mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 03:02:36 +01:00
504d614e86
The PPC test with 2 extra uses seems clearly better by avoiding this transform. With 1 extra use, we also prevent an extra register move (although that might be an RA problem). The general rule should be to only make a change here if it is always profitable. The x86 diffs are all neutral. llvm-svn: 348518
72 lines
1.9 KiB
LLVM
72 lines
1.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu | FileCheck %s
|
|
|
|
; This is good - eliminate an op by hoisting logic.
|
|
|
|
define i32 @lshr_or(i32 %x, i32 %y, i32 %z, i32* %p1, i32* %p2) {
|
|
; CHECK-LABEL: lshr_or:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: or 3, 3, 4
|
|
; CHECK-NEXT: srw 3, 3, 5
|
|
; CHECK-NEXT: blr
|
|
%xt = lshr i32 %x, %z
|
|
%yt = lshr i32 %y, %z
|
|
%r = or i32 %xt, %yt
|
|
ret i32 %r
|
|
}
|
|
|
|
; This is questionable - hoisting doesn't eliminate anything.
|
|
; It might result in an extra register move.
|
|
|
|
define i32 @lshr_or_multiuse1(i32 %x, i32 %y, i32 %z, i32* %p1, i32* %p2) {
|
|
; CHECK-LABEL: lshr_or_multiuse1:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: srw 7, 3, 5
|
|
; CHECK-NEXT: srw 3, 4, 5
|
|
; CHECK-NEXT: or 3, 7, 3
|
|
; CHECK-NEXT: stw 7, 0(6)
|
|
; CHECK-NEXT: blr
|
|
%xt = lshr i32 %x, %z
|
|
%yt = lshr i32 %y, %z
|
|
store i32 %xt, i32* %p1
|
|
%r = or i32 %xt, %yt
|
|
ret i32 %r
|
|
}
|
|
|
|
; This is questionable - hoisting doesn't eliminate anything.
|
|
|
|
define i32 @lshr_multiuse2(i32 %x, i32 %y, i32 %z, i32* %p1, i32* %p2) {
|
|
; CHECK-LABEL: lshr_multiuse2:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: srw 3, 3, 5
|
|
; CHECK-NEXT: srw 4, 4, 5
|
|
; CHECK-NEXT: or 3, 3, 4
|
|
; CHECK-NEXT: stw 4, 0(7)
|
|
; CHECK-NEXT: blr
|
|
%xt = lshr i32 %x, %z
|
|
%yt = lshr i32 %y, %z
|
|
store i32 %yt, i32* %p2
|
|
%r = or i32 %xt, %yt
|
|
ret i32 %r
|
|
}
|
|
|
|
; This is not profitable to hoist. We need an extra shift instruction.
|
|
|
|
define i32 @lshr_multiuse3(i32 %x, i32 %y, i32 %z, i32* %p1, i32* %p2) {
|
|
; CHECK-LABEL: lshr_multiuse3:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: srw 3, 3, 5
|
|
; CHECK-NEXT: srw 4, 4, 5
|
|
; CHECK-NEXT: stw 3, 0(6)
|
|
; CHECK-NEXT: or 3, 3, 4
|
|
; CHECK-NEXT: stw 4, 0(7)
|
|
; CHECK-NEXT: blr
|
|
%xt = lshr i32 %x, %z
|
|
%yt = lshr i32 %y, %z
|
|
store i32 %xt, i32* %p1
|
|
store i32 %yt, i32* %p2
|
|
%r = or i32 %xt, %yt
|
|
ret i32 %r
|
|
}
|
|
|