[llvm] r259420 - [X86][AVX512] Add support for AVX512 VINSERTPS shuffle decoding

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Mon Feb  1 16:05:50 2016
New Revision: 259420

URL: http://llvm.org/viewvc/llvm-project?rev=259420&view=rev
Log:
[X86][AVX512] Add support for AVX512 VINSERTPS shuffle decoding

Modified:
    llvm/trunk/lib/Target/X86/InstPrinter/X86InstComments.cpp
    llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-128.ll
    llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-256.ll

Modified: llvm/trunk/lib/Target/X86/InstPrinter/X86InstComments.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/InstPrinter/X86InstComments.cpp?rev=259420&r1=259419&r2=259420&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/InstPrinter/X86InstComments.cpp (original)
+++ llvm/trunk/lib/Target/X86/InstPrinter/X86InstComments.cpp Mon Feb  1 16:05:50 2016
@@ -278,10 +278,12 @@ bool llvm::EmitAnyX86InstComments(const
 
   case X86::INSERTPSrr:
   case X86::VINSERTPSrr:
+  case X86::VINSERTPSzrr:
     Src2Name = getRegName(MI->getOperand(2).getReg());
     // FALL THROUGH.
   case X86::INSERTPSrm:
   case X86::VINSERTPSrm:
+  case X86::VINSERTPSzrm:
     DestName = getRegName(MI->getOperand(0).getReg());
     Src1Name = getRegName(MI->getOperand(1).getReg());
     if (MI->getOperand(MI->getNumOperands() - 1).isImm())

Modified: llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-128.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-128.ll?rev=259420&r1=259419&r2=259420&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-128.ll (original)
+++ llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-128.ll Mon Feb  1 16:05:50 2016
@@ -188,23 +188,11 @@ define <4 x float> @merge_4f32_f32_012u(
 ; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
 ; SSE41-NEXT:    retq
 ;
-; AVX1-LABEL: merge_4f32_f32_012u:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: merge_4f32_f32_012u:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX2-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
-; AVX2-NEXT:    retq
-;
-; AVX512F-LABEL: merge_4f32_f32_012u:
-; AVX512F:       # BB#0:
-; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX512F-NEXT:    vinsertps $32, 8(%rdi), %xmm0, %xmm0
-; AVX512F-NEXT:    retq
+; AVX-LABEL: merge_4f32_f32_012u:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
+; AVX-NEXT:    retq
   %ptr0 = getelementptr inbounds float, float* %ptr, i64 0
   %ptr1 = getelementptr inbounds float, float* %ptr, i64 1
   %ptr2 = getelementptr inbounds float, float* %ptr, i64 2
@@ -234,23 +222,11 @@ define <4 x float> @merge_4f32_f32_019u(
 ; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
 ; SSE41-NEXT:    retq
 ;
-; AVX1-LABEL: merge_4f32_f32_019u:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX1-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: merge_4f32_f32_019u:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX2-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
-; AVX2-NEXT:    retq
-;
-; AVX512F-LABEL: merge_4f32_f32_019u:
-; AVX512F:       # BB#0:
-; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX512F-NEXT:    vinsertps $32, 36(%rdi), %xmm0, %xmm0
-; AVX512F-NEXT:    retq
+; AVX-LABEL: merge_4f32_f32_019u:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],mem[0],xmm0[3]
+; AVX-NEXT:    retq
   %ptr0 = getelementptr inbounds float, float* %ptr, i64 0
   %ptr1 = getelementptr inbounds float, float* %ptr, i64 1
   %ptr2 = getelementptr inbounds float, float* %ptr, i64 9

Modified: llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-256.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-256.ll?rev=259420&r1=259419&r2=259420&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-256.ll (original)
+++ llvm/trunk/test/CodeGen/X86/merge-consecutive-loads-256.ll Mon Feb  1 16:05:50 2016
@@ -319,7 +319,7 @@ define <8 x float> @merge_8f32_f32_1u3u5
 ; AVX512F-NEXT:    vmovss 20(%rdi), %xmm1
 ; AVX512F-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,1],xmm0[1,0]
 ; AVX512F-NEXT:    vmovss 4(%rdi), %xmm1
-; AVX512F-NEXT:    vinsertps $32, 12(%rdi), %xmm1, %xmm1
+; AVX512F-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
 ; AVX512F-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
 ; AVX512F-NEXT:    retq
   %ptr0 = getelementptr inbounds float, float* %ptr, i64 1