[llvm] r276743 - [X86][SSE] Added extra memory folding tests for cvtsd2ss intrinsic

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Tue Jul 26 07:44:50 2016
New Revision: 276743

URL: http://llvm.org/viewvc/llvm-project?rev=276743&view=rev
Log:
[X86][SSE] Added extra memory folding tests for cvtsd2ss intrinsic

SSE only fold partial reg update instructions when optsize is enabled

Modified:
    llvm/trunk/test/CodeGen/X86/sse2-intrinsics-x86.ll

Modified: llvm/trunk/test/CodeGen/X86/sse2-intrinsics-x86.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sse2-intrinsics-x86.ll?rev=276743&r1=276742&r2=276743&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sse2-intrinsics-x86.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sse2-intrinsics-x86.ll Tue Jul 26 07:44:50 2016
@@ -293,6 +293,24 @@ define <4 x float> @test_x86_sse2_cvtsd2
 }
 
 
+define <4 x float> @test_x86_sse2_cvtsd2ss_load_optsize(<4 x float> %a0, <2 x double>* %p1) optsize {
+; SSE-LABEL: test_x86_sse2_cvtsd2ss_load_optsize:
+; SSE:       ## BB#0:
+; SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; SSE-NEXT:    cvtsd2ss (%eax), %xmm0
+; SSE-NEXT:    retl
+;
+; KNL-LABEL: test_x86_sse2_cvtsd2ss_load_optsize:
+; KNL:       ## BB#0:
+; KNL-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL-NEXT:    vcvtsd2ss (%eax), %xmm0, %xmm0
+; KNL-NEXT:    retl
+  %a1 = load <2 x double>, <2 x double>* %p1
+  %res = call <4 x float> @llvm.x86.sse2.cvtsd2ss(<4 x float> %a0, <2 x double> %a1) ; <<4 x float>> [#uses=1]
+  ret <4 x float> %res
+}
+
+
 define <2 x double> @test_x86_sse2_cvtsi2sd(<2 x double> %a0, i32 %a1) {
 ; SSE-LABEL: test_x86_sse2_cvtsi2sd:
 ; SSE:       ## BB#0:
@@ -337,6 +355,24 @@ define <2 x double> @test_x86_sse2_cvtss
 ; KNL:       ## BB#0:
 ; KNL-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; KNL-NEXT:    vcvtss2sd (%eax), %xmm0, %xmm0
+; KNL-NEXT:    retl
+  %a1 = load <4 x float>, <4 x float>* %p1
+  %res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]
+  ret <2 x double> %res
+}
+
+
+define <2 x double> @test_x86_sse2_cvtss2sd_load_optsize(<2 x double> %a0, <4 x float>* %p1) optsize {
+; SSE-LABEL: test_x86_sse2_cvtss2sd_load_optsize:
+; SSE:       ## BB#0:
+; SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; SSE-NEXT:    cvtss2sd (%eax), %xmm0
+; SSE-NEXT:    retl
+;
+; KNL-LABEL: test_x86_sse2_cvtss2sd_load_optsize:
+; KNL:       ## BB#0:
+; KNL-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; KNL-NEXT:    vcvtss2sd (%eax), %xmm0, %xmm0
 ; KNL-NEXT:    retl
   %a1 = load <4 x float>, <4 x float>* %p1
   %res = call <2 x double> @llvm.x86.sse2.cvtss2sd(<2 x double> %a0, <4 x float> %a1) ; <<2 x double>> [#uses=1]