c9e23187cb
X86 doesn't have absdiff instructions, but we should be able to canonicalize to avoid unnecessary extensions/selects
425 lines
12 KiB
LLVM
425 lines
12 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=i686 -mattr=cmov | FileCheck %s --check-prefixes=X86
|
|
; RUN: llc < %s -mtriple=x86_64-linux | FileCheck %s --check-prefixes=X64
|
|
|
|
;
|
|
; trunc(abs(sub(zext(a),zext(b)))) -> abdu(a,b)
|
|
;
|
|
|
|
define i8 @abd_ext_i8(i8 %a, i8 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i8:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: subl %eax, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: negl %eax
|
|
; X86-NEXT: cmovsl %ecx, %eax
|
|
; X86-NEXT: # kill: def $al killed $al killed $eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i8:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: # kill: def $esi killed $esi def $rsi
|
|
; X64-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; X64-NEXT: movzbl %dil, %ecx
|
|
; X64-NEXT: movzbl %sil, %eax
|
|
; X64-NEXT: subq %rax, %rcx
|
|
; X64-NEXT: movq %rcx, %rax
|
|
; X64-NEXT: negq %rax
|
|
; X64-NEXT: cmovsq %rcx, %rax
|
|
; X64-NEXT: # kill: def $al killed $al killed $rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i8 %a to i64
|
|
%bext = zext i8 %b to i64
|
|
%sub = sub i64 %aext, %bext
|
|
%abs = call i64 @llvm.abs.i64(i64 %sub, i1 false)
|
|
%trunc = trunc i64 %abs to i8
|
|
ret i8 %trunc
|
|
}
|
|
|
|
define i8 @abd_ext_i8_undef(i8 %a, i8 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i8_undef:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: subl %eax, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: negl %eax
|
|
; X86-NEXT: cmovsl %ecx, %eax
|
|
; X86-NEXT: # kill: def $al killed $al killed $eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i8_undef:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: # kill: def $esi killed $esi def $rsi
|
|
; X64-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; X64-NEXT: movzbl %dil, %ecx
|
|
; X64-NEXT: movzbl %sil, %eax
|
|
; X64-NEXT: subq %rax, %rcx
|
|
; X64-NEXT: movq %rcx, %rax
|
|
; X64-NEXT: negq %rax
|
|
; X64-NEXT: cmovsq %rcx, %rax
|
|
; X64-NEXT: # kill: def $al killed $al killed $rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i8 %a to i64
|
|
%bext = zext i8 %b to i64
|
|
%sub = sub i64 %aext, %bext
|
|
%abs = call i64 @llvm.abs.i64(i64 %sub, i1 true)
|
|
%trunc = trunc i64 %abs to i8
|
|
ret i8 %trunc
|
|
}
|
|
|
|
define i16 @abd_ext_i16(i16 %a, i16 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i16:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: subl %eax, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: negl %eax
|
|
; X86-NEXT: cmovsl %ecx, %eax
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i16:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: # kill: def $esi killed $esi def $rsi
|
|
; X64-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; X64-NEXT: movzwl %di, %ecx
|
|
; X64-NEXT: movzwl %si, %eax
|
|
; X64-NEXT: subq %rax, %rcx
|
|
; X64-NEXT: movq %rcx, %rax
|
|
; X64-NEXT: negq %rax
|
|
; X64-NEXT: cmovsq %rcx, %rax
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i16 %a to i64
|
|
%bext = zext i16 %b to i64
|
|
%sub = sub i64 %aext, %bext
|
|
%abs = call i64 @llvm.abs.i64(i64 %sub, i1 false)
|
|
%trunc = trunc i64 %abs to i16
|
|
ret i16 %trunc
|
|
}
|
|
|
|
define i16 @abd_ext_i16_undef(i16 %a, i16 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i16_undef:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: subl %eax, %ecx
|
|
; X86-NEXT: movl %ecx, %eax
|
|
; X86-NEXT: negl %eax
|
|
; X86-NEXT: cmovsl %ecx, %eax
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i16_undef:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: # kill: def $esi killed $esi def $rsi
|
|
; X64-NEXT: # kill: def $edi killed $edi def $rdi
|
|
; X64-NEXT: movzwl %di, %ecx
|
|
; X64-NEXT: movzwl %si, %eax
|
|
; X64-NEXT: subq %rax, %rcx
|
|
; X64-NEXT: movq %rcx, %rax
|
|
; X64-NEXT: negq %rax
|
|
; X64-NEXT: cmovsq %rcx, %rax
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i16 %a to i64
|
|
%bext = zext i16 %b to i64
|
|
%sub = sub i64 %aext, %bext
|
|
%abs = call i64 @llvm.abs.i64(i64 %sub, i1 true)
|
|
%trunc = trunc i64 %abs to i16
|
|
ret i16 %trunc
|
|
}
|
|
|
|
define i32 @abd_ext_i32(i32 %a, i32 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i32:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: xorl %ecx, %ecx
|
|
; X86-NEXT: subl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: sbbl %ecx, %ecx
|
|
; X86-NEXT: sarl $31, %ecx
|
|
; X86-NEXT: xorl %ecx, %eax
|
|
; X86-NEXT: subl %ecx, %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i32:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edi, %ecx
|
|
; X64-NEXT: movl %esi, %eax
|
|
; X64-NEXT: subq %rax, %rcx
|
|
; X64-NEXT: movq %rcx, %rax
|
|
; X64-NEXT: negq %rax
|
|
; X64-NEXT: cmovsq %rcx, %rax
|
|
; X64-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i32 %a to i64
|
|
%bext = zext i32 %b to i64
|
|
%sub = sub i64 %aext, %bext
|
|
%abs = call i64 @llvm.abs.i64(i64 %sub, i1 false)
|
|
%trunc = trunc i64 %abs to i32
|
|
ret i32 %trunc
|
|
}
|
|
|
|
define i32 @abd_ext_i32_undef(i32 %a, i32 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i32_undef:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: xorl %ecx, %ecx
|
|
; X86-NEXT: subl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: sbbl %ecx, %ecx
|
|
; X86-NEXT: sarl $31, %ecx
|
|
; X86-NEXT: xorl %ecx, %eax
|
|
; X86-NEXT: subl %ecx, %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i32_undef:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %edi, %ecx
|
|
; X64-NEXT: movl %esi, %eax
|
|
; X64-NEXT: subq %rax, %rcx
|
|
; X64-NEXT: movq %rcx, %rax
|
|
; X64-NEXT: negq %rax
|
|
; X64-NEXT: cmovsq %rcx, %rax
|
|
; X64-NEXT: # kill: def $eax killed $eax killed $rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i32 %a to i64
|
|
%bext = zext i32 %b to i64
|
|
%sub = sub i64 %aext, %bext
|
|
%abs = call i64 @llvm.abs.i64(i64 %sub, i1 true)
|
|
%trunc = trunc i64 %abs to i32
|
|
ret i32 %trunc
|
|
}
|
|
|
|
define i64 @abd_ext_i64(i64 %a, i64 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i64:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: pushl %esi
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: xorl %ecx, %ecx
|
|
; X86-NEXT: subl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: sbbl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: movl $0, %esi
|
|
; X86-NEXT: sbbl %esi, %esi
|
|
; X86-NEXT: sbbl %ecx, %ecx
|
|
; X86-NEXT: sarl $31, %ecx
|
|
; X86-NEXT: xorl %ecx, %edx
|
|
; X86-NEXT: xorl %ecx, %eax
|
|
; X86-NEXT: subl %ecx, %eax
|
|
; X86-NEXT: sbbl %ecx, %edx
|
|
; X86-NEXT: popl %esi
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i64:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movq %rdi, %rax
|
|
; X64-NEXT: xorl %ecx, %ecx
|
|
; X64-NEXT: subq %rsi, %rax
|
|
; X64-NEXT: sbbq %rcx, %rcx
|
|
; X64-NEXT: sarq $63, %rcx
|
|
; X64-NEXT: xorq %rcx, %rax
|
|
; X64-NEXT: subq %rcx, %rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i64 %a to i128
|
|
%bext = zext i64 %b to i128
|
|
%sub = sub i128 %aext, %bext
|
|
%abs = call i128 @llvm.abs.i128(i128 %sub, i1 false)
|
|
%trunc = trunc i128 %abs to i64
|
|
ret i64 %trunc
|
|
}
|
|
|
|
define i64 @abd_ext_i64_undef(i64 %a, i64 %b) nounwind {
|
|
; X86-LABEL: abd_ext_i64_undef:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: pushl %esi
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: xorl %ecx, %ecx
|
|
; X86-NEXT: subl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: sbbl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: movl $0, %esi
|
|
; X86-NEXT: sbbl %esi, %esi
|
|
; X86-NEXT: sbbl %ecx, %ecx
|
|
; X86-NEXT: sarl $31, %ecx
|
|
; X86-NEXT: xorl %ecx, %edx
|
|
; X86-NEXT: xorl %ecx, %eax
|
|
; X86-NEXT: subl %ecx, %eax
|
|
; X86-NEXT: sbbl %ecx, %edx
|
|
; X86-NEXT: popl %esi
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_ext_i64_undef:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movq %rdi, %rax
|
|
; X64-NEXT: xorl %ecx, %ecx
|
|
; X64-NEXT: subq %rsi, %rax
|
|
; X64-NEXT: sbbq %rcx, %rcx
|
|
; X64-NEXT: sarq $63, %rcx
|
|
; X64-NEXT: xorq %rcx, %rax
|
|
; X64-NEXT: subq %rcx, %rax
|
|
; X64-NEXT: retq
|
|
%aext = zext i64 %a to i128
|
|
%bext = zext i64 %b to i128
|
|
%sub = sub i128 %aext, %bext
|
|
%abs = call i128 @llvm.abs.i128(i128 %sub, i1 true)
|
|
%trunc = trunc i128 %abs to i64
|
|
ret i64 %trunc
|
|
}
|
|
|
|
;
|
|
; sub(umax(a,b),umin(a,b)) -> abdu(a,b)
|
|
;
|
|
|
|
define i8 @abd_minmax_i8(i8 %a, i8 %b) nounwind {
|
|
; X86-LABEL: abd_minmax_i8:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: cmpb %al, %cl
|
|
; X86-NEXT: movl %eax, %edx
|
|
; X86-NEXT: cmovbl %ecx, %edx
|
|
; X86-NEXT: cmoval %ecx, %eax
|
|
; X86-NEXT: subb %dl, %al
|
|
; X86-NEXT: # kill: def $al killed $al killed $eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_minmax_i8:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %esi, %eax
|
|
; X64-NEXT: cmpb %al, %dil
|
|
; X64-NEXT: movl %esi, %ecx
|
|
; X64-NEXT: cmovbl %edi, %ecx
|
|
; X64-NEXT: cmoval %edi, %eax
|
|
; X64-NEXT: subb %cl, %al
|
|
; X64-NEXT: # kill: def $al killed $al killed $eax
|
|
; X64-NEXT: retq
|
|
%min = call i8 @llvm.umin.i8(i8 %a, i8 %b)
|
|
%max = call i8 @llvm.umax.i8(i8 %a, i8 %b)
|
|
%sub = sub i8 %max, %min
|
|
ret i8 %sub
|
|
}
|
|
|
|
define i16 @abd_minmax_i16(i16 %a, i16 %b) nounwind {
|
|
; X86-LABEL: abd_minmax_i16:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: cmpw %ax, %cx
|
|
; X86-NEXT: movl %eax, %edx
|
|
; X86-NEXT: cmovbl %ecx, %edx
|
|
; X86-NEXT: cmoval %ecx, %eax
|
|
; X86-NEXT: subl %edx, %eax
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_minmax_i16:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %esi, %eax
|
|
; X64-NEXT: cmpw %ax, %di
|
|
; X64-NEXT: movl %esi, %ecx
|
|
; X64-NEXT: cmovbl %edi, %ecx
|
|
; X64-NEXT: cmoval %edi, %eax
|
|
; X64-NEXT: subl %ecx, %eax
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $eax
|
|
; X64-NEXT: retq
|
|
%min = call i16 @llvm.umin.i16(i16 %a, i16 %b)
|
|
%max = call i16 @llvm.umax.i16(i16 %a, i16 %b)
|
|
%sub = sub i16 %max, %min
|
|
ret i16 %sub
|
|
}
|
|
|
|
define i32 @abd_minmax_i32(i32 %a, i32 %b) nounwind {
|
|
; X86-LABEL: abd_minmax_i32:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: cmpl %eax, %ecx
|
|
; X86-NEXT: movl %eax, %edx
|
|
; X86-NEXT: cmovbl %ecx, %edx
|
|
; X86-NEXT: cmoval %ecx, %eax
|
|
; X86-NEXT: subl %edx, %eax
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_minmax_i32:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movl %esi, %eax
|
|
; X64-NEXT: cmpl %esi, %edi
|
|
; X64-NEXT: movl %esi, %ecx
|
|
; X64-NEXT: cmovbl %edi, %ecx
|
|
; X64-NEXT: cmoval %edi, %eax
|
|
; X64-NEXT: subl %ecx, %eax
|
|
; X64-NEXT: retq
|
|
%min = call i32 @llvm.umin.i32(i32 %a, i32 %b)
|
|
%max = call i32 @llvm.umax.i32(i32 %a, i32 %b)
|
|
%sub = sub i32 %max, %min
|
|
ret i32 %sub
|
|
}
|
|
|
|
define i64 @abd_minmax_i64(i64 %a, i64 %b) nounwind {
|
|
; X86-LABEL: abd_minmax_i64:
|
|
; X86: # %bb.0:
|
|
; X86-NEXT: pushl %ebp
|
|
; X86-NEXT: pushl %ebx
|
|
; X86-NEXT: pushl %edi
|
|
; X86-NEXT: pushl %esi
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %esi
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
|
; X86-NEXT: cmpl %eax, %esi
|
|
; X86-NEXT: movl %eax, %edi
|
|
; X86-NEXT: cmovbl %esi, %edi
|
|
; X86-NEXT: movl %eax, %ebx
|
|
; X86-NEXT: cmoval %esi, %ebx
|
|
; X86-NEXT: cmpl %edx, %ecx
|
|
; X86-NEXT: movl %eax, %ebp
|
|
; X86-NEXT: cmovbl %esi, %ebp
|
|
; X86-NEXT: cmovel %edi, %ebp
|
|
; X86-NEXT: movl %edx, %edi
|
|
; X86-NEXT: cmovbl %ecx, %edi
|
|
; X86-NEXT: cmoval %esi, %eax
|
|
; X86-NEXT: cmovel %ebx, %eax
|
|
; X86-NEXT: cmoval %ecx, %edx
|
|
; X86-NEXT: subl %ebp, %eax
|
|
; X86-NEXT: sbbl %edi, %edx
|
|
; X86-NEXT: popl %esi
|
|
; X86-NEXT: popl %edi
|
|
; X86-NEXT: popl %ebx
|
|
; X86-NEXT: popl %ebp
|
|
; X86-NEXT: retl
|
|
;
|
|
; X64-LABEL: abd_minmax_i64:
|
|
; X64: # %bb.0:
|
|
; X64-NEXT: movq %rsi, %rax
|
|
; X64-NEXT: cmpq %rsi, %rdi
|
|
; X64-NEXT: movq %rsi, %rcx
|
|
; X64-NEXT: cmovbq %rdi, %rcx
|
|
; X64-NEXT: cmovaq %rdi, %rax
|
|
; X64-NEXT: subq %rcx, %rax
|
|
; X64-NEXT: retq
|
|
%min = call i64 @llvm.umin.i64(i64 %a, i64 %b)
|
|
%max = call i64 @llvm.umax.i64(i64 %a, i64 %b)
|
|
%sub = sub i64 %max, %min
|
|
ret i64 %sub
|
|
}
|
|
|
|
declare i8 @llvm.abs.i8(i8, i1)
|
|
declare i16 @llvm.abs.i16(i16, i1)
|
|
declare i32 @llvm.abs.i32(i32, i1)
|
|
declare i64 @llvm.abs.i64(i64, i1)
|
|
declare i128 @llvm.abs.i128(i128, i1)
|
|
|
|
declare i8 @llvm.umax.i8(i8, i8)
|
|
declare i16 @llvm.umax.i16(i16, i16)
|
|
declare i32 @llvm.umax.i32(i32, i32)
|
|
declare i64 @llvm.umax.i64(i64, i64)
|
|
|
|
declare i8 @llvm.umin.i8(i8, i8)
|
|
declare i16 @llvm.umin.i16(i16, i16)
|
|
declare i32 @llvm.umin.i32(i32, i32)
|
|
declare i64 @llvm.umin.i64(i64, i64)
|