[AVX512] add PSRAW Intrinsic

[delena]

LGTM

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

def int_x86_avx512_mask_psrl_wi_128 : GCCBuiltin<"__builtin_ia32_psrlwi128_mask">,
                          llvm_i8_ty, llvm_v8i16_ty,  llvm_i8_ty], [IntrNoMem]>;
   def int_x86_avx512_mask_psrl_wi_256 : GCCBuiltin<"__builtin_ia32_psrlwi256_mask">,
               Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty,
                          llvm_i8_ty, llvm_v16i16_ty,  llvm_i16_ty], [IntrNoMem]>;
   def int_x86_avx512_mask_psrl_wi_512 : GCCBuiltin<"__builtin_ia32_psrlwi512_mask">,
               Intrinsic<[llvm_v32i16_ty], [llvm_v32i16_ty,
                          llvm_i8_ty, llvm_v32i16_ty,  llvm_i32_ty], [IntrNoMem]>;
 
+  def int_x86_avx512_mask_psra_w_128 : GCCBuiltin<"__builtin_ia32_psraw128_mask">,
+              Intrinsic<[llvm_v8i16_ty], [llvm_v8i16_ty,
+                         llvm_v8i16_ty, llvm_v8i16_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psra_w_256 : GCCBuiltin<"__builtin_ia32_psraw256_mask">,
+              Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty,
+                         llvm_v8i16_ty, llvm_v16i16_ty, llvm_i16_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psra_w_512 : GCCBuiltin<"__builtin_ia32_psraw512_mask">,
+              Intrinsic<[llvm_v32i16_ty], [llvm_v32i16_ty,
+                         llvm_v8i16_ty, llvm_v32i16_ty, llvm_i32_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psra_wi_128 : GCCBuiltin<"__builtin_ia32_psrawi128_mask">,
+              Intrinsic<[llvm_v8i16_ty], [llvm_v8i16_ty,
+                         llvm_i8_ty, llvm_v8i16_ty, llvm_i8_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psra_wi_256 : GCCBuiltin<"__builtin_ia32_psrawi256_mask">,
+              Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty,
+                         llvm_i8_ty, llvm_v16i16_ty, llvm_i16_ty], [IntrNoMem]>;
+  def int_x86_avx512_mask_psra_wi_512 : GCCBuiltin<"__builtin_ia32_psrawi512_mask">,
+              Intrinsic<[llvm_v32i16_ty], [llvm_v32i16_ty,
+                         llvm_i8_ty, llvm_v32i16_ty, llvm_i32_ty], [IntrNoMem]>;
+
   def int_x86_avx512_mask_psll_d : GCCBuiltin<"__builtin_ia32_pslld512_mask">,
               Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty,
                          llvm_v4i32_ty, llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
   def int_x86_avx512_mask_psll_q : GCCBuiltin<"__builtin_ia32_psllq512_mask">,
               Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty,
                          llvm_v2i64_ty, llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>;
   def int_x86_avx512_mask_psrl_d : GCCBuiltin<"__builtin_ia32_psrld512_mask">,
               Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty,

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

static const IntrinsicData  IntrinsicsWithoutChain[] = {
   X86_INTRINSIC_DATA(avx512_mask_psll_d,        INTR_TYPE_2OP_MASK, X86ISD::VSHL, 0),
   X86_INTRINSIC_DATA(avx512_mask_psll_q,        INTR_TYPE_2OP_MASK, X86ISD::VSHL, 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_psra_w_128,    INTR_TYPE_2OP_MASK, X86ISD::VSRA, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psra_w_256,    INTR_TYPE_2OP_MASK, X86ISD::VSRA, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psra_w_512,    INTR_TYPE_2OP_MASK, X86ISD::VSRA, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psra_wi_128,   INTR_TYPE_2OP_MASK, X86ISD::VSRAI, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psra_wi_256,   INTR_TYPE_2OP_MASK, X86ISD::VSRAI, 0),
+  X86_INTRINSIC_DATA(avx512_mask_psra_wi_512,   INTR_TYPE_2OP_MASK, X86ISD::VSRAI, 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),
   X86_INTRINSIC_DATA(avx512_mask_psrav_d,       INTR_TYPE_2OP_MASK, ISD::SRA, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrav_q,       INTR_TYPE_2OP_MASK, ISD::SRA, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrl_d,        INTR_TYPE_2OP_MASK, X86ISD::VSRL, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrl_d_128,    INTR_TYPE_2OP_MASK, X86ISD::VSRL, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrl_d_256,    INTR_TYPE_2OP_MASK, X86ISD::VSRL, 0),
   X86_INTRINSIC_DATA(avx512_mask_psrl_di_128,   INTR_TYPE_2OP_MASK, X86ISD::VSRLI, 0),

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

 ; CHECK-NEXT:    retq
   %res = call <32 x i16> @llvm.x86.avx512.mask.psrlv32hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3)
   %res1 = call <32 x i16> @llvm.x86.avx512.mask.psrlv32hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> zeroinitializer, i32 %x3)
   %res2 = call <32 x i16> @llvm.x86.avx512.mask.psrlv32hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, 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.psra.w.512(<32 x i16>, <8 x i16>, <32 x i16>, i32)
+
+define <32 x i16>@test_int_x86_avx512_mask_psra_w_512(<32 x i16> %x0, <8 x i16> %x1, <32 x i16> %x2, i32 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_w_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovd %edi, %k1
+; CHECK-NEXT:    vpsraw %xmm1, %zmm0, %zmm2 {%k1}
+; CHECK-NEXT:    vpsraw %xmm1, %zmm0, %zmm3 {%k1} {z}
+; CHECK-NEXT:    vpsraw %xmm1, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddw %zmm3, %zmm2, %zmm1
+; CHECK-NEXT:    vpaddw %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <32 x i16> @llvm.x86.avx512.mask.psra.w.512(<32 x i16> %x0, <8 x i16> %x1, <32 x i16> %x2, i32 %x3)
+  %res1 = call <32 x i16> @llvm.x86.avx512.mask.psra.w.512(<32 x i16> %x0, <8 x i16> %x1, <32 x i16> zeroinitializer, i32 %x3)
+  %res2 = call <32 x i16> @llvm.x86.avx512.mask.psra.w.512(<32 x i16> %x0, <8 x i16> %x1, <32 x i16> %x2, 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.psra.wi.512(<32 x i16>, i8, <32 x i16>, i32)
+
+define <32 x i16>@test_int_x86_avx512_mask_psra_wi_512(<32 x i16> %x0, i8 %x1, <32 x i16> %x2, i32 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_wi_512:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovd %esi, %k1
+; CHECK-NEXT:    vpsraw $3, %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vpsraw $3, %zmm0, %zmm2 {%k1} {z}
+; CHECK-NEXT:    vpsraw $3, %zmm0, %zmm0
+; CHECK-NEXT:    vpaddw %zmm2, %zmm1, %zmm1
+; CHECK-NEXT:    vpaddw %zmm0, %zmm1, %zmm0
+; CHECK-NEXT:    retq
+  %res = call <32 x i16> @llvm.x86.avx512.mask.psra.wi.512(<32 x i16> %x0, i8 3, <32 x i16> %x2, i32 %x3)
+  %res1 = call <32 x i16> @llvm.x86.avx512.mask.psra.wi.512(<32 x i16> %x0, i8 3, <32 x i16> zeroinitializer, i32 %x3)
+  %res2 = call <32 x i16> @llvm.x86.avx512.mask.psra.wi.512(<32 x i16> %x0, i8 3, <32 x i16> %x2, i32 -1)
+  %res3 = add <32 x i16> %res, %res1
+  %res4 = add <32 x i16> %res3, %res2
+  ret <32 x i16> %res4
+}

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

 ; CHECK-NEXT:    retq
   %res = call <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3)
   %res1 = call <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> zeroinitializer, i8 %x3)
   %res2 = call <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1)
   %res3 = add <8 x i16> %res, %res1
   %res4 = add <8 x i16> %res3, %res2
   ret <8 x i16> %res4
 }
+
+declare <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16>, <8 x i16>, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psra_w_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_w_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsraw %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vpsraw %xmm1, %xmm0, %xmm3 {%k1} {z}
+; CHECK-NEXT:    vpsraw %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddw %xmm3, %xmm2, %xmm1
+; CHECK-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res3, %res2
+  ret <8 x i16> %res4
+}
+
+declare <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16>, i8, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psra_wi_128(<8 x i16> %x0, i8 %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_wi_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsraw $3, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vpsraw $3, %xmm0, %xmm2 {%k1} {z}
+; CHECK-NEXT:    vpsraw $3, %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.psra.wi.128(<8 x i16> %x0, i8 3, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16> %x0, i8 3, <8 x i16> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16> %x0, i8 3, <8 x i16> %x2, 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.psra.w.256(<16 x i16>, <8 x i16>, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psra_w_256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_w_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vpsraw %xmm1, %ymm0, %ymm2 {%k1}
+; CHECK-NEXT:    vpsraw %xmm1, %ymm0, %ymm3 {%k1} {z}
+; CHECK-NEXT:    vpsraw %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddw %ymm3, %ymm2, %ymm1
+; CHECK-NEXT:    vpaddw %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 -1)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16>, i8, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psra_wi_256(<16 x i16> %x0, i8 %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_wi_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %esi, %k1
+; CHECK-NEXT:    vpsraw $3, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vpsraw $3, %ymm0, %ymm2 {%k1} {z}
+; CHECK-NEXT:    vpsraw $3, %ymm0, %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.psra.wi.256(<16 x i16> %x0, i8 3, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16> %x0, i8 3, <16 x i16> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16> %x0, i8 3, <16 x i16> %x2, i16 -1)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}