[AVX512] Adding PMOVSXBD/W/Q , PMOVZSDQ and PMOVZSWD/Q Intrinsics .

[m_zuckerman]

ping

[igorb]

LGTM

llvm/trunk/include/llvm/IR/IntrinsicsX86.td

def int_x86_avx512_kunpck_dq : GCCBuiltin<"__builtin_ia32_kunpckdi">,
               Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty],
                          [IntrNoMem]>;
   def int_x86_avx512_kortestz_w : GCCBuiltin<"__builtin_ia32_kortestzhi">,
               Intrinsic<[llvm_i32_ty], [llvm_i16_ty, llvm_i16_ty],
                         [IntrNoMem]>;
   def int_x86_avx512_kortestc_w : GCCBuiltin<"__builtin_ia32_kortestchi">,
               Intrinsic<[llvm_i32_ty], [llvm_i16_ty, llvm_i16_ty],
                         [IntrNoMem]>;
+                        
+  def int_x86_avx512_mask_pmovsxb_d_128 : GCCBuiltin<"__builtin_ia32_pmovsxbd128_mask">,
+              Intrinsic<[llvm_v4i32_ty], [llvm_v16i8_ty,
+                        llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_d_256 : GCCBuiltin<"__builtin_ia32_pmovsxbd256_mask">,
+              Intrinsic<[llvm_v8i32_ty], [llvm_v16i8_ty,
+                        llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_d_512 : GCCBuiltin<"__builtin_ia32_pmovsxbd512_mask">,
+              Intrinsic<[llvm_v16i32_ty], [llvm_v16i8_ty,
+                        llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_q_128 : GCCBuiltin<"__builtin_ia32_pmovsxbq128_mask">,
+              Intrinsic<[llvm_v2i64_ty], [llvm_v16i8_ty,
+                        llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_q_256 : GCCBuiltin<"__builtin_ia32_pmovsxbq256_mask">,
+              Intrinsic<[llvm_v4i64_ty], [llvm_v16i8_ty,
+                        llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_q_512 : GCCBuiltin<"__builtin_ia32_pmovsxbq512_mask">,
+              Intrinsic<[llvm_v8i64_ty], [llvm_v16i8_ty,
+                        llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_w_128 : GCCBuiltin<"__builtin_ia32_pmovsxbw128_mask">,
+              Intrinsic<[llvm_v8i16_ty], [llvm_v16i8_ty,
+                        llvm_v8i16_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_w_256 : GCCBuiltin<"__builtin_ia32_pmovsxbw256_mask">,
+              Intrinsic<[llvm_v16i16_ty], [llvm_v16i8_ty,
+                        llvm_v16i16_ty, llvm_i16_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxb_w_512 : GCCBuiltin<"__builtin_ia32_pmovsxbw512_mask">,
+              Intrinsic<[llvm_v32i16_ty], [llvm_v32i8_ty,
+                        llvm_v32i16_ty, llvm_i32_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxd_q_128 : GCCBuiltin<"__builtin_ia32_pmovsxdq128_mask">,
+              Intrinsic<[llvm_v2i64_ty], [llvm_v4i32_ty,
+                        llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxd_q_256 : GCCBuiltin<"__builtin_ia32_pmovsxdq256_mask">,
+              Intrinsic<[llvm_v4i64_ty], [llvm_v4i32_ty,
+                        llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxd_q_512 : GCCBuiltin<"__builtin_ia32_pmovsxdq512_mask">,
+              Intrinsic<[llvm_v8i64_ty], [llvm_v8i32_ty,
+                        llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxw_d_128 : GCCBuiltin<"__builtin_ia32_pmovsxwd128_mask">,
+              Intrinsic<[llvm_v4i32_ty], [llvm_v8i16_ty,
+                        llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxw_d_256 : GCCBuiltin<"__builtin_ia32_pmovsxwd256_mask">,
+              Intrinsic<[llvm_v8i32_ty], [llvm_v8i16_ty,
+                        llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxw_d_512 : GCCBuiltin<"__builtin_ia32_pmovsxwd512_mask">,
+              Intrinsic<[llvm_v16i32_ty], [llvm_v16i16_ty,
+                        llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxw_q_128 : GCCBuiltin<"__builtin_ia32_pmovsxwq128_mask">,
+              Intrinsic<[llvm_v2i64_ty], [llvm_v8i16_ty,
+                        llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxw_q_256 : GCCBuiltin<"__builtin_ia32_pmovsxwq256_mask">,
+              Intrinsic<[llvm_v4i64_ty], [llvm_v8i16_ty,
+                        llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_pmovsxw_q_512 : GCCBuiltin<"__builtin_ia32_pmovsxwq512_mask">,
+              Intrinsic<[llvm_v8i64_ty], [llvm_v8i16_ty,
+                        llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>;
 }
 
 // Conversion ops
 let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_avx512_cvtss2usi : GCCBuiltin<"__builtin_ia32_cvtss2usi">,
               Intrinsic<[llvm_i32_ty], [llvm_v4f32_ty], [IntrNoMem]>;
   def int_x86_avx512_cvtss2usi64 : GCCBuiltin<"__builtin_ia32_cvtss2usi64">,
               Intrinsic<[llvm_i64_ty], [llvm_v4f32_ty], [IntrNoMem]>;

llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h

static const IntrinsicData  IntrinsicsWithoutChain[] = {
   X86_INTRINSIC_DATA(avx512_mask_pmovs_qw_512, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovs_wb_128, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovs_wb_256, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovs_wb_512, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCS, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_d_128,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_d_256,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_d_512,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_q_128,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_q_256,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_q_512,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_w_128,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_w_256,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxb_w_512,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxd_q_128,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxd_q_256,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxd_q_512,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxw_d_128,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxw_d_256,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxw_d_512,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxw_q_128,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxw_q_256,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
+  X86_INTRINSIC_DATA(avx512_mask_pmovsxw_q_512,  INTR_TYPE_1OP_MASK,
+                     X86ISD::VSEXT, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovus_db_128, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCUS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovus_db_256, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCUS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovus_db_512, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCUS, 0),
   X86_INTRINSIC_DATA(avx512_mask_pmovus_dw_128, INTR_TYPE_1OP_MASK,
                      X86ISD::VTRUNCUS, 0),

llvm/trunk/test/CodeGen/X86/avx512-intrinsics.ll

 ; CHECK-NEXT:    retq 
   %res = call <8 x i64> @llvm.x86.avx512.mask.pmovzxw.q.512(<8 x i16> %x0, <8 x i64> %x1, i8 %x2)
   %res1 = call <8 x i64> @llvm.x86.avx512.mask.pmovzxw.q.512(<8 x i16> %x0, <8 x i64> zeroinitializer, i8 %x2)
   %res2 = call <8 x i64> @llvm.x86.avx512.mask.pmovzxw.q.512(<8 x i16> %x0, <8 x i64> %x1, i8 -1)
   %res3 = add <8 x i64> %res, %res1
   %res4 = add <8 x i64> %res3, %res2
   ret <8 x i64> %res4
 }
+
+declare <16 x i32> @llvm.x86.avx512.mask.pmovsxb.d.512(<16 x i8>, <16 x i32>, i16)
+
+define <16 x i32>@test_int_x86_avx512_mask_pmovsxb_d_512(<16 x i8> %x0, <16 x i32> %x1, i16 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_d_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %zmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %zmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %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.pmovsxb.d.512(<16 x i8> %x0, <16 x i32> %x1, i16 %x2)
+  %res1 = call <16 x i32> @llvm.x86.avx512.mask.pmovsxb.d.512(<16 x i8> %x0, <16 x i32> zeroinitializer, i16 %x2)
+  %res2 = call <16 x i32> @llvm.x86.avx512.mask.pmovsxb.d.512(<16 x i8> %x0, <16 x i32> %x1, 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.pmovsxb.q.512(<16 x i8>, <8 x i64>, i8)
+
+define <8 x i64>@test_int_x86_avx512_mask_pmovsxb_q_512(<16 x i8> %x0, <8 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_q_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %zmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %zmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %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.pmovsxb.q.512(<16 x i8> %x0, <8 x i64> %x1, i8 %x2)
+  %res1 = call <8 x i64> @llvm.x86.avx512.mask.pmovsxb.q.512(<16 x i8> %x0, <8 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <8 x i64> @llvm.x86.avx512.mask.pmovsxb.q.512(<16 x i8> %x0, <8 x i64> %x1, i8 -1)
+  %res3 = add <8 x i64> %res, %res1
+  %res4 = add <8 x i64> %res3, %res2
+  ret <8 x i64> %res4
+}
+
+declare <8 x i64> @llvm.x86.avx512.mask.pmovsxd.q.512(<8 x i32>, <8 x i64>, i8)
+
+define <8 x i64>@test_int_x86_avx512_mask_pmovsxd_q_512(<8 x i32> %x0, <8 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxd_q_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxdq %ymm0, %zmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxdq %ymm0, %zmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxdq %ymm0, %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.pmovsxd.q.512(<8 x i32> %x0, <8 x i64> %x1, i8 %x2)
+  %res1 = call <8 x i64> @llvm.x86.avx512.mask.pmovsxd.q.512(<8 x i32> %x0, <8 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <8 x i64> @llvm.x86.avx512.mask.pmovsxd.q.512(<8 x i32> %x0, <8 x i64> %x1, i8 -1)
+  %res3 = add <8 x i64> %res, %res1
+  %res4 = add <8 x i64> %res3, %res2
+  ret <8 x i64> %res4
+}
+
+
+declare <16 x i32> @llvm.x86.avx512.mask.pmovsxw.d.512(<16 x i16>, <16 x i32>, i16)
+
+define <16 x i32>@test_int_x86_avx512_mask_pmovsxw_d_512(<16 x i16> %x0, <16 x i32> %x1, i16 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxw_d_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1 
+; CHECK-NEXT:    vpmovsxwd %ymm0, %zmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxwd %ymm0, %zmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxwd %ymm0, %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.pmovsxw.d.512(<16 x i16> %x0, <16 x i32> %x1, i16 %x2)
+  %res1 = call <16 x i32> @llvm.x86.avx512.mask.pmovsxw.d.512(<16 x i16> %x0, <16 x i32> zeroinitializer, i16 %x2)
+  %res2 = call <16 x i32> @llvm.x86.avx512.mask.pmovsxw.d.512(<16 x i16> %x0, <16 x i32> %x1, 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.pmovsxw.q.512(<8 x i16>, <8 x i64>, i8)
+
+define <8 x i64>@test_int_x86_avx512_mask_pmovsxw_q_512(<8 x i16> %x0, <8 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxw_q_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %zmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %zmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %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.pmovsxw.q.512(<8 x i16> %x0, <8 x i64> %x1, i8 %x2)
+  %res1 = call <8 x i64> @llvm.x86.avx512.mask.pmovsxw.q.512(<8 x i16> %x0, <8 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <8 x i64> @llvm.x86.avx512.mask.pmovsxw.q.512(<8 x i16> %x0, <8 x i64> %x1, i8 -1)
+  %res3 = add <8 x i64> %res, %res1
+  %res4 = add <8 x i64> %res3, %res2
+  ret <8 x i64> %res4
+}
+
+

llvm/trunk/test/CodeGen/X86/avx512bw-intrinsics.ll

; AVX512BW-NEXT:    retq
   %res = call <32 x i16> @llvm.x86.avx512.mask.pmovzxb.w.512(<32 x i8> %x0, <32 x i16> %x1, i32 %x2)
   %res1 = call <32 x i16> @llvm.x86.avx512.mask.pmovzxb.w.512(<32 x i8> %x0, <32 x i16> zeroinitializer, i32 %x2)
   %res2 = call <32 x i16> @llvm.x86.avx512.mask.pmovzxb.w.512(<32 x i8> %x0, <32 x i16> %x1, i32 -1)
   %res3 = add <32 x i16> %res, %res1
   %res4 = add <32 x i16> %res3, %res2
   ret <32 x i16> %res4
 }
 
+
+declare <32 x i16> @llvm.x86.avx512.mask.pmovsxb.w.512(<32 x i8>, <32 x i16>, i32)
+
+define <32 x i16>@test_int_x86_avx512_mask_pmovsxb_w_512(<32 x i8> %x0, <32 x i16> %x1, i32 %x2) {
+; AVX512BW-LABEL: test_int_x86_avx512_mask_pmovsxb_w_512:
+; AVX512BW:       ## BB#0:
+; AVX512BW-NEXT:    kmovd %edi, %k1 
+; AVX512BW-NEXT:    vpmovsxbw %ymm0, %zmm1 {%k1} 
+; AVX512BW-NEXT:    vpmovsxbw %ymm0, %zmm2 {%k1} {z} 
+; AVX512BW-NEXT:    vpmovsxbw %ymm0, %zmm0 
+; AVX512BW-NEXT:    vpaddw %zmm2, %zmm1, %zmm1 
+; AVX512BW-NEXT:    vpaddw %zmm0, %zmm1, %zmm0 
+; AVX512BW-NEXT:    retq 
+  %res = call <32 x i16> @llvm.x86.avx512.mask.pmovsxb.w.512(<32 x i8> %x0, <32 x i16> %x1, i32 %x2)
+  %res1 = call <32 x i16> @llvm.x86.avx512.mask.pmovsxb.w.512(<32 x i8> %x0, <32 x i16> zeroinitializer, i32 %x2)
+  %res2 = call <32 x i16> @llvm.x86.avx512.mask.pmovsxb.w.512(<32 x i8> %x0, <32 x i16> %x1, i32 -1)
+  %res3 = add <32 x i16> %res, %res1
+  %res4 = add <32 x i16> %res3, %res2
+  ret <32 x i16> %res4
+}
 No newline at end of file

llvm/trunk/test/CodeGen/X86/avx512bwvl-intrinsics.ll

; CHECK-NEXT:    retq
   %res = call <16 x i16> @llvm.x86.avx512.mask.pmovzxb.w.256(<16 x i8> %x0, <16 x i16> %x1, i16 %x2)
   %res1 = call <16 x i16> @llvm.x86.avx512.mask.pmovzxb.w.256(<16 x i8> %x0, <16 x i16> zeroinitializer, i16 %x2)
   %res2 = call <16 x i16> @llvm.x86.avx512.mask.pmovzxb.w.256(<16 x i8> %x0, <16 x i16> %x1, i16 -1)
   %res3 = add <16 x i16> %res, %res1
   %res4 = add <16 x i16> %res3, %res2
   ret <16 x i16> %res4
 }
 
+
+declare <8 x i16> @llvm.x86.avx512.mask.pmovsxb.w.128(<16 x i8>, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_pmovsxb_w_128(<16 x i8> %x0, <8 x i16> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_w_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxbw %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbw %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbw %xmm0, %xmm0 
+; CHECK-NEXT:    vpaddw %xmm2, %xmm1, %xmm1 
+; CHECK-NEXT:    vpaddw %xmm0, %xmm1, %xmm0 
+; CHECK-NEXT:    retq 
+  %res = call <8 x i16> @llvm.x86.avx512.mask.pmovsxb.w.128(<16 x i8> %x0, <8 x i16> %x1, i8 %x2)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.pmovsxb.w.128(<16 x i8> %x0, <8 x i16> zeroinitializer, i8 %x2)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.pmovsxb.w.128(<16 x i8> %x0, <8 x i16> %x1, i8 -1)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res3, %res2
+  ret <8 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.pmovsxb.w.256(<16 x i8>, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_pmovsxb_w_256(<16 x i8> %x0, <16 x i16> %x1, i16 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_w_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1 
+; CHECK-NEXT:    vpmovsxbw %xmm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbw %xmm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbw %xmm0, %ymm0 
+; CHECK-NEXT:    vpaddw %ymm2, %ymm1, %ymm1 
+; CHECK-NEXT:    vpaddw %ymm0, %ymm1, %ymm0 
+; CHECK-NEXT:    retq 
+  %res = call <16 x i16> @llvm.x86.avx512.mask.pmovsxb.w.256(<16 x i8> %x0, <16 x i16> %x1, i16 %x2)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.pmovsxb.w.256(<16 x i8> %x0, <16 x i16> zeroinitializer, i16 %x2)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.pmovsxb.w.256(<16 x i8> %x0, <16 x i16> %x1, i16 -1)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <2 x i64> @llvm.x86.avx512.mask.pmovsxd.q.128(<4 x i32>, <2 x i64>, i8)
+
+define <2 x i64>@test_int_x86_avx512_mask_pmovsxd_q_128(<4 x i32> %x0, <2 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxd_q_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxdq %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxdq %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxdq %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.pmovsxd.q.128(<4 x i32> %x0, <2 x i64> %x1, i8 %x2)
+  %res1 = call <2 x i64> @llvm.x86.avx512.mask.pmovsxd.q.128(<4 x i32> %x0, <2 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <2 x i64> @llvm.x86.avx512.mask.pmovsxd.q.128(<4 x i32> %x0, <2 x i64> %x1, 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.pmovsxd.q.256(<4 x i32>, <4 x i64>, i8)
+
+define <4 x i64>@test_int_x86_avx512_mask_pmovsxd_q_256(<4 x i32> %x0, <4 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxd_q_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxdq %xmm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpmovsxdq %xmm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxdq %xmm0, %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.pmovsxd.q.256(<4 x i32> %x0, <4 x i64> %x1, i8 %x2)
+  %res1 = call <4 x i64> @llvm.x86.avx512.mask.pmovsxd.q.256(<4 x i32> %x0, <4 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <4 x i64> @llvm.x86.avx512.mask.pmovsxd.q.256(<4 x i32> %x0, <4 x i64> %x1, i8 -1)
+  %res3 = add <4 x i64> %res, %res1
+  %res4 = add <4 x i64> %res3, %res2
+  ret <4 x i64> %res4
+}
+

llvm/trunk/test/CodeGen/X86/avx512vl-intrinsics.ll

; CHECK-NEXT:    retq
   %res = call <4 x i64> @llvm.x86.avx512.mask.pmovzxw.q.256(<8 x i16> %x0, <4 x i64> %x1, i8 %x2)
   %res1 = call <4 x i64> @llvm.x86.avx512.mask.pmovzxw.q.256(<8 x i16> %x0, <4 x i64> zeroinitializer, i8 %x2)
   %res2 = call <4 x i64> @llvm.x86.avx512.mask.pmovzxw.q.256(<8 x i16> %x0, <4 x i64> %x1, i8 -1)
   %res3 = add <4 x i64> %res, %res1
   %res4 = add <4 x i64> %res3, %res2
   ret <4 x i64> %res4
 }
 
+declare <4 x i32> @llvm.x86.avx512.mask.pmovsxb.d.128(<16 x i8>, <4 x i32>, i8)
+
+define <4 x i32>@test_int_x86_avx512_mask_pmovsxb_d_128(<16 x i8> %x0, <4 x i32> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_d_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbd %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.pmovsxb.d.128(<16 x i8> %x0, <4 x i32> %x1, i8 %x2)
+  %res1 = call <4 x i32> @llvm.x86.avx512.mask.pmovsxb.d.128(<16 x i8> %x0, <4 x i32> zeroinitializer, i8 %x2)
+  %res2 = call <4 x i32> @llvm.x86.avx512.mask.pmovsxb.d.128(<16 x i8> %x0, <4 x i32> %x1, 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.pmovsxb.d.256(<16 x i8>, <8 x i32>, i8)
+
+define <8 x i32>@test_int_x86_avx512_mask_pmovsxb_d_256(<16 x i8> %x0, <8 x i32> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_d_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbd %xmm0, %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.pmovsxb.d.256(<16 x i8> %x0, <8 x i32> %x1, i8 %x2)
+  %res1 = call <8 x i32> @llvm.x86.avx512.mask.pmovsxb.d.256(<16 x i8> %x0, <8 x i32> zeroinitializer, i8 %x2)
+  %res2 = call <8 x i32> @llvm.x86.avx512.mask.pmovsxb.d.256(<16 x i8> %x0, <8 x i32> %x1, i8 -1)
+  %res3 = add <8 x i32> %res, %res1
+  %res4 = add <8 x i32> %res3, %res2
+  ret <8 x i32> %res4
+}
+
+declare <2 x i64> @llvm.x86.avx512.mask.pmovsxb.q.128(<16 x i8>, <2 x i64>, i8)
+
+define <2 x i64>@test_int_x86_avx512_mask_pmovsxb_q_128(<16 x i8> %x0, <2 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_q_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbq %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.pmovsxb.q.128(<16 x i8> %x0, <2 x i64> %x1, i8 %x2)
+  %res1 = call <2 x i64> @llvm.x86.avx512.mask.pmovsxb.q.128(<16 x i8> %x0, <2 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <2 x i64> @llvm.x86.avx512.mask.pmovsxb.q.128(<16 x i8> %x0, <2 x i64> %x1, 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.pmovsxb.q.256(<16 x i8>, <4 x i64>, i8)
+
+define <4 x i64>@test_int_x86_avx512_mask_pmovsxb_q_256(<16 x i8> %x0, <4 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxb_q_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxbq %xmm0, %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.pmovsxb.q.256(<16 x i8> %x0, <4 x i64> %x1, i8 %x2)
+  %res1 = call <4 x i64> @llvm.x86.avx512.mask.pmovsxb.q.256(<16 x i8> %x0, <4 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <4 x i64> @llvm.x86.avx512.mask.pmovsxb.q.256(<16 x i8> %x0, <4 x i64> %x1, i8 -1)
+  %res3 = add <4 x i64> %res, %res1
+  %res4 = add <4 x i64> %res3, %res2
+  ret <4 x i64> %res4
+}
+
+declare <4 x i32> @llvm.x86.avx512.mask.pmovsxw.d.128(<8 x i16>, <4 x i32>, i8)
+
+define <4 x i32>@test_int_x86_avx512_mask_pmovsxw_d_128(<8 x i16> %x0, <4 x i32> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxw_d_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxwd %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxwd %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxwd %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.pmovsxw.d.128(<8 x i16> %x0, <4 x i32> %x1, i8 %x2)
+  %res1 = call <4 x i32> @llvm.x86.avx512.mask.pmovsxw.d.128(<8 x i16> %x0, <4 x i32> zeroinitializer, i8 %x2)
+  %res2 = call <4 x i32> @llvm.x86.avx512.mask.pmovsxw.d.128(<8 x i16> %x0, <4 x i32> %x1, 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.pmovsxw.d.256(<8 x i16>, <8 x i32>, i8)
+
+define <8 x i32>@test_int_x86_avx512_mask_pmovsxw_d_256(<8 x i16> %x0, <8 x i32> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxw_d_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxwd %xmm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpmovsxwd %xmm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxwd %xmm0, %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.pmovsxw.d.256(<8 x i16> %x0, <8 x i32> %x1, i8 %x2)
+  %res1 = call <8 x i32> @llvm.x86.avx512.mask.pmovsxw.d.256(<8 x i16> %x0, <8 x i32> zeroinitializer, i8 %x2)
+  %res2 = call <8 x i32> @llvm.x86.avx512.mask.pmovsxw.d.256(<8 x i16> %x0, <8 x i32> %x1, i8 -1)
+  %res3 = add <8 x i32> %res, %res1
+  %res4 = add <8 x i32> %res3, %res2
+  ret <8 x i32> %res4
+}
+
+declare <2 x i64> @llvm.x86.avx512.mask.pmovsxw.q.128(<8 x i16>, <2 x i64>, i8)
+
+define <2 x i64>@test_int_x86_avx512_mask_pmovsxw_q_128(<8 x i16> %x0, <2 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxw_q_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxwq %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.pmovsxw.q.128(<8 x i16> %x0, <2 x i64> %x1, i8 %x2)
+  %res1 = call <2 x i64> @llvm.x86.avx512.mask.pmovsxw.q.128(<8 x i16> %x0, <2 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <2 x i64> @llvm.x86.avx512.mask.pmovsxw.q.128(<8 x i16> %x0, <2 x i64> %x1, 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.pmovsxw.q.256(<8 x i16>, <4 x i64>, i8)
+
+define <4 x i64>@test_int_x86_avx512_mask_pmovsxw_q_256(<8 x i16> %x0, <4 x i64> %x1, i8 %x2) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pmovsxw_q_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpmovsxwq %xmm0, %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.pmovsxw.q.256(<8 x i16> %x0, <4 x i64> %x1, i8 %x2)
+  %res1 = call <4 x i64> @llvm.x86.avx512.mask.pmovsxw.q.256(<8 x i16> %x0, <4 x i64> zeroinitializer, i8 %x2)
+  %res2 = call <4 x i64> @llvm.x86.avx512.mask.pmovsxw.q.256(<8 x i16> %x0, <4 x i64> %x1, i8 -1)
+  %res3 = add <4 x i64> %res, %res1
+  %res4 = add <4 x i64> %res3, %res2
+  ret <4 x i64> %res4
+}