[llvm] r289035 - [X86][SSE] Add vector test for (shl (or x, c1), c2) -> (or (shl x, c2), c1 << c2) detailed in D19325

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Thu Dec 8 02:17:26 PST 2016


Author: rksimon
Date: Thu Dec  8 04:17:25 2016
New Revision: 289035

URL: http://llvm.org/viewvc/llvm-project?rev=289035&view=rev
Log:
[X86][SSE] Add vector test for (shl (or x, c1), c2) -> (or (shl x, c2), c1 << c2) detailed in D19325

Modified:
    llvm/trunk/test/CodeGen/X86/combine-shl.ll

Modified: llvm/trunk/test/CodeGen/X86/combine-shl.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/combine-shl.ll?rev=289035&r1=289034&r2=289035&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/combine-shl.ll (original)
+++ llvm/trunk/test/CodeGen/X86/combine-shl.ll Thu Dec  8 04:17:25 2016
@@ -543,6 +543,42 @@ define <4 x i32> @combine_vec_shl_add1(<
   ret <4 x i32> %2
 }
 
+; FIXME: fold (shl (or x, c1), c2) -> (or (shl x, c2), c1 << c2)
+define <4 x i32> @combine_vec_shl_or0(<4 x i32> %x) {
+; SSE-LABEL: combine_vec_shl_or0:
+; SSE:       # BB#0:
+; SSE-NEXT:    por {{.*}}(%rip), %xmm0
+; SSE-NEXT:    pslld $2, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: combine_vec_shl_or0:
+; AVX:       # BB#0:
+; AVX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm1
+; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpslld $2, %xmm0, %xmm0
+; AVX-NEXT:    retq
+  %1 = or  <4 x i32> %x, <i32 5, i32 5, i32 5, i32 5>
+  %2 = shl <4 x i32> %1, <i32 2, i32 2, i32 2, i32 2>
+  ret <4 x i32> %2
+}
+
+define <4 x i32> @combine_vec_shl_or1(<4 x i32> %x) {
+; SSE-LABEL: combine_vec_shl_or1:
+; SSE:       # BB#0:
+; SSE-NEXT:    por {{.*}}(%rip), %xmm0
+; SSE-NEXT:    pmulld {{.*}}(%rip), %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: combine_vec_shl_or1:
+; AVX:       # BB#0:
+; AVX-NEXT:    vpor {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vpsllvd {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    retq
+  %1 = or  <4 x i32> %x, <i32 5, i32 6, i32 7, i32 8>
+  %2 = shl <4 x i32> %1, <i32 1, i32 2, i32 3, i32 4>
+  ret <4 x i32> %2
+}
+
 ; fold (shl (mul x, c1), c2) -> (mul x, c1 << c2)
 define <4 x i32> @combine_vec_shl_mul0(<4 x i32> %x) {
 ; SSE-LABEL: combine_vec_shl_mul0:




More information about the llvm-commits mailing list