[AVX512] add PSLLD and PSLLQ Intrinsic

[delena]

LGTM

include/llvm/IR/IntrinsicsX86.td

def int_x86_avx512_mask_psrlv_q : GCCBuiltin<"__builtin_ia32_psrlv8di_mask">,
                          llvm_v8i64_ty, llvm_v8i64_ty, llvm_i8_ty], 
                         [IntrNoMem]>;
   def int_x86_avx512_psll_dq_512 : GCCBuiltin<"__builtin_ia32_pslldq512">,
               Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty, llvm_i32_ty], 
                         [IntrNoMem]>;
   def int_x86_avx512_psrl_dq_512 : GCCBuiltin<"__builtin_ia32_psrldq512">,
               Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty, llvm_i32_ty], 
                         [IntrNoMem]>;                        
+
+  def int_x86_avx512_mask_psll_d_128 : GCCBuiltin<"__builtin_ia32_pslld128_mask">,
+              Intrinsic<[llvm_v4i32_ty], [llvm_v4i32_ty,
+                         llvm_v4i32_ty, llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_d_256 : GCCBuiltin<"__builtin_ia32_pslld256_mask">,
+              Intrinsic<[llvm_v8i32_ty], [llvm_v8i32_ty, 
+                         llvm_v4i32_ty, llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_di_128 : GCCBuiltin<"__builtin_ia32_pslldi128_mask">,
+              Intrinsic<[llvm_v4i32_ty], [llvm_v4i32_ty, 
+                         llvm_i8_ty, llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_di_256 : GCCBuiltin<"__builtin_ia32_pslldi256_mask">,
+              Intrinsic<[llvm_v8i32_ty], [llvm_v8i32_ty, 
+                         llvm_i8_ty, llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_di_512 : GCCBuiltin<"__builtin_ia32_pslldi512_mask">,
+              Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty, 
+                         llvm_i8_ty, llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_q_128 : GCCBuiltin<"__builtin_ia32_psllq128_mask">,
+              Intrinsic<[llvm_v2i64_ty], [llvm_v2i64_ty,
+                         llvm_v2i64_ty, llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_q_256 : GCCBuiltin<"__builtin_ia32_psllq256_mask">,
+              Intrinsic<[llvm_v4i64_ty], [llvm_v4i64_ty, 
+                         llvm_v2i64_ty, llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_qi_128 : GCCBuiltin<"__builtin_ia32_psllqi128_mask">,
+              Intrinsic<[llvm_v2i64_ty], [llvm_v2i64_ty, 
+                         llvm_i8_ty, llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_qi_256 : GCCBuiltin<"__builtin_ia32_psllqi256_mask">,
+              Intrinsic<[llvm_v4i64_ty], [llvm_v4i64_ty, 
+                         llvm_i8_ty, llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psll_qi_512 : GCCBuiltin<"__builtin_ia32_psllqi512_mask">,
+              Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty, 
+                         llvm_i8_ty, llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>;
 }
 
 // Gather ops
 let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_avx2_gather_d_pd : GCCBuiltin<"__builtin_ia32_gatherd_pd">,
       Intrinsic<[llvm_v2f64_ty],
         [llvm_v2f64_ty, llvm_ptr_ty, llvm_v4i32_ty, llvm_v2f64_ty, llvm_i8_ty],
         [IntrReadArgMem]>;

lib/Target/X86/X86IntrinsicsInfo.h

static const IntrinsicData  IntrinsicsWithoutChain[] = {
   X86_INTRINSIC_DATA(avx512_mask_por_q_512, INTR_TYPE_2OP_MASK, ISD::OR, 0),
   X86_INTRINSIC_DATA(avx512_mask_pshuf_b_128, INTR_TYPE_2OP_MASK,
                      X86ISD::PSHUFB, 0),
   X86_INTRINSIC_DATA(avx512_mask_pshuf_b_256, INTR_TYPE_2OP_MASK,
                     X86ISD::PSHUFB, 0),
   X86_INTRINSIC_DATA(avx512_mask_pshuf_b_512, INTR_TYPE_2OP_MASK,
                     X86ISD::PSHUFB, 0),
   X86_INTRINSIC_DATA(avx512_mask_psll_d,        INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_d_128,    INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_d_256,    INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_di_128,   INTR_TYPE_2OP_MASK, X86ISD::VSHLI, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_di_256,   INTR_TYPE_2OP_MASK, X86ISD::VSHLI, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_di_512,   INTR_TYPE_2OP_MASK, X86ISD::VSHLI, 0),
   X86_INTRINSIC_DATA(avx512_mask_psll_q,        INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_q_128,    INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_q_256,    INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_qi_128,   INTR_TYPE_2OP_MASK, X86ISD::VSHLI, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_qi_256,   INTR_TYPE_2OP_MASK, X86ISD::VSHLI, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psll_qi_512,   INTR_TYPE_2OP_MASK, X86ISD::VSHLI, 0),
   X86_INTRINSIC_DATA(avx512_mask_pslli_d,       VSHIFT_MASK, X86ISD::VSHLI, 0),
   X86_INTRINSIC_DATA(avx512_mask_pslli_q,       VSHIFT_MASK, X86ISD::VSHLI, 0),
   X86_INTRINSIC_DATA(avx512_mask_psllv_d,       INTR_TYPE_2OP_MASK, ISD::SHL, 0),
   X86_INTRINSIC_DATA(avx512_mask_psllv_q,       INTR_TYPE_2OP_MASK, ISD::SHL, 0),
   X86_INTRINSIC_DATA(avx512_mask_psra_d,        INTR_TYPE_2OP_MASK, X86ISD::VSRA, 0),
   X86_INTRINSIC_DATA(avx512_mask_psra_q,        INTR_TYPE_2OP_MASK, X86ISD::VSRA, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrai_d,       VSHIFT_MASK, X86ISD::VSRAI, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrai_q,       VSHIFT_MASK, X86ISD::VSRAI, 0),

test/CodeGen/X86/avx512-intrinsics.ll

; CHECK: vpaddq %zmm0, %zmm2, %zmm0
   %res1 = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x4.512(<4 x i64> %x0, <8 x i64> %x2, i8 -1)
   %res2 = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x4.512(<4 x i64> %x0, <8 x i64> %x2, i8 %mask)
   %res3 = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x4.512(<4 x i64> %x0, <8 x i64> zeroinitializer, i8 %mask)
   %res4 = add <8 x i64> %res1, %res2
   %res5 = add <8 x i64> %res3, %res4
   ret <8 x i64> %res5
 }
 
+declare <16 x i32> @llvm.x86.avx512.mask.psll.di.512(<16 x i32>, i8, <16 x i32>, i16)
+
+define <16 x i32>@test_int_x86_avx512_mask_psll_di_512(<16 x i32> %x0, i8 %x1, <16 x i32> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_di_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %esi, %k1
+; CHECK-NEXT:    vpslld $3, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vpslld $3, %zmm0, %zmm2 {%k1} {z}
+; CHECK-NEXT:    vpslld $3, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddd %zmm2, %zmm1, %zmm1
+; CHECK-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i32> @llvm.x86.avx512.mask.psll.di.512(<16 x i32> %x0, i8 3, <16 x i32> %x2, i16 %x3)
+  %res1 = call <16 x i32> @llvm.x86.avx512.mask.psll.di.512(<16 x i32> %x0, i8 3, <16 x i32> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i32> @llvm.x86.avx512.mask.psll.di.512(<16 x i32> %x0, i8 3, <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 <8 x i64> @llvm.x86.avx512.mask.psll.qi.512(<8 x i64>, i8, <8 x i64>, i8)
+
+define <8 x i64>@test_int_x86_avx512_mask_psll_qi_512(<8 x i64> %x0, i8 %x1, <8 x i64> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_qi_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsllq $3, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vpsllq $3, %zmm0, %zmm2 {%k1} {z}
+; CHECK-NEXT:    vpsllq $3, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddq %zmm2, %zmm1, %zmm1
+; CHECK-NEXT:    vpaddq %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i64> @llvm.x86.avx512.mask.psll.qi.512(<8 x i64> %x0, i8 3, <8 x i64> %x2, i8 %x3)
+  %res1 = call <8 x i64> @llvm.x86.avx512.mask.psll.qi.512(<8 x i64> %x0, i8 3, <8 x i64> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i64> @llvm.x86.avx512.mask.psll.qi.512(<8 x i64> %x0, i8 3, <8 x i64> %x2, i8 -1)
+  %res3 = add <8 x i64> %res, %res1
+  %res4 = add <8 x i64> %res3, %res2
+  ret <8 x i64> %res4
+}

test/CodeGen/X86/avx512vl-intrinsics.ll

; CHECK: vpaddd %ymm0, %ymm2, %ymm0
 
   %res1 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x4.256(<4 x i32> %x0, <8 x i32> %x2, i8 -1)
   %res2 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x4.256(<4 x i32> %x0, <8 x i32> %x2, i8 %mask)
   %res3 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x4.256(<4 x i32> %x0, <8 x i32> zeroinitializer, i8 %mask)
   %res4 = add <8 x i32> %res1, %res2
   %res5 = add <8 x i32> %res3, %res4
   ret <8 x i32> %res5
 }
+
+declare <4 x i32> @llvm.x86.avx512.mask.psll.d.128(<4 x i32>, <4 x i32>, <4 x i32>, i8)
+
+define <4 x i32>@test_int_x86_avx512_mask_psll_d_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_d_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpslld %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vpslld %xmm1, %xmm0, %xmm3 {%k1} {z}
+; CHECK-NEXT:    vpslld %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddd %xmm3, %xmm2, %xmm1
+; CHECK-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <4 x i32> @llvm.x86.avx512.mask.psll.d.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3)
+  %res1 = call <4 x i32> @llvm.x86.avx512.mask.psll.d.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> zeroinitializer, i8 %x3)
+  %res2 = call <4 x i32> @llvm.x86.avx512.mask.psll.d.128(<4 x i32> %x0, <4 x i32> %x1, <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 <8 x i32> @llvm.x86.avx512.mask.psll.d.256(<8 x i32>, <4 x i32>, <8 x i32>, i8)
+
+define <8 x i32>@test_int_x86_avx512_mask_psll_d_256(<8 x i32> %x0, <4 x i32> %x1, <8 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_d_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpslld %xmm1, %ymm0, %ymm2 {%k1}
+; CHECK-NEXT:    vpslld %xmm1, %ymm0, %ymm3 {%k1} {z}
+; CHECK-NEXT:    vpslld %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddd %ymm3, %ymm2, %ymm1
+; CHECK-NEXT:    vpaddd %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i32> @llvm.x86.avx512.mask.psll.d.256(<8 x i32> %x0, <4 x i32> %x1, <8 x i32> %x2, i8 %x3)
+  %res1 = call <8 x i32> @llvm.x86.avx512.mask.psll.d.256(<8 x i32> %x0, <4 x i32> %x1, <8 x i32> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i32> @llvm.x86.avx512.mask.psll.d.256(<8 x i32> %x0, <4 x i32> %x1, <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.psll.di.128(<4 x i32>, i8, <4 x i32>, i8)
+
+define <4 x i32>@test_int_x86_avx512_mask_psll_di_128(<4 x i32> %x0, i8 %x1, <4 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_di_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpslld $3, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vpslld $3, %xmm0, %xmm2 {%k1} {z}
+; CHECK-NEXT:    vpslld $3, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddd %xmm2, %xmm1, %xmm1
+; CHECK-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <4 x i32> @llvm.x86.avx512.mask.psll.di.128(<4 x i32> %x0, i8 3, <4 x i32> %x2, i8 %x3)
+  %res1 = call <4 x i32> @llvm.x86.avx512.mask.psll.di.128(<4 x i32> %x0, i8 3, <4 x i32> zeroinitializer, i8 %x3)
+  %res2 = call <4 x i32> @llvm.x86.avx512.mask.psll.di.128(<4 x i32> %x0, i8 3, <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 <8 x i32> @llvm.x86.avx512.mask.psll.di.256(<8 x i32>, i8, <8 x i32>, i8)
+
+define <8 x i32>@test_int_x86_avx512_mask_psll_di_256(<8 x i32> %x0, i8 %x1, <8 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_di_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpslld $3, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vpslld $3, %ymm0, %ymm2 {%k1} {z}
+; CHECK-NEXT:    vpslld $3, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddd %ymm2, %ymm1, %ymm1
+; CHECK-NEXT:    vpaddd %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i32> @llvm.x86.avx512.mask.psll.di.256(<8 x i32> %x0, i8 3, <8 x i32> %x2, i8 %x3)
+  %res1 = call <8 x i32> @llvm.x86.avx512.mask.psll.di.256(<8 x i32> %x0, i8 3, <8 x i32> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i32> @llvm.x86.avx512.mask.psll.di.256(<8 x i32> %x0, i8 3, <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 i64> @llvm.x86.avx512.mask.psll.q.256(<4 x i64>, <2 x i64>, <4 x i64>, i8)
+
+define <4 x i64>@test_int_x86_avx512_mask_psll_q_256(<4 x i64> %x0, <2 x i64> %x1, <4 x i64> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_q_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsllq %xmm1, %ymm0, %ymm2 {%k1}
+; CHECK-NEXT:    vpsllq %xmm1, %ymm0, %ymm3 {%k1} {z}
+; CHECK-NEXT:    vpsllq %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddq %ymm3, %ymm2, %ymm1
+; CHECK-NEXT:    vpaddq %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <4 x i64> @llvm.x86.avx512.mask.psll.q.256(<4 x i64> %x0, <2 x i64> %x1, <4 x i64> %x2, i8 %x3)
+  %res1 = call <4 x i64> @llvm.x86.avx512.mask.psll.q.256(<4 x i64> %x0, <2 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
+  %res2 = call <4 x i64> @llvm.x86.avx512.mask.psll.q.256(<4 x i64> %x0, <2 x i64> %x1, <4 x i64> %x2, i8 -1)
+  %res3 = add <4 x i64> %res, %res1
+  %res4 = add <4 x i64> %res3, %res2
+  ret <4 x i64> %res4
+}
+
+declare <2 x i64> @llvm.x86.avx512.mask.psll.qi.128(<2 x i64>, i8, <2 x i64>, i8)
+
+define <2 x i64>@test_int_x86_avx512_mask_psll_qi_128(<2 x i64> %x0, i8 %x1, <2 x i64> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_qi_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsllq $3, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vpsllq $3, %xmm0, %xmm2 {%k1} {z}
+; CHECK-NEXT:    vpsllq $3, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddq %xmm2, %xmm1, %xmm1
+; CHECK-NEXT:    vpaddq %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <2 x i64> @llvm.x86.avx512.mask.psll.qi.128(<2 x i64> %x0, i8 3, <2 x i64> %x2, i8 %x3)
+  %res1 = call <2 x i64> @llvm.x86.avx512.mask.psll.qi.128(<2 x i64> %x0, i8 3, <2 x i64> zeroinitializer, i8 %x3)
+  %res2 = call <2 x i64> @llvm.x86.avx512.mask.psll.qi.128(<2 x i64> %x0, i8 3, <2 x i64> %x2, i8 -1)
+  %res3 = add <2 x i64> %res, %res1
+  %res4 = add <2 x i64> %res3, %res2
+  ret <2 x i64> %res4
+}
+
+declare <4 x i64> @llvm.x86.avx512.mask.psll.qi.256(<4 x i64>, i8, <4 x i64>, i8)
+
+define <4 x i64>@test_int_x86_avx512_mask_psll_qi_256(<4 x i64> %x0, i8 %x1, <4 x i64> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psll_qi_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsllq $3, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vpsllq $3, %ymm0, %ymm2 {%k1} {z}
+; CHECK-NEXT:    vpsllq $3, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddq %ymm2, %ymm1, %ymm1
+; CHECK-NEXT:    vpaddq %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <4 x i64> @llvm.x86.avx512.mask.psll.qi.256(<4 x i64> %x0, i8 3, <4 x i64> %x2, i8 %x3)
+  %res1 = call <4 x i64> @llvm.x86.avx512.mask.psll.qi.256(<4 x i64> %x0, i8 3, <4 x i64> zeroinitializer, i8 %x3)
+  %res2 = call <4 x i64> @llvm.x86.avx512.mask.psll.qi.256(<4 x i64> %x0, i8 3, <4 x i64> %x2, i8 -1)
+  %res3 = add <4 x i64> %res, %res1
+  %res4 = add <4 x i64> %res3, %res2
+  ret <4 x i64> %res4
+}
+