mirror of
https://github.com/RPCS3/llvm.git
synced 2024-12-13 14:47:00 +00:00
1f4a5003d2
xor/and/or (bitcast(A), bitcast(B)) -> bitcast(op (A,B)) was only being combined at the AfterLegalizeTypes stage, this patch permits the combine to occur anytime before then as well. The main aim with this to improve the ability to recognise bitmasks that can be converted to shuffles. I had to modify a number of AVX512 mask tests as the basic bitcast to/from scalar pattern was being stripped out, preventing testing of the mmask bitops. By replacing the bitcasts with loads we can get almost the same result. Differential Revision: http://reviews.llvm.org/D18944 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@265998 91177308-0d34-0410-b5e6-96231b3b80d8
236 lines
6.2 KiB
LLVM
236 lines
6.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=X32-SSE --check-prefix=X32-SSE42
|
|
; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse4.2 | FileCheck %s --check-prefix=X64-SSE --check-prefix=X64-SSE42
|
|
|
|
;
|
|
; AND/XOR/OR i32 as v4i8
|
|
;
|
|
|
|
define i32 @and_i32_as_v4i8(i32 %a, i32 %b) nounwind {
|
|
; X32-SSE-LABEL: and_i32_as_v4i8:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: andl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: and_i32_as_v4i8:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: andl %esi, %edi
|
|
; X64-SSE-NEXT: movl %edi, %eax
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast i32 %a to <4 x i8>
|
|
%2 = bitcast i32 %b to <4 x i8>
|
|
%3 = and <4 x i8> %1, %2
|
|
%4 = bitcast <4 x i8> %3 to i32
|
|
ret i32 %4
|
|
}
|
|
|
|
define i32 @xor_i32_as_v4i8(i32 %a, i32 %b) nounwind {
|
|
; X32-SSE-LABEL: xor_i32_as_v4i8:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: xorl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: xor_i32_as_v4i8:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: xorl %esi, %edi
|
|
; X64-SSE-NEXT: movl %edi, %eax
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast i32 %a to <4 x i8>
|
|
%2 = bitcast i32 %b to <4 x i8>
|
|
%3 = xor <4 x i8> %1, %2
|
|
%4 = bitcast <4 x i8> %3 to i32
|
|
ret i32 %4
|
|
}
|
|
|
|
define i32 @or_i32_as_v4i8(i32 %a, i32 %b) nounwind {
|
|
; X32-SSE-LABEL: or_i32_as_v4i8:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: orl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: or_i32_as_v4i8:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: orl %esi, %edi
|
|
; X64-SSE-NEXT: movl %edi, %eax
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast i32 %a to <4 x i8>
|
|
%2 = bitcast i32 %b to <4 x i8>
|
|
%3 = or <4 x i8> %1, %2
|
|
%4 = bitcast <4 x i8> %3 to i32
|
|
ret i32 %4
|
|
}
|
|
|
|
;
|
|
; AND/XOR/OR i32 as v8i4
|
|
;
|
|
|
|
define i32 @and_i32_as_v8i4(i32 %a, i32 %b) nounwind {
|
|
; X32-SSE-LABEL: and_i32_as_v8i4:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: andl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: and_i32_as_v8i4:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: andl %esi, %edi
|
|
; X64-SSE-NEXT: movl %edi, %eax
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast i32 %a to <8 x i4>
|
|
%2 = bitcast i32 %b to <8 x i4>
|
|
%3 = and <8 x i4> %1, %2
|
|
%4 = bitcast <8 x i4> %3 to i32
|
|
ret i32 %4
|
|
}
|
|
|
|
define i32 @xor_i32_as_v8i4(i32 %a, i32 %b) nounwind {
|
|
; X32-SSE-LABEL: xor_i32_as_v8i4:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: xorl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: xor_i32_as_v8i4:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: xorl %esi, %edi
|
|
; X64-SSE-NEXT: movl %edi, %eax
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast i32 %a to <8 x i4>
|
|
%2 = bitcast i32 %b to <8 x i4>
|
|
%3 = xor <8 x i4> %1, %2
|
|
%4 = bitcast <8 x i4> %3 to i32
|
|
ret i32 %4
|
|
}
|
|
|
|
define i32 @or_i32_as_v8i4(i32 %a, i32 %b) nounwind {
|
|
; X32-SSE-LABEL: or_i32_as_v8i4:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: orl {{[0-9]+}}(%esp), %eax
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: or_i32_as_v8i4:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: orl %esi, %edi
|
|
; X64-SSE-NEXT: movl %edi, %eax
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast i32 %a to <8 x i4>
|
|
%2 = bitcast i32 %b to <8 x i4>
|
|
%3 = or <8 x i4> %1, %2
|
|
%4 = bitcast <8 x i4> %3 to i32
|
|
ret i32 %4
|
|
}
|
|
|
|
;
|
|
; AND/XOR/OR v4i8 as i32
|
|
;
|
|
|
|
define <4 x i8> @and_v4i8_as_i32(<4 x i8> %a, <4 x i8> %b) nounwind {
|
|
; X32-SSE-LABEL: and_v4i8_as_i32:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: andps %xmm1, %xmm0
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: and_v4i8_as_i32:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: andps %xmm1, %xmm0
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast <4 x i8> %a to i32
|
|
%2 = bitcast <4 x i8> %b to i32
|
|
%3 = and i32 %1, %2
|
|
%4 = bitcast i32 %3 to <4 x i8>
|
|
ret <4 x i8> %4
|
|
}
|
|
|
|
define <4 x i8> @xor_v4i8_as_i32(<4 x i8> %a, <4 x i8> %b) nounwind {
|
|
; X32-SSE-LABEL: xor_v4i8_as_i32:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: xorps %xmm1, %xmm0
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: xor_v4i8_as_i32:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: xorps %xmm1, %xmm0
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast <4 x i8> %a to i32
|
|
%2 = bitcast <4 x i8> %b to i32
|
|
%3 = xor i32 %1, %2
|
|
%4 = bitcast i32 %3 to <4 x i8>
|
|
ret <4 x i8> %4
|
|
}
|
|
|
|
define <4 x i8> @or_v4i8_as_i32(<4 x i8> %a, <4 x i8> %b) nounwind {
|
|
; X32-SSE-LABEL: or_v4i8_as_i32:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: orps %xmm1, %xmm0
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: or_v4i8_as_i32:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: orps %xmm1, %xmm0
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast <4 x i8> %a to i32
|
|
%2 = bitcast <4 x i8> %b to i32
|
|
%3 = or i32 %1, %2
|
|
%4 = bitcast i32 %3 to <4 x i8>
|
|
ret <4 x i8> %4
|
|
}
|
|
|
|
;
|
|
; AND/XOR/OR v8i4 as i32
|
|
;
|
|
|
|
define <8 x i4> @and_v8i4_as_i32(<8 x i4> %a, <8 x i4> %b) nounwind {
|
|
; X32-SSE-LABEL: and_v8i4_as_i32:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: andps %xmm1, %xmm0
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: and_v8i4_as_i32:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: andps %xmm1, %xmm0
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast <8 x i4> %a to i32
|
|
%2 = bitcast <8 x i4> %b to i32
|
|
%3 = and i32 %1, %2
|
|
%4 = bitcast i32 %3 to <8 x i4>
|
|
ret <8 x i4> %4
|
|
}
|
|
|
|
define <8 x i4> @xor_v8i4_as_i32(<8 x i4> %a, <8 x i4> %b) nounwind {
|
|
; X32-SSE-LABEL: xor_v8i4_as_i32:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: xorps %xmm1, %xmm0
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: xor_v8i4_as_i32:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: xorps %xmm1, %xmm0
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast <8 x i4> %a to i32
|
|
%2 = bitcast <8 x i4> %b to i32
|
|
%3 = xor i32 %1, %2
|
|
%4 = bitcast i32 %3 to <8 x i4>
|
|
ret <8 x i4> %4
|
|
}
|
|
|
|
define <8 x i4> @or_v8i4_as_i32(<8 x i4> %a, <8 x i4> %b) nounwind {
|
|
; X32-SSE-LABEL: or_v8i4_as_i32:
|
|
; X32-SSE: # BB#0:
|
|
; X32-SSE-NEXT: orps %xmm1, %xmm0
|
|
; X32-SSE-NEXT: retl
|
|
;
|
|
; X64-SSE-LABEL: or_v8i4_as_i32:
|
|
; X64-SSE: # BB#0:
|
|
; X64-SSE-NEXT: orps %xmm1, %xmm0
|
|
; X64-SSE-NEXT: retq
|
|
%1 = bitcast <8 x i4> %a to i32
|
|
%2 = bitcast <8 x i4> %b to i32
|
|
%3 = or i32 %1, %2
|
|
%4 = bitcast i32 %3 to <8 x i4>
|
|
ret <8 x i4> %4
|
|
}
|