1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; Use CPU parameters to ensure that a CPU-specific attribute is not overriding the AVX definition.
4 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s
5 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=corei7-avx | FileCheck %s
6 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=btver2 | FileCheck %s
7 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=-avx | FileCheck %s --check-prefix=SSE
8 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=corei7-avx -mattr=-avx | FileCheck %s --check-prefix=SSE
9 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=btver2 -mattr=-avx | FileCheck %s --check-prefix=SSE
11 ; No need to load unaligned operand from memory using an explicit instruction with AVX.
12 ; The operand should be folded into the AND instr.
14 ; With SSE, folding memory operands into math/logic ops requires 16-byte alignment
15 ; unless specially configured on some CPUs such as AMD Family 10H.
17 define <4 x i32> @test1(<4 x i32>* %p0, <4 x i32> %in1) nounwind {
20 ; CHECK-NEXT: vandps (%rdi), %xmm0, %xmm0
25 ; SSE-NEXT: movups (%rdi), %xmm1
26 ; SSE-NEXT: andps %xmm1, %xmm0
28 %in0 = load <4 x i32>, <4 x i32>* %p0, align 2
29 %a = and <4 x i32> %in0, %in1