1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 02:52:53 +02:00
llvm-mirror/test/Transforms/InstCombine/sitofp.ll

185 lines
3.9 KiB
LLVM
Raw Normal View History

; RUN: opt < %s -instcombine -S | FileCheck %s
; CHECK-LABEL: test1
; CHECK: ret i1 true
define i1 @test1(i8 %A) {
%B = sitofp i8 %A to double
%C = fcmp ult double %B, 128.0
ret i1 %C
}
; CHECK-LABEL: test2
; CHECK: ret i1 true
define i1 @test2(i8 %A) {
%B = sitofp i8 %A to double
%C = fcmp ugt double %B, -128.1
ret i1 %C
}
; CHECK-LABEL: test3
; CHECK: ret i1 true
define i1 @test3(i8 %A) {
%B = sitofp i8 %A to double
%C = fcmp ule double %B, 127.0
ret i1 %C
}
; CHECK-LABEL: test4
; CHECK: icmp ne i8 %A, 127
; CHECK-NEXT: ret i1
define i1 @test4(i8 %A) {
%B = sitofp i8 %A to double
%C = fcmp ult double %B, 127.0
ret i1 %C
}
; CHECK-LABEL: test5
; CHECK: ret i32
define i32 @test5(i32 %A) {
%B = sitofp i32 %A to double
%C = fptosi double %B to i32
%D = uitofp i32 %C to double
%E = fptoui double %D to i32
ret i32 %E
}
; CHECK-LABEL: test6
; CHECK: and i32 %A, 39
; CHECK-NEXT: ret i32
Teach instcombine 4 new xforms: (add (sext x), cst) --> (sext (add x, cst')) (add (sext x), (sext y)) --> (sext (add int x, y)) (add double (sitofp x), fpcst) --> (sitofp (add int x, intcst)) (add double (sitofp x), (sitofp y)) --> (sitofp (add int x, y)) This generally reduces conversions. For example MiBench/telecomm-gsm gets these simplifications: HACK2: %tmp67.i142.i.i = sext i16 %tmp6.i141.i.i to i32 ; <i32> [#uses=1] %tmp23.i139.i.i = sext i16 %tmp2.i138.i.i to i32 ; <i32> [#uses=1] %tmp8.i143.i.i = add i32 %tmp67.i142.i.i, %tmp23.i139.i.i ; <i32> [#uses=3] HACK2: %tmp67.i121.i.i = sext i16 %tmp6.i120.i.i to i32 ; <i32> [#uses=1] %tmp23.i118.i.i = sext i16 %tmp2.i117.i.i to i32 ; <i32> [#uses=1] %tmp8.i122.i.i = add i32 %tmp67.i121.i.i, %tmp23.i118.i.i ; <i32> [#uses=3] HACK2: %tmp67.i.i190.i = sext i16 %tmp6.i.i189.i to i32 ; <i32> [#uses=1] %tmp23.i.i187.i = sext i16 %tmp2.i.i186.i to i32 ; <i32> [#uses=1] %tmp8.i.i191.i = add i32 %tmp67.i.i190.i, %tmp23.i.i187.i ; <i32> [#uses=3] HACK2: %tmp67.i173.i.i.i = sext i16 %tmp6.i172.i.i.i to i32 ; <i32> [#uses=1] %tmp23.i170.i.i.i = sext i16 %tmp2.i169.i.i.i to i32 ; <i32> [#uses=1] %tmp8.i174.i.i.i = add i32 %tmp67.i173.i.i.i, %tmp23.i170.i.i.i ; <i32> [#uses=3] HACK2: %tmp67.i152.i.i.i = sext i16 %tmp6.i151.i.i.i to i32 ; <i32> [#uses=1] %tmp23.i149.i.i.i = sext i16 %tmp2.i148.i.i.i to i32 ; <i32> [#uses=1] %tmp8.i153.i.i.i = add i32 %tmp67.i152.i.i.i, %tmp23.i149.i.i.i ; <i32> [#uses=3] HACK2: %tmp67.i.i.i.i = sext i16 %tmp6.i.i.i.i to i32 ; <i32> [#uses=1] %tmp23.i.i5.i.i = sext i16 %tmp2.i.i.i.i to i32 ; <i32> [#uses=1] %tmp8.i.i7.i.i = add i32 %tmp67.i.i.i.i, %tmp23.i.i5.i.i ; <i32> [#uses=3] This also fixes a bug in ComputeNumSignBits handling select and makes it more aggressive with and/or. llvm-svn: 51302
2008-05-20 07:46:13 +02:00
define i32 @test6(i32 %A) {
%B = and i32 %A, 7
%C = and i32 %A, 32
%D = sitofp i32 %B to double
%E = sitofp i32 %C to double
%F = fadd double %D, %E
%G = fptosi double %F to i32
ret i32 %G
Teach instcombine 4 new xforms: (add (sext x), cst) --> (sext (add x, cst')) (add (sext x), (sext y)) --> (sext (add int x, y)) (add double (sitofp x), fpcst) --> (sitofp (add int x, intcst)) (add double (sitofp x), (sitofp y)) --> (sitofp (add int x, y)) This generally reduces conversions. For example MiBench/telecomm-gsm gets these simplifications: HACK2: %tmp67.i142.i.i = sext i16 %tmp6.i141.i.i to i32 ; <i32> [#uses=1] %tmp23.i139.i.i = sext i16 %tmp2.i138.i.i to i32 ; <i32> [#uses=1] %tmp8.i143.i.i = add i32 %tmp67.i142.i.i, %tmp23.i139.i.i ; <i32> [#uses=3] HACK2: %tmp67.i121.i.i = sext i16 %tmp6.i120.i.i to i32 ; <i32> [#uses=1] %tmp23.i118.i.i = sext i16 %tmp2.i117.i.i to i32 ; <i32> [#uses=1] %tmp8.i122.i.i = add i32 %tmp67.i121.i.i, %tmp23.i118.i.i ; <i32> [#uses=3] HACK2: %tmp67.i.i190.i = sext i16 %tmp6.i.i189.i to i32 ; <i32> [#uses=1] %tmp23.i.i187.i = sext i16 %tmp2.i.i186.i to i32 ; <i32> [#uses=1] %tmp8.i.i191.i = add i32 %tmp67.i.i190.i, %tmp23.i.i187.i ; <i32> [#uses=3] HACK2: %tmp67.i173.i.i.i = sext i16 %tmp6.i172.i.i.i to i32 ; <i32> [#uses=1] %tmp23.i170.i.i.i = sext i16 %tmp2.i169.i.i.i to i32 ; <i32> [#uses=1] %tmp8.i174.i.i.i = add i32 %tmp67.i173.i.i.i, %tmp23.i170.i.i.i ; <i32> [#uses=3] HACK2: %tmp67.i152.i.i.i = sext i16 %tmp6.i151.i.i.i to i32 ; <i32> [#uses=1] %tmp23.i149.i.i.i = sext i16 %tmp2.i148.i.i.i to i32 ; <i32> [#uses=1] %tmp8.i153.i.i.i = add i32 %tmp67.i152.i.i.i, %tmp23.i149.i.i.i ; <i32> [#uses=3] HACK2: %tmp67.i.i.i.i = sext i16 %tmp6.i.i.i.i to i32 ; <i32> [#uses=1] %tmp23.i.i5.i.i = sext i16 %tmp2.i.i.i.i to i32 ; <i32> [#uses=1] %tmp8.i.i7.i.i = add i32 %tmp67.i.i.i.i, %tmp23.i.i5.i.i ; <i32> [#uses=3] This also fixes a bug in ComputeNumSignBits handling select and makes it more aggressive with and/or. llvm-svn: 51302
2008-05-20 07:46:13 +02:00
}
; CHECK-LABEL: test7
; CHECK: ret i32
define i32 @test7(i32 %A) nounwind {
%B = sitofp i32 %A to double
%C = fptoui double %B to i32
ret i32 %C
}
; CHECK-LABEL: test8
; CHECK: ret i32
define i32 @test8(i32 %A) nounwind {
%B = uitofp i32 %A to double
%C = fptosi double %B to i32
ret i32 %C
}
; CHECK-LABEL: test9
; CHECK: zext i8
; CHECK-NEXT: ret i32
define i32 @test9(i8 %A) nounwind {
%B = sitofp i8 %A to float
%C = fptoui float %B to i32
ret i32 %C
}
; CHECK-LABEL: test10
; CHECK: sext i8
; CHECK-NEXT: ret i32
define i32 @test10(i8 %A) nounwind {
%B = sitofp i8 %A to float
%C = fptosi float %B to i32
ret i32 %C
}
; If the input value is outside of the range of the output cast, it's
; undefined behavior, so we can assume it fits.
; CHECK-LABEL: test11
; CHECK: trunc
; CHECK-NEXT: ret i8
define i8 @test11(i32 %A) nounwind {
%B = sitofp i32 %A to float
%C = fptosi float %B to i8
ret i8 %C
}
; If the input value is negative, it'll be outside the range of the
; output cast, and thus undefined behavior.
; CHECK-LABEL: test12
; CHECK: zext i8
; CHECK-NEXT: ret i32
define i32 @test12(i8 %A) nounwind {
%B = sitofp i8 %A to float
%C = fptoui float %B to i32
ret i32 %C
}
; This can't fold because the 25-bit input doesn't fit in the mantissa.
; CHECK-LABEL: test13
; CHECK: uitofp
; CHECK-NEXT: fptoui
define i32 @test13(i25 %A) nounwind {
%B = uitofp i25 %A to float
%C = fptoui float %B to i32
ret i32 %C
}
; But this one can.
; CHECK-LABEL: test14
; CHECK: zext i24
; CHECK-NEXT: ret i32
define i32 @test14(i24 %A) nounwind {
%B = uitofp i24 %A to float
%C = fptoui float %B to i32
ret i32 %C
}
; And this one can too.
; CHECK-LABEL: test15
; CHECK: trunc i32
; CHECK-NEXT: ret i24
define i24 @test15(i32 %A) nounwind {
%B = uitofp i32 %A to float
%C = fptoui float %B to i24
ret i24 %C
}
; This can fold because the 25-bit input is signed and we disard the sign bit.
; CHECK-LABEL: test16
; CHECK: zext
define i32 @test16(i25 %A) nounwind {
%B = sitofp i25 %A to float
%C = fptoui float %B to i32
ret i32 %C
}
; This can't fold because the 26-bit input won't fit the mantissa
; even after disarding the signed bit.
; CHECK-LABEL: test17
; CHECK: sitofp
; CHECK-NEXT: fptoui
define i32 @test17(i26 %A) nounwind {
%B = sitofp i26 %A to float
%C = fptoui float %B to i32
ret i32 %C
}
; This can fold because the 54-bit output is signed and we disard the sign bit.
; CHECK-LABEL: test18
; CHECK: trunc
define i54 @test18(i64 %A) nounwind {
%B = sitofp i64 %A to double
%C = fptosi double %B to i54
ret i54 %C
}
; This can't fold because the 55-bit output won't fit the mantissa
; even after disarding the sign bit.
; CHECK-LABEL: test19
; CHECK: sitofp
; CHECK-NEXT: fptosi
define i55 @test19(i64 %A) nounwind {
%B = sitofp i64 %A to double
%C = fptosi double %B to i55
ret i55 %C
}