[AVX512] Enabling bit logic lowering

[oota-llvm.git] / test / CodeGen / X86 / avx2-arith.ll

diff --git a/test/CodeGen/X86/avx2-arith.ll b/test/CodeGen/X86/avx2-arith.ll
index 2c0b6685e56d482e0fcfc5d9fe35c52cd0ac31f5..72bdd9d04729a9fc67eefb1f134c70244ee4e9a3 100644 (file)
--- a/test/CodeGen/X86/avx2-arith.ll
+++ b/test/CodeGen/X86/avx2-arith.ll
@@ -1,65 +1,66 @@
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=core-avx2 -mattr=+avx2 | FileCheck %s
 
 ; CHECK: vpaddq %ymm
-define <4 x i64> @vpaddq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+define <4 x i64> @test_vpaddq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
   %x = add <4 x i64> %i, %j
   ret <4 x i64> %x
 }
 
 ; CHECK: vpaddd %ymm
-define <8 x i32> @vpaddd(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+define <8 x i32> @test_vpaddd(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
   %x = add <8 x i32> %i, %j
   ret <8 x i32> %x
 }
 
 ; CHECK: vpaddw %ymm
-define <16 x i16> @vpaddw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+define <16 x i16> @test_vpaddw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
   %x = add <16 x i16> %i, %j
   ret <16 x i16> %x
 }
 
 ; CHECK: vpaddb %ymm
-define <32 x i8> @vpaddb(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
+define <32 x i8> @test_vpaddb(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
   %x = add <32 x i8> %i, %j
   ret <32 x i8> %x
 }
 
 ; CHECK: vpsubq %ymm
-define <4 x i64> @vpsubq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+define <4 x i64> @test_vpsubq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
   %x = sub <4 x i64> %i, %j
   ret <4 x i64> %x
 }
 
 ; CHECK: vpsubd %ymm
-define <8 x i32> @vpsubd(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+define <8 x i32> @test_vpsubd(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
   %x = sub <8 x i32> %i, %j
   ret <8 x i32> %x
 }
 
 ; CHECK: vpsubw %ymm
-define <16 x i16> @vpsubw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+define <16 x i16> @test_vpsubw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
   %x = sub <16 x i16> %i, %j
   ret <16 x i16> %x
 }
 
 ; CHECK: vpsubb %ymm
-define <32 x i8> @vpsubb(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
+define <32 x i8> @test_vpsubb(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
   %x = sub <32 x i8> %i, %j
   ret <32 x i8> %x
 }
 
 ; CHECK: vpmulld %ymm
-define <8 x i32> @vpmulld(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+define <8 x i32> @test_vpmulld(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
   %x = mul <8 x i32> %i, %j
   ret <8 x i32> %x
 }
 
 ; CHECK: vpmullw %ymm
-define <16 x i16> @vpmullw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+define <16 x i16> @test_vpmullw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
   %x = mul <16 x i16> %i, %j
   ret <16 x i16> %x
 }
 
+; CHECK: mul-v4i64
 ; CHECK: vpmuludq %ymm
 ; CHECK-NEXT: vpsrlq $32, %ymm
 ; CHECK-NEXT: vpmuludq %ymm
@@ -146,4 +147,22 @@ define <8 x i16> @mul_const8(<8 x i16> %x) {
 define <8 x i32> @mul_const9(<8 x i32> %x) {
   %y = mul <8 x i32> %x, <i32 2, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <8 x i32> %y
-}
\ No newline at end of file
+}
+
+; CHECK: mul_const10
+; CHECK: vpmulld
+; CHECK: ret
+define <4 x i32> @mul_const10(<4 x i32> %x) {
+  ; %x * 0x01010101
+  %m = mul <4 x i32> %x, <i32 16843009, i32 16843009, i32 16843009, i32 16843009>
+  ret <4 x i32> %m
+}
+
+; CHECK: mul_const11
+; CHECK: vpmulld
+; CHECK: ret
+define <4 x i32> @mul_const11(<4 x i32> %x) {
+  ; %x * 0x80808080
+  %m = mul <4 x i32> %x, <i32 2155905152, i32 2155905152, i32 2155905152, i32 2155905152>
+  ret <4 x i32> %m
+}