mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-24 19:52:54 +01:00
b6a12c439f
This is almost the same as an existing IR canonicalization in instcombine, so I'm assuming this is a good early generic DAG combine too. The motivation comes from reduced bit-hacking for select-of-constants in IR after rL331486. We want to restore that functionality in the DAG as noted in the commit comments for that change and the llvm-dev discussion here: http://lists.llvm.org/pipermail/llvm-dev/2018-July/124433.html The PPC and AArch tests show that those targets are already doing something similar. x86 will be neutral in the minimal case and generally better when this pattern is extended with other ops as shown in the signbit-shift.ll tests. Note the asymmetry: we don't include the (extend (ifneg X)) transform because it already exists in SimplifySelectCC(), and that is verified in the later unchanged tests in the signbit-shift.ll files. Without the 'not' op, the general transform to use a shift is always a win because that's a single instruction. Alive proofs: https://rise4fun.com/Alive/ysli Name: if pos, get -1 %c = icmp sgt i16 %x, -1 %r = sext i1 %c to i16 => %n = xor i16 %x, -1 %r = ashr i16 %n, 15 Name: if pos, get 1 %c = icmp sgt i16 %x, -1 %r = zext i1 %c to i16 => %n = xor i16 %x, -1 %r = lshr i16 %n, 15 Differential Revision: https://reviews.llvm.org/D48970 llvm-svn: 337130
129 lines
3.7 KiB
LLVM
129 lines
3.7 KiB
LLVM
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu -O2 \
|
|
; RUN: -ppc-gpr-icmps=all -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
|
|
; RUN: --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu -O2 \
|
|
; RUN: -ppc-gpr-icmps=all -ppc-asm-full-reg-names -mcpu=pwr8 < %s | FileCheck %s \
|
|
; RUN: --implicit-check-not cmpw --implicit-check-not cmpd --implicit-check-not cmpl
|
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
@glob = common local_unnamed_addr global i64 0, align 8
|
|
|
|
define signext i32 @test_igesll(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: test_igesll:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: sradi r5, r3, 63
|
|
; CHECK-NEXT: rldicl r6, r4, 1, 63
|
|
; CHECK-NEXT: subfc r3, r4, r3
|
|
; CHECK-NEXT: adde r3, r5, r6
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sge i64 %a, %b
|
|
%conv = zext i1 %cmp to i32
|
|
ret i32 %conv
|
|
}
|
|
|
|
define signext i32 @test_igesll_sext(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: test_igesll_sext:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: sradi r5, r3, 63
|
|
; CHECK-NEXT: rldicl r6, r4, 1, 63
|
|
; CHECK-NEXT: subfc r3, r4, r3
|
|
; CHECK-NEXT: adde r3, r5, r6
|
|
; CHECK-NEXT: neg r3, r3
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sge i64 %a, %b
|
|
%sub = sext i1 %cmp to i32
|
|
ret i32 %sub
|
|
}
|
|
|
|
define signext i32 @test_igesll_z(i64 %a) {
|
|
; CHECK-LABEL: test_igesll_z:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: rldicl r3, r3, 1, 63
|
|
; CHECK-NEXT: xori r3, r3, 1
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sgt i64 %a, -1
|
|
%conv = zext i1 %cmp to i32
|
|
ret i32 %conv
|
|
}
|
|
|
|
define signext i32 @test_igesll_sext_z(i64 %a) {
|
|
; CHECK-LABEL: test_igesll_sext_z:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: sradi r3, r3, 63
|
|
; CHECK-NEXT: not r3, r3
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sgt i64 %a, -1
|
|
%sub = sext i1 %cmp to i32
|
|
ret i32 %sub
|
|
}
|
|
|
|
define void @test_igesll_store(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: test_igesll_store:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK: sradi r6, r3, 63
|
|
; CHECK: subfc r3, r4, r3
|
|
; CHECK: rldicl r3, r4, 1, 63
|
|
; CHECK: adde r3, r6, r3
|
|
; CHECK: std r3
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sge i64 %a, %b
|
|
%conv1 = zext i1 %cmp to i64
|
|
store i64 %conv1, i64* @glob, align 8
|
|
ret void
|
|
}
|
|
|
|
define void @test_igesll_sext_store(i64 %a, i64 %b) {
|
|
; CHECK-LABEL: test_igesll_sext_store:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: sradi r6, r3, 63
|
|
; CHECK-NEXT: addis r5, r2, .LC0@toc@ha
|
|
; CHECK-NEXT: subfc r3, r4, r3
|
|
; CHECK-NEXT: rldicl r3, r4, 1, 63
|
|
; CHECK-NEXT: ld r4, .LC0@toc@l(r5)
|
|
; CHECK-NEXT: adde r3, r6, r3
|
|
; CHECK-NEXT: neg r3, r3
|
|
; CHECK-NEXT: std r3, 0(r4)
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sge i64 %a, %b
|
|
%conv1 = sext i1 %cmp to i64
|
|
store i64 %conv1, i64* @glob, align 8
|
|
ret void
|
|
}
|
|
|
|
define void @test_igesll_z_store(i64 %a) {
|
|
; CHECK-LABEL: test_igesll_z_store:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: addis r4, r2, .LC0@toc@ha
|
|
; CHECK-NEXT: not r3, r3
|
|
; CHECK-NEXT: ld r4, .LC0@toc@l(r4)
|
|
; CHECK-NEXT: rldicl r3, r3, 1, 63
|
|
; CHECK-NEXT: std r3, 0(r4)
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sgt i64 %a, -1
|
|
%conv1 = zext i1 %cmp to i64
|
|
store i64 %conv1, i64* @glob, align 8
|
|
ret void
|
|
}
|
|
|
|
define void @test_igesll_sext_z_store(i64 %a) {
|
|
; CHECK-LABEL: test_igesll_sext_z_store:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: addis r4, r2, .LC0@toc@ha
|
|
; CHECK-NEXT: not r3, r3
|
|
; CHECK-NEXT: ld r4, .LC0@toc@l(r4)
|
|
; CHECK-NEXT: sradi r3, r3, 63
|
|
; CHECK-NEXT: std r3,
|
|
; CHECK-NEXT: blr
|
|
entry:
|
|
%cmp = icmp sgt i64 %a, -1
|
|
%conv1 = sext i1 %cmp to i64
|
|
store i64 %conv1, i64* @glob, align 8
|
|
ret void
|
|
}
|