+declare <16 x i16> @llvm.bswap.v16i16(<16 x i16>)
+declare <8 x i32> @llvm.bswap.v8i32(<8 x i32>)
+declare <4 x i64> @llvm.bswap.v4i64(<4 x i64>)
+
+define <16 x i16> @test4(<16 x i16> %v) {
+; CHECK-NOSSSE3-LABEL: test4:
+; CHECK-NOSSSE3: # BB#0: # %entry
+; CHECK-NOSSSE3-NEXT: pxor %xmm2, %xmm2
+; CHECK-NOSSSE3-NEXT: movdqa %xmm0, %xmm3
+; CHECK-NOSSSE3-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm3 = xmm3[1,0,3,2,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,5,4,7,6]
+; CHECK-NOSSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,0,3,2,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,4,7,6]
+; CHECK-NOSSSE3-NEXT: packuswb %xmm3, %xmm0
+; CHECK-NOSSSE3-NEXT: movdqa %xmm1, %xmm3
+; CHECK-NOSSSE3-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm3 = xmm3[1,0,3,2,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,5,4,7,6]
+; CHECK-NOSSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[1,0,3,2,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,4,7,6]
+; CHECK-NOSSSE3-NEXT: packuswb %xmm3, %xmm1
+; CHECK-NOSSSE3-NEXT: retq
+;
+; CHECK-SSSE3-LABEL: test4:
+; CHECK-SSSE3: # BB#0: # %entry
+; CHECK-SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14]
+; CHECK-SSSE3-NEXT: pshufb %xmm2, %xmm0
+; CHECK-SSSE3-NEXT: pshufb %xmm2, %xmm1
+; CHECK-SSSE3-NEXT: retq
+;
+; CHECK-AVX2-LABEL: test4:
+; CHECK-AVX2: # BB#0: # %entry
+; CHECK-AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14,17,16,19,18,21,20,23,22,25,24,27,26,29,28,31,30]
+; CHECK-AVX2-NEXT: retq
+;
+; CHECK-WIDE-AVX2-LABEL: test4:
+; CHECK-WIDE-AVX2: # BB#0: # %entry
+; CHECK-WIDE-AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14,17,16,19,18,21,20,23,22,25,24,27,26,29,28,31,30]
+; CHECK-WIDE-AVX2-NEXT: retq
+entry:
+ %r = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> %v)
+ ret <16 x i16> %r
+}
+
+define <8 x i32> @test5(<8 x i32> %v) {
+; CHECK-NOSSSE3-LABEL: test5:
+; CHECK-NOSSSE3: # BB#0: # %entry
+; CHECK-NOSSSE3-NEXT: pxor %xmm2, %xmm2
+; CHECK-NOSSSE3-NEXT: movdqa %xmm0, %xmm3
+; CHECK-NOSSSE3-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm3 = xmm3[3,2,1,0,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,7,6,5,4]
+; CHECK-NOSSSE3-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
+; CHECK-NOSSSE3-NEXT: packuswb %xmm3, %xmm0
+; CHECK-NOSSSE3-NEXT: movdqa %xmm1, %xmm3
+; CHECK-NOSSSE3-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm3 = xmm3[3,2,1,0,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,7,6,5,4]
+; CHECK-NOSSSE3-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; CHECK-NOSSSE3-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[3,2,1,0,4,5,6,7]
+; CHECK-NOSSSE3-NEXT: pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,7,6,5,4]
+; CHECK-NOSSSE3-NEXT: packuswb %xmm3, %xmm1
+; CHECK-NOSSSE3-NEXT: retq
+;
+; CHECK-SSSE3-LABEL: test5:
+; CHECK-SSSE3: # BB#0: # %entry
+; CHECK-SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12]
+; CHECK-SSSE3-NEXT: pshufb %xmm2, %xmm0
+; CHECK-SSSE3-NEXT: pshufb %xmm2, %xmm1
+; CHECK-SSSE3-NEXT: retq
+;
+; CHECK-AVX2-LABEL: test5:
+; CHECK-AVX2: # BB#0: # %entry
+; CHECK-AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12,19,18,17,16,23,22,21,20,27,26,25,24,31,30,29,28]
+; CHECK-AVX2-NEXT: retq
+;
+; CHECK-WIDE-AVX2-LABEL: test5:
+; CHECK-WIDE-AVX2: # BB#0: # %entry
+; CHECK-WIDE-AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12,19,18,17,16,23,22,21,20,27,26,25,24,31,30,29,28]
+; CHECK-WIDE-AVX2-NEXT: retq
+entry:
+ %r = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> %v)
+ ret <8 x i32> %r
+}