1. Hoist minus sign as high as possible in an attempt to reveal

some optimization opportunities (in the enclosing supper-expressions).

   rule 1. (-0.0 - X ) * Y => -0.0 - (X * Y)
     if expression "-0.0 - X" has only one reference.

   rule 2. (0.0 - X ) * Y => -0.0 - (X * Y)
     if expression "0.0 - X" has only one reference, and
        the instruction is marked "noSignedZero".

2. Eliminate negation (The compiler was already able to handle these
    opt if the 0.0s are replaced with -0.0.)

   rule 3: (0.0 - X) * (0.0 - Y) => X * Y
   rule 4: (0.0 - X) * C => X * -C
   if the expr is flagged "noSignedZero".

3. 
  Rule 5: (X*Y) * X => (X*X) * Y
   if X!=Y and the expression is flagged with "UnsafeAlgebra".

   The purpose of this transformation is two-fold:
    a) to form a power expression (of X).
    b) potentially shorten the critical path: After transformation, the
       latency of the instruction Y is amortized by the expression of X*X,
       and therefore Y is in a "less critical" position compared to what it
      was before the transformation. 

4. Remove the InstCombine code about simplifiying "X * select".
   
   The reasons are following:
    a) The "select" is somewhat architecture-dependent, therefore the
       higher level optimizers are not able to precisely predict if
       the simplification really yields any performance improvement
       or not.

    b) The "select" operator is bit complicate, and tends to obscure
       optimization opportunities. It is btter to keep it as low as
       possible in expr tree, and let CodeGen to tackle the optimization.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172551 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Shuxin Yang
2013-01-15 21:09:32 +00:00
parent 3d69041abe
commit a1444219b2
3 changed files with 152 additions and 64 deletions

View File

@@ -130,37 +130,6 @@ define double @fail2(double %f1, double %f2) {
; CHECK: ret
}
; rdar://12753946: x * cond ? 1.0 : 0.0 => cond ? x : 0.0
define double @select1(i32 %cond, double %x, double %y) {
%tobool = icmp ne i32 %cond, 0
%cond1 = select i1 %tobool, double 1.000000e+00, double 0.000000e+00
%mul = fmul nnan nsz double %cond1, %x
%add = fadd double %mul, %y
ret double %add
; CHECK: @select1
; CHECK: select i1 %tobool, double %x, double 0.000000e+00
}
define double @select2(i32 %cond, double %x, double %y) {
%tobool = icmp ne i32 %cond, 0
%cond1 = select i1 %tobool, double 0.000000e+00, double 1.000000e+00
%mul = fmul nnan nsz double %cond1, %x
%add = fadd double %mul, %y
ret double %add
; CHECK: @select2
; CHECK: select i1 %tobool, double 0.000000e+00, double %x
}
define double @select3(i32 %cond, double %x, double %y) {
%tobool = icmp ne i32 %cond, 0
%cond1 = select i1 %tobool, double 0.000000e+00, double 2.000000e+00
%mul = fmul nnan nsz double %cond1, %x
%add = fadd double %mul, %y
ret double %add
; CHECK: @select3
; CHECK: fmul nnan nsz double %cond1, %x
}
; =========================================================================
;
; Testing-cases about fmul begin
@@ -243,6 +212,25 @@ define float @fmul5(float %f1, float %f2) {
; CHECK: fdiv fast float %f1, 0x47E8000000000000
}
; (X*Y) * X => (X*X) * Y
define float @fmul6(float %f1, float %f2) {
%mul = fmul float %f1, %f2
%mul1 = fmul fast float %mul, %f1
ret float %mul1
; CHECK: @fmul6
; CHECK: fmul fast float %f1, %f1
}
; "(X*Y) * X => (X*X) * Y" is disabled if "X*Y" has multiple uses
define float @fmul7(float %f1, float %f2) {
%mul = fmul float %f1, %f2
%mul1 = fmul fast float %mul, %f1
%add = fadd float %mul1, %mul
ret float %add
; CHECK: @fmul7
; CHECK: fmul fast float %mul, %f1
}
; =========================================================================
;
; Testing-cases about negation
@@ -262,8 +250,8 @@ define float @fneg1(float %f1, float %f2) {
; Testing-cases about div
;
; =========================================================================
; X/C1 / C2 => X * (1/(C2*C1))
; X/C1 / C2 => X * (1/(C2*C1))
define float @fdiv1(float %x) {
%div = fdiv float %x, 0x3FF3333340000000
%div1 = fdiv fast float %div, 0x4002666660000000
@@ -351,4 +339,3 @@ define float @fdiv9(float %x) {
; CHECK: @fdiv9
; CHECK: fmul fast float %x, 5.000000e+00
}