } -void %splat_imm_i32(%i4* %P, %i4* %Q, int %X) { - %q = load %i4* %Q - %R = add %i4 %q, <int -1, int -1, int -1, int -1> +define void @splat_imm_i32(%i4* %P, %i4* %Q, i32 %X) { + %q = load %i4* %Q ; <%i4> [#uses=1] + %R = add %i4 %q, < i32 -1, i32 -1, i32 -1, i32 -1 > ; <%i4> [#uses=1] store %i4 %R, %i4* %P ret void } -void %splat_imm_i16(%i4* %P, %i4* %Q, int %X) { - %q = load %i4* %Q - %R = add %i4 %q, <int 65537, int 65537, int 65537, int 65537> +define void @splat_imm_i16(%i4* %P, %i4* %Q, i32 %X) { + %q = load %i4* %Q ; <%i4> [#uses=1] + %R = add %i4 %q, < i32 65537, i32 65537, i32 65537, i32 65537 > ; <%i4> [#uses=1] store %i4 %R, %i4* %P ret void } -void %splat_h(short %tmp, <16 x ubyte>* %dst) { - %tmp = insertelement <8 x short> undef, short %tmp, uint 0 - %tmp72 = insertelement <8 x short> %tmp, short %tmp, uint 1 - %tmp73 = insertelement <8 x short> %tmp72, short %tmp, uint 2 - %tmp74 = insertelement <8 x short> %tmp73, short %tmp, uint 3 - %tmp75 = insertelement <8 x short> %tmp74, short %tmp, uint 4 - %tmp76 = insertelement <8 x short> %tmp75, short %tmp, uint 5 - %tmp77 = insertelement <8 x short> %tmp76, short %tmp, uint 6 - %tmp78 = insertelement <8 x short> %tmp77, short %tmp, uint 7 - %tmp78 = cast <8 x short> %tmp78 to <16 x ubyte> - store <16 x ubyte> %tmp78, <16 x ubyte>* %dst - ret void +define void @splat_h(i16 %tmp, <16 x i8>* %dst) { + %tmp.upgrd.1 = insertelement <8 x i16> undef, i16 %tmp, i32 0 + %tmp72 = insertelement <8 x i16> %tmp.upgrd.1, i16 %tmp, i32 1 + %tmp73 = insertelement <8 x i16> %tmp72, i16 %tmp, i32 2 + %tmp74 = insertelement <8 x i16> %tmp73, i16 %tmp, i32 3 + %tmp75 = insertelement <8 x i16> %tmp74, i16 %tmp, i32 4 + %tmp76 = insertelement <8 x i16> %tmp75, i16 %tmp, i32 5 + %tmp77 = insertelement <8 x i16> %tmp76, i16 %tmp, i32 6 + %tmp78 = insertelement <8 x i16> %tmp77, i16 %tmp, i32 7 + %tmp78.upgrd.2 = bitcast <8 x i16> %tmp78 to <16 x i8> + store <16 x i8> %tmp78.upgrd.2, <16 x i8>* %dst + ret void } -void %spltish(<16 x ubyte>* %A, <16 x ubyte>* %B) { - ; Gets converted to 16 x ubyte - %tmp = load <16 x ubyte>* %B - %tmp.s = cast <16 x ubyte> %tmp to <16 x sbyte> - %tmp4 = sub <16 x sbyte> %tmp.s, cast (<8 x short> < short 15, short 15, short 15, short 15, short 15, short 15, short 15, short 15 > to <16 x sbyte>) - %tmp4.u = cast <16 x sbyte> %tmp4 to <16 x ubyte> - store <16 x ubyte> %tmp4.u, <16 x ubyte>* %A +define void @spltish(<16 x i8>* %A, <16 x i8>* %B) { + %tmp = load <16 x i8>* %B ; <<16 x i8>> [#uses=1] + %tmp.s = bitcast <16 x i8> %tmp to <16 x i8> ; <<16 x i8>> [#uses=1] + %tmp4 = sub <16 x i8> %tmp.s, bitcast (<8 x i16> < i16 15, i16 15, i16 15, i16 15, i16 15, i16 + 15, i16 15, i16 15 > to <16 x i8>) ; <<16 x i8>> [#uses=1] + %tmp4.u = bitcast <16 x i8> %tmp4 to <16 x i8> ; <<16 x i8>> [#uses=1] + store <16 x i8> %tmp4.u, <16 x i8>* %A ret void } diff --git a/test/CodeGen/PowerPC/vec_vrsave.ll b/test/CodeGen/PowerPC/vec_vrsave.ll index 63e3eba..06769f6 100644 --- a/test/CodeGen/PowerPC/vec_vrsave.ll +++ b/test/CodeGen/PowerPC/vec_vrsave.ll @@ -1,14 +1,14 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 -o %t -f +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 -o %t -f ; RUN: grep vrlw %t ; RUN: not grep spr %t ; RUN: not grep vrsave %t -<4 x int> %test_rol() { - ret <4 x int> < int -11534337, int -11534337, int -11534337, int -11534337 > +define <4 x i32> @test_rol() { + ret <4 x i32> < i32 -11534337, i32 -11534337, i32 -11534337, i32 -11534337 > } -<4 x int> %test_arg(<4 x int> %A, <4 x int> %B) { - %C = add <4 x int> %A, %B - ret <4 x int> %C +define <4 x i32> @test_arg(<4 x i32> %A, <4 x i32> %B) { + %C = add <4 x i32> %A, %B ; <<4 x i32>> [#uses=1] + ret <4 x i32> %C } diff --git a/test/CodeGen/PowerPC/vec_zero.ll b/test/CodeGen/PowerPC/vec_zero.ll index c845c0e..8d06a7d 100644 --- a/test/CodeGen/PowerPC/vec_zero.ll +++ b/test/CodeGen/PowerPC/vec_zero.ll @@ -1,8 +1,9 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | grep vxor +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | grep vxor -void %foo(<4 x float> *%P) { - %T = load <4 x float> * %P - %S = add <4 x float> zeroinitializer, %T - store <4 x float> %S, <4 x float>* %P - ret void +define void @foo(<4 x float>* %P) { + %T = load <4 x float>* %P ; <<4 x float>> [#uses=1] + %S = add <4 x float> zeroinitializer, %T ; <<4 x float>> [#uses=1] + store <4 x float> %S, <4 x float>* %P + ret void } + diff --git a/test/CodeGen/PowerPC/vector-identity-shuffle.ll b/test/CodeGen/PowerPC/vector-identity-shuffle.ll index af5cc02..aefd266 100644 --- a/test/CodeGen/PowerPC/vector-identity-shuffle.ll +++ b/test/CodeGen/PowerPC/vector-identity-shuffle.ll @@ -1,16 +1,17 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | grep test: -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 | not grep vperm +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | grep test: +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 | not grep vperm -void %test(<4 x float> *%tmp2.i) { - %tmp2.i = load <4x float>* %tmp2.i - %xFloat0.48 = extractelement <4 x float> %tmp2.i, uint 0 ; <float> [#uses=1] - %inFloat0.49 = insertelement <4 x float> undef, float %xFloat0.48, uint 0 ; <<4 x float>> [#uses=1] - %xFloat1.50 = extractelement <4 x float> %tmp2.i, uint 1 ; <float> [#uses=1] - %inFloat1.52 = insertelement <4 x float> %inFloat0.49, float %xFloat1.50, uint 1 ; <<4 x float>> [#uses=1] - %xFloat2.53 = extractelement <4 x float> %tmp2.i, uint 2 ; <float> [#uses=1] - %inFloat2.55 = insertelement <4 x float> %inFloat1.52, float %xFloat2.53, uint 2 ; <<4 x float>> [#uses=1] - %xFloat3.56 = extractelement <4 x float> %tmp2.i, uint 3 ; <float> [#uses=1] - %inFloat3.58 = insertelement <4 x float> %inFloat2.55, float %xFloat3.56, uint 3 ; <<4 x float>> [#uses=4] - store <4 x float> %inFloat3.58, <4x float>* %tmp2.i - ret void +define void @test(<4 x float>* %tmp2.i) { + %tmp2.i.upgrd.1 = load <4 x float>* %tmp2.i ; <<4 x float>> [#uses=4] + %xFloat0.48 = extractelement <4 x float> %tmp2.i.upgrd.1, i32 0 ; <float> [#uses=1] + %inFloat0.49 = insertelement <4 x float> undef, float %xFloat0.48, i32 0 ; <<4 x float>> [#uses=1] + %xFloat1.50 = extractelement <4 x float> %tmp2.i.upgrd.1, i32 1 ; <float> [#uses=1] + %inFloat1.52 = insertelement <4 x float> %inFloat0.49, float %xFloat1.50, i32 1 ; <<4 x float>> [#uses=1] + %xFloat2.53 = extractelement <4 x float> %tmp2.i.upgrd.1, i32 2 ; <float> [#uses=1] + %inFloat2.55 = insertelement <4 x float> %inFloat1.52, float %xFloat2.53, i32 2 ; <<4 x float>> [#uses=1] + %xFloat3.56 = extractelement <4 x float> %tmp2.i.upgrd.1, i32 3 ; <float> [#uses=1] + %inFloat3.58 = insertelement <4 x float> %inFloat2.55, float %xFloat3.56, i32 3 ; <<4 x float>> [#uses=1] + store <4 x float> %inFloat3.58, <4 x float>* %tmp2.i + ret void } + diff --git a/test/CodeGen/PowerPC/vector.ll b/test/CodeGen/PowerPC/vector.ll index 789b543..f121306 100644 --- a/test/CodeGen/PowerPC/vector.ll +++ b/test/CodeGen/PowerPC/vector.ll @@ -1,156 +1,157 @@ ; Test that vectors are scalarized/lowered correctly. -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5 -; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g3 +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g5 +; RUN: llvm-as < %s | llc -march=ppc32 -mcpu=g3 +%d8 = type <8 x double> %f1 = type <1 x float> %f2 = type <2 x float> %f4 = type <4 x float> -%i4 = type <4 x int> %f8 = type <8 x float> -%d8 = type <8 x double> - -implementation +%i4 = type <4 x i32> ;;; TEST HANDLING OF VARIOUS VECTOR SIZES -void %test_f1(%f1 *%P, %f1* %Q, %f1 *%S) { - %p = load %f1 *%P - %q = load %f1* %Q - %R = add %f1 %p, %q - store %f1 %R, %f1 *%S - ret void +define void @test_f1(%f1* %P, %f1* %Q, %f1* %S) { + %p = load %f1* %P ; <%f1> [#uses=1] + %q = load %f1* %Q ; <%f1> [#uses=1] + %R = add %f1 %p, %q ; <%f1> [#uses=1] + store %f1 %R, %f1* %S + ret void } -void %test_f2(%f2 *%P, %f2* %Q, %f2 *%S) { - %p = load %f2* %P - %q = load %f2* %Q - %R = add %f2 %p, %q - store %f2 %R, %f2 *%S - ret void +define void @test_f2(%f2* %P, %f2* %Q, %f2* %S) { + %p = load %f2* %P ; <%f2> [#uses=1] + %q = load %f2* %Q ; <%f2> [#uses=1] + %R = add %f2 %p, %q ; <%f2> [#uses=1] + store %f2 %R, %f2* %S + ret void } -void %test_f4(%f4 *%P, %f4* %Q, %f4 *%S) { - %p = load %f4* %P - %q = load %f4* %Q - %R = add %f4 %p, %q - store %f4 %R, %f4 *%S - ret void +define void @test_f4(%f4* %P, %f4* %Q, %f4* %S) { + %p = load %f4* %P ; <%f4> [#uses=1] + %q = load %f4* %Q ; <%f4> [#uses=1] + %R = add %f4 %p, %q ; <%f4> [#uses=1] + store %f4 %R, %f4* %S + ret void } -void %test_f8(%f8 *%P, %f8* %Q, %f8 *%S) { - %p = load %f8* %P - %q = load %f8* %Q - %R = add %f8 %p, %q - store %f8 %R, %f8 *%S - ret void +define void @test_f8(%f8* %P, %f8* %Q, %f8* %S) { + %p = load %f8* %P ; <%f8> [#uses=1] + %q = load %f8* %Q ; <%f8> [#uses=1] + %R = add %f8 %p, %q ; <%f8> [#uses=1] + store %f8 %R, %f8* %S + ret void } -void %test_fmul(%f8 *%P, %f8* %Q, %f8 *%S) { - %p = load %f8* %P - %q = load %f8* %Q - %R = mul %f8 %p, %q - store %f8 %R, %f8 *%S - ret void +define void @test_fmul(%f8* %P, %f8* %Q, %f8* %S) { + %p = load %f8* %P ; <%f8> [#uses=1] + %q = load %f8* %Q ; <%f8> [#uses=1] + %R = mul %f8 %p, %q ; <%f8> [#uses=1] + store %f8 %R, %f8* %S + ret void } -void %test_div(%f8 *%P, %f8* %Q, %f8 *%S) { - %p = load %f8* %P - %q = load %f8* %Q - %R = div %f8 %p, %q - store %f8 %R, %f8 *%S - ret void +define void @test_div(%f8* %P, %f8* %Q, %f8* %S) { + %p = load %f8* %P ; <%f8> [#uses=1] + %q = load %f8* %Q ; <%f8> [#uses=1] + %R = fdiv %f8 %p, %q ; <%f8> [#uses=1] + store %f8 %R, %f8* %S + ret void } ;;; TEST VECTOR CONSTRUCTS -void %test_cst(%f4 *%P, %f4 *%S) { - %p = load %f4* %P - %R = add %f4 %p, <float 0x3FB99999A0000000, float 1.0, float 2.0, float 4.5> - store %f4 %R, %f4 *%S - ret void +define void @test_cst(%f4* %P, %f4* %S) { + %p = load %f4* %P ; <%f4> [#uses=1] + %R = add %f4 %p, < float 0x3FB99999A0000000, float 1.000000e+00, float + 2.000000e+00, float 4.500000e+00 > ; <%f4> [#uses=1] + store %f4 %R, %f4* %S + ret void } -void %test_zero(%f4 *%P, %f4 *%S) { - %p = load %f4* %P - %R = add %f4 %p, zeroinitializer - store %f4 %R, %f4 *%S - ret void +define void @test_zero(%f4* %P, %f4* %S) { + %p = load %f4* %P ; <%f4> [#uses=1] + %R = add %f4 %p, zeroinitializer ; <%f4> [#uses=1] + store %f4 %R, %f4* %S + ret void } -void %test_undef(%f4 *%P, %f4 *%S) { - %p = load %f4* %P - %R = add %f4 %p, undef - store %f4 %R, %f4 *%S - ret void +define void @test_undef(%f4* %P, %f4* %S) { + %p = load %f4* %P ; <%f4> [#uses=1] + %R = add %f4 %p, undef ; <%f4> [#uses=1] + store %f4 %R, %f4* %S + ret void } -void %test_constant_insert(%f4 *%S) { - %R = insertelement %f4 zeroinitializer, float 10.0, uint 0 - store %f4 %R, %f4 *%S - ret void +define void @test_constant_insert(%f4* %S) { + %R = insertelement %f4 zeroinitializer, float 1.000000e+01, i32 0 + ; <%f4> [#uses=1] + store %f4 %R, %f4* %S + ret void } -void %test_variable_buildvector(float %F, %f4 *%S) { - %R = insertelement %f4 zeroinitializer, float %F, uint 0 - store %f4 %R, %f4 *%S - ret void +define void @test_variable_buildvector(float %F, %f4* %S) { + %R = insertelement %f4 zeroinitializer, float %F, i32 0 + store %f4 %R, %f4* %S + ret void } -void %test_scalar_to_vector(float %F, %f4 *%S) { - %R = insertelement %f4 undef, float %F, uint 0 ;; R = scalar_to_vector F - store %f4 %R, %f4 *%S - ret void +define void @test_scalar_to_vector(float %F, %f4* %S) { + %R = insertelement %f4 undef, float %F, i32 0 + store %f4 %R, %f4* %S + ret void } -float %test_extract_elt(%f8 *%P) { - %p = load %f8* %P - %R = extractelement %f8 %p, uint 3 - ret float %R +define float @test_extract_elt(%f8* %P) { + %p = load %f8* %P ; <%f8> [#uses=1] + %R = extractelement %f8 %p, i32 3 ; <float> [#uses=1] + ret float %R } -double %test_extract_elt2(%d8 *%P) { - %p = load %d8* %P - %R = extractelement %d8 %p, uint 3 - ret double %R +define double @test_extract_elt2(%d8* %P) { + %p = load %d8* %P ; <%d8> [#uses=1] + %R = extractelement %d8 %p, i32 3 ; <double> [#uses=1] + ret double %R } -void %test_cast_1(<4 x float>* %b, <4 x int>* %a) { - %tmp = load <4 x float>* %b - %tmp2 = add <4 x float> %tmp, <float 1.0, float 2.0, float 3.0, float 4.0> - %tmp3 = cast <4 x float> %tmp2 to <4 x int> - %tmp4 = add <4 x int> %tmp3, <int 1, int 2, int 3, int 4> - store <4 x int> %tmp4, <4 x int>* %a - ret void +define void @test_cast_1(%f4* %b, %i4* %a) { + %tmp = load %f4* %b ; <%f4> [#uses=1] + %tmp2 = add %f4 %tmp, < float 1.000000e+00, float 2.000000e+00, float +3.000000e+00, float 4.000000e+00 > ; <%f4> [#uses=1] + %tmp3 = bitcast %f4 %tmp2 to %i4 ; <%i4> [#uses=1] + %tmp4 = add %i4 %tmp3, < i32 1, i32 2, i32 3, i32 4 > + store %i4 %tmp4, %i4* %a + ret void } -void %test_cast_2(<8 x float>* %a, <8 x int>* %b) { - %T = load <8 x float>* %a - %T2 = cast <8 x float> %T to <8 x int> - store <8 x int> %T2, <8 x int>* %b - ret void +define void @test_cast_2(%f8* %a, <8 x i32>* %b) { + %T = load %f8* %a ; <%f8> [#uses=1] + %T2 = bitcast %f8 %T to <8 x i32> + store <8 x i32> %T2, <8 x i32>* %b + ret void } ;;; TEST IMPORTANT IDIOMS -void %splat(%f4* %P, %f4* %Q, float %X) { - %tmp = insertelement %f4 undef, float %X, uint 0 - %tmp2 = insertelement %f4 %tmp, float %X, uint 1 - %tmp4 = insertelement %f4 %tmp2, float %X, uint 2 - %tmp6 = insertelement %f4 %tmp4, float %X, uint 3 - %q = load %f4* %Q - %R = add %f4 %q, %tmp6 +define void @splat(%f4* %P, %f4* %Q, float %X) { + %tmp = insertelement %f4 undef, float %X, i32 0 + %tmp2 = insertelement %f4 %tmp, float %X, i32 1 + %tmp4 = insertelement %f4 %tmp2, float %X, i32 2 + %tmp6 = insertelement %f4 %tmp4, float %X, i32 3 + %q = load %f4* %Q ; <%f4> [#uses=1] + %R = add %f4 %q, %tmp6 ; <%f4> [#uses=1] store %f4 %R, %f4* %P ret void } -void %splat_i4(%i4* %P, %i4* %Q, int %X) { - %tmp = insertelement %i4 undef, int %X, uint 0 - %tmp2 = insertelement %i4 %tmp, int %X, uint 1 - %tmp4 = insertelement %i4 %tmp2, int %X, uint 2 - %tmp6 = insertelement %i4 %tmp4, int %X, uint 3 - %q = load %i4* %Q - %R = add %i4 %q, %tmp6 +define void @splat_i4(%i4* %P, %i4* %Q, i32 %X) { + %tmp = insertelement %i4 undef, i32 %X, i32 0 + %tmp2 = insertelement %i4 %tmp, i32 %X, i32 1 + %tmp4 = insertelement %i4 %tmp2, i32 %X, i32 2 + %tmp6 = insertelement %i4 %tmp4, i32 %X, i32 3 + %q = load %i4* %Q ; <%i4> [#uses=1] + %R = add %i4 %q, %tmp6 ; <%i4> [#uses=1] store %i4 %R, %i4* %P ret void } |