replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Add an instcombine to clean up a common pattern produced 

by the SRoA "promote to large integer" code, eliminating
some type conversions like this:

   %94 = zext i16 %93 to i32                       ; <i32> [#uses=2]
   %96 = lshr i32 %94, 8                           ; <i32> [#uses=1]
   %101 = trunc i32 %96 to i8                      ; <i8> [#uses=1]

This also unblocks other xforms from happening, now clang is able to compile:

struct S { float A, B, C, D; };
float foo(struct S A) { return A.A + A.B+A.C+A.D; }

into:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	pshufd	$1, %xmm0, %xmm2
	addss	%xmm0, %xmm2
	movdqa	%xmm1, %xmm3
	addss	%xmm2, %xmm3
	pshufd	$1, %xmm1, %xmm0
	addss	%xmm3, %xmm0
	ret

on x86-64, instead of:

_foo:                                   ## @foo
## BB#0:                                ## %entry
	movd	%xmm0, %rax
	shrq	$32, %rax
	movd	%eax, %xmm2
	addss	%xmm0, %xmm2
	movapd	%xmm1, %xmm3
	addss	%xmm2, %xmm3
	movd	%xmm1, %rax
	shrq	$32, %rax
	movd	%eax, %xmm0
	addss	%xmm3, %xmm0
	ret

This seems pretty close to optimal to me, at least without
using horizontal adds.  This also triggers in lots of other
code, including SPEC.

llvm-svn: 112278
This commit is contained in:
Chris Lattner 2010-08-27 18:31:05 +00:00
parent 5082c5fdf6
commit 90cd746e63
2 changed files with 55 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
View File

@ -454,6 +454,29 @@ Instruction *InstCombiner::visitTrunc(TruncInst &CI) {
Value *Zero = Constant::getNullValue(Src->getType());
return new ICmpInst(ICmpInst::ICMP_NE, Src, Zero);
}
// Transform trunc(lshr (zext A), Cst) to eliminate one type conversion.
Value *A = 0; ConstantInt *Cst = 0;
if (match(Src, m_LShr(m_ZExt(m_Value(A)), m_ConstantInt(Cst))) &&
Src->hasOneUse()) {
// We have three types to worry about here, the type of A, the source of
// the truncate (MidSize), and the destination of the truncate. We know that
// ASize < MidSize and MidSize > ResultSize, but don't know the relation
// between ASize and ResultSize.
unsigned ASize = A->getType()->getPrimitiveSizeInBits();
// If the shift amount is larger than the size of A, then the result is
// known to be zero because all the input bits got shifted out.
if (Cst->getZExtValue() >= ASize)
return ReplaceInstUsesWith(CI, Constant::getNullValue(CI.getType()));
// Since we're doing an lshr and a zero extend, and know that the shift
// amount is smaller than ASize, it is always safe to do the shift in A's
// type, then zero extend or truncate to the result.
Value *Shift = Builder->CreateLShr(A, Cst->getZExtValue());
Shift->takeName(Src);
return CastInst::CreateIntegerCast(Shift, CI.getType(), false);
}
return 0;
}

32
llvm/test/Transforms/InstCombine/trunc.ll
View File

@ -48,3 +48,35 @@ define i64 @test4(i64 %a) {
; CHECK: %d = xor i64 {{.*}}, 8
; CHECK: ret i64 %d
}
define i32 @test5(i32 %A) {
%B = zext i32 %A to i128
%C = lshr i128 %B, 16
%D = trunc i128 %C to i32
ret i32 %D
; CHECK: @test5
; CHECK: %C = lshr i32 %A, 16
; CHECK: ret i32 %C
}
define i32 @test6(i64 %A) {
%B = zext i64 %A to i128
%C = lshr i128 %B, 32
%D = trunc i128 %C to i32
ret i32 %D
; CHECK: @test6
; CHECK: %C = lshr i64 %A, 32
; CHECK: %D = trunc i64 %C to i32
; CHECK: ret i32 %D
}
define i92 @test7(i64 %A) {
%B = zext i64 %A to i128
%C = lshr i128 %B, 32
%D = trunc i128 %C to i92
ret i92 %D
; CHECK: @test7
; CHECK: %C = lshr i64 %A, 32
; CHECK: %D = zext i64 %C to i92
; CHECK: ret i92 %D
}