[X86] Finishing broadcast[f|i]32x2 intrinsics lowering to IR.

[uriel.k]

Summary

Removing X86 broadcast(f/i)32x2 intrinsics from llvm.
Adding autoUpgrade support.
Moving matching tests from avx512dq-intrinsics.ll to avx512dq-intrinsics-upgrade.ll and from avx512dqvl-intrinsics.ll to avx512dqvl-intrinsics-upgrade.ll.

[m_zuckerman]

lib/IR/AutoUpgrade.cpp

250

can you use the "avx512.mask.broadcastf" and "avx512.mask.broadcasti" instead of adding a new line?

253

same as above

[uriel.k]

change after Michael comment

[m_zuckerman]

LGTM THANKS ,

[uriel.k]

committed in 314195

include/llvm/IR/IntrinsicsX86.td

 let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   // TODO: Remove the broadcast intrinsics with no gcc builtin and autoupgrade
   def int_x86_avx512_vbroadcast_ss_512 :
         Intrinsic<[llvm_v16f32_ty], [llvm_ptr_ty], [IntrReadMem, IntrArgMemOnly]>;
 
   def int_x86_avx512_vbroadcast_sd_512 :
         Intrinsic<[llvm_v8f64_ty], [llvm_ptr_ty], [IntrReadMem, IntrArgMemOnly]>;
 
-  def int_x86_avx512_mask_broadcastf32x2_256 :
-          Intrinsic<[llvm_v8f32_ty],
-                    [llvm_v4f32_ty, llvm_v8f32_ty, llvm_i8_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcastf32x2_512 :
-          Intrinsic<[llvm_v16f32_ty],
-                    [llvm_v4f32_ty, llvm_v16f32_ty, llvm_i16_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcasti32x2_128 :
-          Intrinsic<[llvm_v4i32_ty],
-                    [llvm_v4i32_ty, llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcasti32x2_256 :
-          Intrinsic<[llvm_v8i32_ty],
-                    [llvm_v4i32_ty, llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcasti32x2_512 :
-          Intrinsic<[llvm_v16i32_ty],
-                    [llvm_v4i32_ty, llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
-
    def int_x86_avx512_broadcastmw_512 :
           GCCBuiltin<"__builtin_ia32_broadcastmw512">,
           Intrinsic<[llvm_v16i32_ty], [llvm_i16_ty], [IntrNoMem]>;
    def int_x86_avx512_broadcastmw_256 :
           GCCBuiltin<"__builtin_ia32_broadcastmw256">,
           Intrinsic<[llvm_v8i32_ty], [llvm_i16_ty], [IntrNoMem]>;
    def int_x86_avx512_broadcastmw_128 :
           GCCBuiltin<"__builtin_ia32_broadcastmw128">,

lib/IR/AutoUpgrade.cpp

if (Name=="ssse3.pabs.b.128" || // Added in 6.0
       Name.startswith("avx512.psrl.dq") || // Added in 3.9
       Name == "sse41.pblendw" || // Added in 3.7
       Name.startswith("sse41.blendp") || // Added in 3.7
       Name.startswith("avx.blend.p") || // Added in 3.7
       Name == "avx2.pblendw" || // Added in 3.7
       Name.startswith("avx2.pblendd.") || // Added in 3.7
       Name.startswith("avx.vbroadcastf128") || // Added in 4.0
       Name == "avx2.vbroadcasti128" || // Added in 3.7
+      Name.startswith("avx512.mask.broadcastf32x2.") || // Added in 6.0
       Name.startswith("avx512.mask.broadcastf32x4.") || // Added in 6.0
       Name.startswith("avx512.mask.broadcastf64x2.") || // Added in 6.0
+      Name.startswith("avx512.mask.broadcasti32x2.") || // Added in 6.0
       Name.startswith("avx512.mask.broadcasti32x4.") || // Added in 6.0
       Name.startswith("avx512.mask.broadcasti64x2.") || // Added in 6.0
       Name == "avx512.mask.broadcastf32x8.512" || // Added in 6.0
       Name == "avx512.mask.broadcasti32x8.512" || // Added in 6.0
       Name == "avx512.mask.broadcastf64x4.512" || // Added in 6.0
       Name == "avx512.mask.broadcasti64x4.512" || // Added in 6.0
       Name == "xop.vpcmov" || // Added in 3.8
       Name == "xop.vpcmov.256" || // Added in 5.0

lib/Target/X86/X86IntrinsicsInfo.h

static const IntrinsicData  IntrinsicsWithoutChain[] = {
   X86_INTRINSIC_DATA(avx512_mask_add_pd_512, INTR_TYPE_2OP_MASK, ISD::FADD,
   X86ISD::FADD_RND),
   X86_INTRINSIC_DATA(avx512_mask_add_ps_512, INTR_TYPE_2OP_MASK, ISD::FADD,
   X86ISD::FADD_RND),
   X86_INTRINSIC_DATA(avx512_mask_add_sd_round, INTR_TYPE_SCALAR_MASK_RM,
                      X86ISD::FADDS_RND, 0),
   X86_INTRINSIC_DATA(avx512_mask_add_ss_round, INTR_TYPE_SCALAR_MASK_RM,
                      X86ISD::FADDS_RND, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcastf32x2_256, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcastf32x2_512, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcasti32x2_128, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcasti32x2_256, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcasti32x2_512, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_pd_128, CMP_MASK_CC, X86ISD::CMPM, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_pd_256, CMP_MASK_CC, X86ISD::CMPM, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_pd_512, CMP_MASK_CC, X86ISD::CMPM,
                      X86ISD::CMPM_RND),
   X86_INTRINSIC_DATA(avx512_mask_cmp_ps_128, CMP_MASK_CC, X86ISD::CMPM, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_ps_256, CMP_MASK_CC, X86ISD::CMPM, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_ps_512, CMP_MASK_CC, X86ISD::CMPM,
                      X86ISD::CMPM_RND),

test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll

 ; CHECK-NEXT:    kmovw %esi, %k1
 ; CHECK-NEXT:    vbroadcasti64x2 {{.*#+}} zmm0 {%k1} = mem[0,1,0,1,0,1,0,1]
 ; CHECK-NEXT:    retq
 
   %x0 = load <2 x i64>, <2 x i64>* %x0ptr
   %res = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x2.512(<2 x i64> %x0, <8 x i64> %x2, i8 %mask)
   ret <8 x i64> %res
 }
+
+declare <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float>, <16 x float>, i16)
+
+define <16 x float>@test_int_x86_avx512_mask_broadcastf32x2_512(<4 x float> %x0, <16 x float> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_broadcastf32x2_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    ## kill: %XMM0<def> %XMM0<kill> %YMM0<def>
+; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; CHECK-NEXT:    vinsertf64x4 $1, %ymm0, %zmm0, %zmm2
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vinsertf32x8 $1, %ymm0, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vinsertf32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z}
+; CHECK-NEXT:    vaddps %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    vaddps %zmm2, %zmm0, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float>  %x0, <16 x float> %x2, i16 %x3)
+  %res1 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> zeroinitializer, i16 %x3)
+  %res2 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> %x2, i16 -1)
+  %res3 = fadd <16 x float> %res, %res1
+  %res4 = fadd <16 x float> %res3, %res2
+  ret <16 x float> %res4
+}
+
+declare <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32>, <16 x i32>, i16)
+
+define <16 x i32>@test_int_x86_avx512_mask_broadcasti32x2_512(<4 x i32> %x0, <16 x i32> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_broadcasti32x2_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    ## kill: %XMM0<def> %XMM0<kill> %YMM0<def>
+; CHECK-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
+; CHECK-NEXT:    vinserti64x4 $1, %ymm0, %zmm0, %zmm2
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vinserti32x8 $1, %ymm0, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vinserti32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z}
+; CHECK-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32>  %x0, <16 x i32> %x2, i16 %x3)
+  %res1 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> %x2, i16 -1)
+  %res3 = add <16 x i32> %res, %res1
+  %res4 = add <16 x i32> %res3, %res2
+  ret <16 x i32> %res4
+}
+

test/CodeGen/X86/avx512dq-intrinsics.ll

 ; CHECK-NEXT:    ## kill: %AL<def> %AL<kill> %EAX<kill>
 ; CHECK-NEXT:    retq
   %res = call i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float> %x0, i32 4, i8 %x1)
   %res1 = call i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float> %x0, i32 4, i8 -1)
   %res2 = add i8 %res, %res1
   ret i8 %res2
 }
 
-declare <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float>, <16 x float>, i16)
-
-define <16 x float>@test_int_x86_avx512_mask_broadcastf32x2_512(<4 x float> %x0, <16 x float> %x2, i16 %x3) {
-; CHECK-LABEL: test_int_x86_avx512_mask_broadcastf32x2_512:
-; CHECK:       ## BB#0:
-; CHECK-NEXT:    kmovw %edi, %k1
-; CHECK-NEXT:    vbroadcastf32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vbroadcastf32x2 {{.*#+}} zmm2 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vaddps %zmm2, %zmm1, %zmm1
-; CHECK-NEXT:    vbroadcastf32x2 {{.*#+}} zmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vaddps %zmm0, %zmm1, %zmm0
-; CHECK-NEXT:    retq
-  %res = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float>  %x0, <16 x float> %x2, i16 %x3)
-  %res1 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> zeroinitializer, i16 %x3)
-  %res2 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> %x2, i16 -1)
-  %res3 = fadd <16 x float> %res, %res1
-  %res4 = fadd <16 x float> %res3, %res2
-  ret <16 x float> %res4
-}
-
-declare <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32>, <16 x i32>, i16)
-
-define <16 x i32>@test_int_x86_avx512_mask_broadcasti32x2_512(<4 x i32> %x0, <16 x i32> %x2, i16 %x3) {
-; CHECK-LABEL: test_int_x86_avx512_mask_broadcasti32x2_512:
-; CHECK:       ## BB#0:
-; CHECK-NEXT:    kmovw %edi, %k1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} zmm1 {%k1} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} zmm2 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vpaddd %zmm2, %zmm1, %zmm1
-; CHECK-NEXT:    vbroadcasti32x2 {{.*#+}} zmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
-; CHECK-NEXT:    retq
-  %res = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32>  %x0, <16 x i32> %x2, i16 %x3)
-  %res1 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> zeroinitializer, i16 %x3)
-  %res2 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> %x2, i16 -1)
-  %res3 = add <16 x i32> %res, %res1
-  %res4 = add <16 x i32> %res3, %res2
-  ret <16 x i32> %res4
-}
-
 declare i16 @llvm.x86.avx512.cvtd2mask.512(<16 x i32>)
 
 define i16@test_int_x86_avx512_cvtd2mask_512(<16 x i32> %x0) {
 ; CHECK-LABEL: test_int_x86_avx512_cvtd2mask_512:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    vpmovd2m %zmm0, %k0
 ; CHECK-NEXT:    kmovw %k0, %eax
 ; CHECK-NEXT:    ## kill: %AX<def> %AX<kill> %EAX<kill>

test/CodeGen/X86/avx512dqvl-intrinsics-upgrade.ll

 ; CHECK-NEXT:    vbroadcasti64x2 (%rdi), %ymm0 {%k1} ## encoding: [0x62,0xf2,0xfd,0x29,0x5a,0x07]
 ; CHECK-NEXT:    ## ymm0 {%k1} = mem[0,1,0,1]
 ; CHECK-NEXT:    retq ## encoding: [0xc3]
 
   %x0 = load <2 x i64>, <2 x i64>* %x0ptr
   %res = call <4 x i64> @llvm.x86.avx512.mask.broadcasti64x2.256(<2 x i64> %x0, <4 x i64> %x2, i8 %mask)
   ret <4 x i64> %res
 }
+
+declare <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float>, <8 x float>, i8)
+
+define <8 x float>@test_int_x86_avx512_mask_broadcastf32x2_256(<4 x float> %x0, <8 x float> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_broadcastf32x2_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    ## kill: %XMM0<def> %XMM0<kill> %YMM0<def>
+; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm2 ## EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd0,0x01]
+; CHECK-NEXT:    kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
+; CHECK-NEXT:    vinsertf32x4 $1, %xmm0, %ymm0, %ymm1 {%k1} ## encoding: [0x62,0xf3,0x7d,0x29,0x18,0xc8,0x01]
+; CHECK-NEXT:    vinsertf32x4 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf3,0x7d,0xa9,0x18,0xc0,0x01]
+; CHECK-NEXT:    vaddps %ymm0, %ymm1, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
+; CHECK-NEXT:    vaddps %ymm2, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfc,0x58,0xc2]
+; CHECK-NEXT:    retq ## encoding: [0xc3]
+  %res = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float>  %x0, <8 x float> %x2, i8 %x3)
+  %res1 = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float> %x0, <8 x float> zeroinitializer, i8 %x3)
+  %res2 = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float> %x0, <8 x float> %x2, i8 -1)
+  %res3 = fadd <8 x float> %res, %res1
+  %res4 = fadd <8 x float> %res3, %res2
+  ret <8 x float> %res4
+}
+
+declare <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32>, <8 x i32>, i8)
+
+define <8 x i32>@test_int_x86_avx512_mask_broadcasti32x2_256(<4 x i32> %x0, <8 x i32> %x2, i8 %x3, i64 * %y_ptr) {
+; CHECK-LABEL: test_int_x86_avx512_mask_broadcasti32x2_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    ## kill: %XMM0<def> %XMM0<kill> %YMM0<def>
+; CHECK-NEXT:    vmovq (%rsi), %xmm2 ## EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x16]
+; CHECK-NEXT:    ## xmm2 = mem[0],zero
+; CHECK-NEXT:    kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
+; CHECK-NEXT:    vinserti32x4 $1, %xmm2, %ymm2, %ymm1 {%k1} ## encoding: [0x62,0xf3,0x6d,0x29,0x38,0xca,0x01]
+; CHECK-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm2 ## EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd0,0x01]
+; CHECK-NEXT:    vinserti32x4 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf3,0x7d,0xa9,0x38,0xc0,0x01]
+; CHECK-NEXT:    vpaddd %ymm2, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfe,0xc2]
+; CHECK-NEXT:    vpaddd %ymm0, %ymm1, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf5,0xfe,0xc0]
+; CHECK-NEXT:    retq ## encoding: [0xc3]
+  %y_64  = load i64, i64 * %y_ptr
+  %y_v2i64 = insertelement <2 x i64> undef, i64 %y_64, i32 0
+  %y = bitcast <2 x i64> %y_v2i64 to <4 x i32>
+  %res = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32>  %y, <8 x i32> %x2, i8 %x3)
+  %res1 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32> %x0, <8 x i32> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32> %x0, <8 x i32> %x2, i8 -1)
+  %res3 = add <8 x i32> %res, %res1
+  %res4 = add <8 x i32> %res3, %res2
+  ret <8 x i32> %res4
+}
+
+declare <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32>, <4 x i32>, i8)
+
+define <4 x i32>@test_int_x86_avx512_mask_broadcasti32x2_128(<4 x i32> %x0, <4 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_broadcasti32x2_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
+; CHECK-NEXT:    vmovdqa32 %xmm0, %xmm1 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x6f,0xc8]
+; CHECK-NEXT:    vmovdqa32 %xmm0, %xmm2 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0x89,0x6f,0xd0]
+; CHECK-NEXT:    vpaddd %xmm2, %xmm1, %xmm1 ## EVEX TO VEX Compression encoding: [0xc5,0xf1,0xfe,0xca]
+; CHECK-NEXT:    vpaddd %xmm0, %xmm1, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf1,0xfe,0xc0]
+; CHECK-NEXT:    retq ## encoding: [0xc3]
+  %res = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32>  %x0, <4 x i32> %x2, i8 %x3)
+  %res1 = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32> %x0, <4 x i32> zeroinitializer, i8 %x3)
+  %res2 = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32> %x0, <4 x i32> %x2, i8 -1)
+  %res3 = add <4 x i32> %res, %res1
+  %res4 = add <4 x i32> %res3, %res2
+  ret <4 x i32> %res4
+}
+

test/CodeGen/X86/avx512dqvl-intrinsics.ll

 ; CHECK-NEXT:    ## kill: %AL<def> %AL<kill> %EAX<kill>
 ; CHECK-NEXT:    retq ## encoding: [0xc3]
   %res = call i8 @llvm.x86.avx512.mask.fpclass.pd.256(<4 x double> %x0, i32 2, i8 %x1)
   %res1 = call i8 @llvm.x86.avx512.mask.fpclass.pd.256(<4 x double> %x0, i32 4, i8 -1)
   %res2 = add i8 %res, %res1
   ret i8 %res2
 }
 
-declare <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float>, <8 x float>, i8)
-
-define <8 x float>@test_int_x86_avx512_mask_broadcastf32x2_256(<4 x float> %x0, <8 x float> %x2, i8 %x3) {
-; CHECK-LABEL: test_int_x86_avx512_mask_broadcastf32x2_256:
-; CHECK:       ## BB#0:
-; CHECK-NEXT:    kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
-; CHECK-NEXT:    vbroadcastf32x2 %xmm0, %ymm1 {%k1} ## encoding: [0x62,0xf2,0x7d,0x29,0x19,0xc8]
-; CHECK-NEXT:    ## ymm1 {%k1} = xmm0[0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vbroadcastf32x2 %xmm0, %ymm2 {%k1} {z} ## encoding: [0x62,0xf2,0x7d,0xa9,0x19,0xd0]
-; CHECK-NEXT:    ## ymm2 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vaddps %ymm2, %ymm1, %ymm1 ## EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xca]
-; CHECK-NEXT:    vbroadcastsd %xmm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0x19,0xc0]
-; CHECK-NEXT:    vaddps %ymm0, %ymm1, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
-; CHECK-NEXT:    retq ## encoding: [0xc3]
-  %res = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float>  %x0, <8 x float> %x2, i8 %x3)
-  %res1 = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float> %x0, <8 x float> zeroinitializer, i8 %x3)
-  %res2 = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float> %x0, <8 x float> %x2, i8 -1)
-  %res3 = fadd <8 x float> %res, %res1
-  %res4 = fadd <8 x float> %res3, %res2
-  ret <8 x float> %res4
-}
-
-declare <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32>, <8 x i32>, i8)
-
-define <8 x i32>@test_int_x86_avx512_mask_broadcasti32x2_256(<4 x i32> %x0, <8 x i32> %x2, i8 %x3, i64 * %y_ptr) {
-; CHECK-LABEL: test_int_x86_avx512_mask_broadcasti32x2_256:
-; CHECK:       ## BB#0:
-; CHECK-NEXT:    kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
-; CHECK-NEXT:    vbroadcasti32x2 (%rsi), %ymm1 {%k1} ## encoding: [0x62,0xf2,0x7d,0x29,0x59,0x0e]
-; CHECK-NEXT:    ## ymm1 {%k1} = mem[0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vbroadcasti32x2 %xmm0, %ymm2 {%k1} {z} ## encoding: [0x62,0xf2,0x7d,0xa9,0x59,0xd0]
-; CHECK-NEXT:    ## ymm2 {%k1} {z} = xmm0[0,1,0,1,0,1,0,1]
-; CHECK-NEXT:    vpbroadcastq %xmm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0x59,0xc0]
-; CHECK-NEXT:    vpaddd %ymm0, %ymm2, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xed,0xfe,0xc0]
-; CHECK-NEXT:    vpaddd %ymm0, %ymm1, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf5,0xfe,0xc0]
-; CHECK-NEXT:    retq ## encoding: [0xc3]
-  %y_64  = load i64, i64 * %y_ptr
-  %y_v2i64 = insertelement <2 x i64> undef, i64 %y_64, i32 0
-  %y = bitcast <2 x i64> %y_v2i64 to <4 x i32>
-  %res = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32>  %y, <8 x i32> %x2, i8 %x3)
-  %res1 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32> %x0, <8 x i32> zeroinitializer, i8 %x3)
-  %res2 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32> %x0, <8 x i32> %x2, i8 -1)
-  %res3 = add <8 x i32> %res, %res1
-  %res4 = add <8 x i32> %res3, %res2
-  ret <8 x i32> %res4
-}
-
-declare <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32>, <4 x i32>, i8)
-
-define <4 x i32>@test_int_x86_avx512_mask_broadcasti32x2_128(<4 x i32> %x0, <4 x i32> %x2, i8 %x3) {
-; CHECK-LABEL: test_int_x86_avx512_mask_broadcasti32x2_128:
-; CHECK:       ## BB#0:
-; CHECK-NEXT:    kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
-; CHECK-NEXT:    vbroadcasti32x2 %xmm0, %xmm1 {%k1} ## encoding: [0x62,0xf2,0x7d,0x09,0x59,0xc8]
-; CHECK-NEXT:    vbroadcasti32x2 %xmm0, %xmm2 {%k1} {z} ## encoding: [0x62,0xf2,0x7d,0x89,0x59,0xd0]
-; CHECK-NEXT:    vpaddd %xmm2, %xmm1, %xmm1 ## EVEX TO VEX Compression encoding: [0xc5,0xf1,0xfe,0xca]
-; CHECK-NEXT:    vpbroadcastq %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x59,0xc0]
-; CHECK-NEXT:    vpaddd %xmm0, %xmm1, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf1,0xfe,0xc0]
-; CHECK-NEXT:    retq ## encoding: [0xc3]
-  %res = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32>  %x0, <4 x i32> %x2, i8 %x3)
-  %res1 = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32> %x0, <4 x i32> zeroinitializer, i8 %x3)
-  %res2 = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32> %x0, <4 x i32> %x2, i8 -1)
-  %res3 = add <4 x i32> %res, %res1
-  %res4 = add <4 x i32> %res3, %res2
-  ret <4 x i32> %res4
-}
-
 declare i8 @llvm.x86.avx512.cvtd2mask.128(<4 x i32>)
 
 define i8@test_int_x86_avx512_cvtd2mask_128(<4 x i32> %x0) {
 ; CHECK-LABEL: test_int_x86_avx512_cvtd2mask_128:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    vpmovd2m %xmm0, %k0 ## encoding: [0x62,0xf2,0x7e,0x08,0x39,0xc0]
 ; CHECK-NEXT:    kmovw %k0, %eax ## encoding: [0xc5,0xf8,0x93,0xc0]
 ; CHECK-NEXT:    ## kill: %AL<def> %AL<kill> %EAX<kill>