+
+;
+; Shuffle to logical bit shifts
+;
+define <8 x i16> @shuffle_v8i16_z0z2z4z6(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_z0z2z4z6:
+; SSE: # BB#0:
+; SSE-NEXT: pslld $16, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_z0z2z4z6:
+; AVX: # BB#0:
+; AVX-NEXT: vpslld $16, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 0, i32 8, i32 2, i32 8, i32 4, i32 8, i32 6>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_zzz0zzz4(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_zzz0zzz4:
+; SSE: # BB#0:
+; SSE-NEXT: psllq $48, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_zzz0zzz4:
+; AVX: # BB#0:
+; AVX-NEXT: vpsllq $48, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 8, i32 8, i32 0, i32 8, i32 8, i32 8, i32 4>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_zz01zX4X(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_zz01zX4X:
+; SSE: # BB#0:
+; SSE-NEXT: psllq $32, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_zz01zX4X:
+; AVX: # BB#0:
+; AVX-NEXT: vpsllq $32, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 8, i32 0, i32 1, i32 8, i32 undef, i32 4, i32 undef>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_z0X2z456(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_z0X2z456:
+; SSE: # BB#0:
+; SSE-NEXT: psllq $16, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_z0X2z456:
+; AVX: # BB#0:
+; AVX-NEXT: vpsllq $16, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 8, i32 0, i32 undef, i32 2, i32 8, i32 4, i32 5, i32 6>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_1z3zXz7z(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_1z3zXz7z:
+; SSE: # BB#0:
+; SSE-NEXT: psrld $16, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_1z3zXz7z:
+; AVX: # BB#0:
+; AVX-NEXT: vpsrld $16, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 1, i32 8, i32 3, i32 8, i32 undef, i32 8, i32 7, i32 8>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_1X3z567z(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_1X3z567z:
+; SSE: # BB#0:
+; SSE-NEXT: psrlq $16, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_1X3z567z:
+; AVX: # BB#0:
+; AVX-NEXT: vpsrlq $16, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 1, i32 undef, i32 3, i32 8, i32 5, i32 6, i32 7, i32 8>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_23zz67zz(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_23zz67zz:
+; SSE: # BB#0:
+; SSE-NEXT: psrlq $32, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_23zz67zz:
+; AVX: # BB#0:
+; AVX-NEXT: vpsrlq $32, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 2, i32 3, i32 8, i32 8, i32 6, i32 7, i32 8, i32 8>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_3zXXXzzz(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_3zXXXzzz:
+; SSE: # BB#0:
+; SSE-NEXT: psrlq $48, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_3zXXXzzz:
+; AVX: # BB#0:
+; AVX-NEXT: vpsrlq $48, %xmm0, %xmm0
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32><i32 3, i32 8, i32 undef, i32 undef, i32 undef, i32 8, i32 8, i32 8>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_01u3zzuz(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_01u3zzuz:
+; SSE: # BB#0:
+; SSE-NEXT: movq {{.*#+}} xmm0 = xmm0[0],zero
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_01u3zzuz:
+; AVX: # BB#0:
+; AVX-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 undef, i32 3, i32 8, i32 8, i32 undef, i32 8>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_0z234567(<8 x i16> %a) {
+; SSE2-LABEL: shuffle_v8i16_0z234567:
+; SSE2: # BB#0:
+; SSE2-NEXT: andps {{.*}}(%rip), %xmm0
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: shuffle_v8i16_0z234567:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: andps {{.*}}(%rip), %xmm0
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: shuffle_v8i16_0z234567:
+; SSE41: # BB#0:
+; SSE41-NEXT: pxor %xmm1, %xmm1
+; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
+; SSE41-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_0z234567:
+; AVX: # BB#0:
+; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_0zzzz5z7(<8 x i16> %a) {
+; SSE2-LABEL: shuffle_v8i16_0zzzz5z7:
+; SSE2: # BB#0:
+; SSE2-NEXT: andps {{.*}}(%rip), %xmm0
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: shuffle_v8i16_0zzzz5z7:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: andps {{.*}}(%rip), %xmm0
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: shuffle_v8i16_0zzzz5z7:
+; SSE41: # BB#0:
+; SSE41-NEXT: pxor %xmm1, %xmm1
+; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3,4],xmm0[5],xmm1[6],xmm0[7]
+; SSE41-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_0zzzz5z7:
+; AVX: # BB#0:
+; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3,4],xmm0[5],xmm1[6],xmm0[7]
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 8, i32 8, i32 8, i32 8, i32 5, i32 8, i32 7>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_0123456z(<8 x i16> %a) {
+; SSE2-LABEL: shuffle_v8i16_0123456z:
+; SSE2: # BB#0:
+; SSE2-NEXT: andps {{.*}}(%rip), %xmm0
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: shuffle_v8i16_0123456z:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: andps {{.*}}(%rip), %xmm0
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: shuffle_v8i16_0123456z:
+; SSE41: # BB#0:
+; SSE41-NEXT: pxor %xmm1, %xmm1
+; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6],xmm1[7]
+; SSE41-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_0123456z:
+; AVX: # BB#0:
+; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6],xmm1[7]
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 0, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 15>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_fu3ucc5u(<8 x i16> %a, <8 x i16> %b) {
+; SSE-LABEL: shuffle_v8i16_fu3ucc5u:
+; SSE: # BB#0:
+; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
+; SSE-NEXT: pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,7,5,4,4]
+; SSE-NEXT: punpckhdq {{.*#+}} xmm1 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; SSE-NEXT: movdqa %xmm1, %xmm0
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_fu3ucc5u:
+; AVX: # BB#0:
+; AVX-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
+; AVX-NEXT: vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,7,5,4,4]
+; AVX-NEXT: vpunpckhdq {{.*#+}} xmm0 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 15, i32 undef, i32 3, i32 undef, i32 12, i32 12, i32 5, i32 undef>
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @shuffle_v8i16_8012345u(<8 x i16> %a) {
+; SSE-LABEL: shuffle_v8i16_8012345u:
+; SSE: # BB#0:
+; SSE-NEXT: pslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
+; SSE-NEXT: retq
+;
+; AVX-LABEL: shuffle_v8i16_8012345u:
+; AVX: # BB#0:
+; AVX-NEXT: vpslldq {{.*#+}} xmm0 = zero,zero,xmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13]
+; AVX-NEXT: retq
+ %shuffle = shufflevector <8 x i16> %a, <8 x i16> zeroinitializer, <8 x i32> <i32 8, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 undef>
+
+ ret <8 x i16> %shuffle
+}
+
+define <8 x i16> @insert_dup_mem_v8i16_i32(i32* %ptr) {
+; SSE2-LABEL: insert_dup_mem_v8i16_i32:
+; SSE2: # BB#0:
+; SSE2-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
+; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: insert_dup_mem_v8i16_i32:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: insert_dup_mem_v8i16_i32:
+; SSE41: # BB#0:
+; SSE41-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: insert_dup_mem_v8i16_i32:
+; AVX1: # BB#0:
+; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: insert_dup_mem_v8i16_i32:
+; AVX2: # BB#0:
+; AVX2-NEXT: vpbroadcastw (%rdi), %xmm0
+; AVX2-NEXT: retq
+ %tmp = load i32, i32* %ptr, align 4
+ %tmp1 = insertelement <4 x i32> zeroinitializer, i32 %tmp, i32 0
+ %tmp2 = bitcast <4 x i32> %tmp1 to <8 x i16>
+ %tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> undef, <8 x i32> zeroinitializer
+ ret <8 x i16> %tmp3
+}
+
+define <8 x i16> @insert_dup_mem_v8i16_sext_i16(i16* %ptr) {
+; SSE2-LABEL: insert_dup_mem_v8i16_sext_i16:
+; SSE2: # BB#0:
+; SSE2-NEXT: movswl (%rdi), %eax
+; SSE2-NEXT: movd %eax, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
+; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: insert_dup_mem_v8i16_sext_i16:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movswl (%rdi), %eax
+; SSSE3-NEXT: movd %eax, %xmm0
+; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: insert_dup_mem_v8i16_sext_i16:
+; SSE41: # BB#0:
+; SSE41-NEXT: movswl (%rdi), %eax
+; SSE41-NEXT: movd %eax, %xmm0
+; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: insert_dup_mem_v8i16_sext_i16:
+; AVX1: # BB#0:
+; AVX1-NEXT: movswl (%rdi), %eax
+; AVX1-NEXT: vmovd %eax, %xmm0
+; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: insert_dup_mem_v8i16_sext_i16:
+; AVX2: # BB#0:
+; AVX2-NEXT: movswl (%rdi), %eax
+; AVX2-NEXT: vmovd %eax, %xmm0
+; AVX2-NEXT: vpbroadcastw %xmm0, %xmm0
+; AVX2-NEXT: retq
+ %tmp = load i16, i16* %ptr, align 2
+ %tmp1 = sext i16 %tmp to i32
+ %tmp2 = insertelement <4 x i32> zeroinitializer, i32 %tmp1, i32 0
+ %tmp3 = bitcast <4 x i32> %tmp2 to <8 x i16>
+ %tmp4 = shufflevector <8 x i16> %tmp3, <8 x i16> undef, <8 x i32> zeroinitializer
+ ret <8 x i16> %tmp4
+}
+
+define <8 x i16> @insert_dup_elt1_mem_v8i16_i32(i32* %ptr) {
+; SSE2-LABEL: insert_dup_elt1_mem_v8i16_i32:
+; SSE2: # BB#0:
+; SSE2-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
+; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,1,1,4,5,6,7]
+; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,5,5]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: insert_dup_elt1_mem_v8i16_i32:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: insert_dup_elt1_mem_v8i16_i32:
+; SSE41: # BB#0:
+; SSE41-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: insert_dup_elt1_mem_v8i16_i32:
+; AVX1: # BB#0:
+; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: insert_dup_elt1_mem_v8i16_i32:
+; AVX2: # BB#0:
+; AVX2-NEXT: vpbroadcastw 2(%rdi), %xmm0
+; AVX2-NEXT: retq
+ %tmp = load i32, i32* %ptr, align 4
+ %tmp1 = insertelement <4 x i32> zeroinitializer, i32 %tmp, i32 0
+ %tmp2 = bitcast <4 x i32> %tmp1 to <8 x i16>
+ %tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
+ ret <8 x i16> %tmp3
+}
+
+define <8 x i16> @insert_dup_elt3_mem_v8i16_i32(i32* %ptr) {
+; SSE2-LABEL: insert_dup_elt3_mem_v8i16_i32:
+; SSE2: # BB#0:
+; SSE2-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,0,1,0]
+; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,3,3,3,4,5,6,7]
+; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,7,7,7]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: insert_dup_elt3_mem_v8i16_i32:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: insert_dup_elt3_mem_v8i16_i32:
+; SSE41: # BB#0:
+; SSE41-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: insert_dup_elt3_mem_v8i16_i32:
+; AVX1: # BB#0:
+; AVX1-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: insert_dup_elt3_mem_v8i16_i32:
+; AVX2: # BB#0:
+; AVX2-NEXT: vpbroadcastw 2(%rdi), %xmm0
+; AVX2-NEXT: retq
+ %tmp = load i32, i32* %ptr, align 4
+ %tmp1 = insertelement <4 x i32> zeroinitializer, i32 %tmp, i32 1
+ %tmp2 = bitcast <4 x i32> %tmp1 to <8 x i16>
+ %tmp3 = shufflevector <8 x i16> %tmp2, <8 x i16> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+ ret <8 x i16> %tmp3
+}
+
+define <8 x i16> @insert_dup_elt1_mem_v8i16_sext_i16(i16* %ptr) {
+; SSE2-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
+; SSE2: # BB#0:
+; SSE2-NEXT: movswl (%rdi), %eax
+; SSE2-NEXT: movd %eax, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
+; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,1,1,4,5,6,7]
+; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,5,5]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movswl (%rdi), %eax
+; SSSE3-NEXT: movd %eax, %xmm0
+; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
+; SSE41: # BB#0:
+; SSE41-NEXT: movswl (%rdi), %eax
+; SSE41-NEXT: movd %eax, %xmm0
+; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
+; AVX1: # BB#0:
+; AVX1-NEXT: movswl (%rdi), %eax
+; AVX1-NEXT: vmovd %eax, %xmm0
+; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: insert_dup_elt1_mem_v8i16_sext_i16:
+; AVX2: # BB#0:
+; AVX2-NEXT: movswl (%rdi), %eax
+; AVX2-NEXT: shrl $16, %eax
+; AVX2-NEXT: vmovd %eax, %xmm0
+; AVX2-NEXT: vpbroadcastw %xmm0, %xmm0
+; AVX2-NEXT: retq
+ %tmp = load i16, i16* %ptr, align 2
+ %tmp1 = sext i16 %tmp to i32
+ %tmp2 = insertelement <4 x i32> zeroinitializer, i32 %tmp1, i32 0
+ %tmp3 = bitcast <4 x i32> %tmp2 to <8 x i16>
+ %tmp4 = shufflevector <8 x i16> %tmp3, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
+ ret <8 x i16> %tmp4
+}
+
+define <8 x i16> @insert_dup_elt3_mem_v8i16_sext_i16(i16* %ptr) {
+; SSE2-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
+; SSE2: # BB#0:
+; SSE2-NEXT: movswl (%rdi), %eax
+; SSE2-NEXT: movd %eax, %xmm0
+; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,0,1,0]
+; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,3,3,3,4,5,6,7]
+; SSE2-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,7,7,7]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: movswl (%rdi), %eax
+; SSSE3-NEXT: movd %eax, %xmm0
+; SSSE3-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
+; SSE41: # BB#0:
+; SSE41-NEXT: movswl (%rdi), %eax
+; SSE41-NEXT: movd %eax, %xmm0
+; SSE41-NEXT: pshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; SSE41-NEXT: retq
+;
+; AVX1-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
+; AVX1: # BB#0:
+; AVX1-NEXT: movswl (%rdi), %eax
+; AVX1-NEXT: vmovd %eax, %xmm0
+; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[2,3,2,3,2,3,2,3,2,3,2,3,2,3,2,3]
+; AVX1-NEXT: retq
+;
+; AVX2-LABEL: insert_dup_elt3_mem_v8i16_sext_i16:
+; AVX2: # BB#0:
+; AVX2-NEXT: movswl (%rdi), %eax
+; AVX2-NEXT: shrl $16, %eax
+; AVX2-NEXT: vmovd %eax, %xmm0
+; AVX2-NEXT: vpbroadcastw %xmm0, %xmm0
+; AVX2-NEXT: retq
+ %tmp = load i16, i16* %ptr, align 2
+ %tmp1 = sext i16 %tmp to i32
+ %tmp2 = insertelement <4 x i32> zeroinitializer, i32 %tmp1, i32 1
+ %tmp3 = bitcast <4 x i32> %tmp2 to <8 x i16>
+ %tmp4 = shufflevector <8 x i16> %tmp3, <8 x i16> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+ ret <8 x i16> %tmp4
+}