-; SSE2-NEXT: pxor %xmm8, %xmm8
-; SSE2-NEXT: movdqa %xmm0, %xmm4
-; SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm8[8],xmm4[9],xmm8[9],xmm4[10],xmm8[10],xmm4[11],xmm8[11],xmm4[12],xmm8[12],xmm4[13],xmm8[13],xmm4[14],xmm8[14],xmm4[15],xmm8[15]
-; SSE2-NEXT: movdqa %xmm0, %xmm6
-; SSE2-NEXT: punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm8[0],xmm6[1],xmm8[1],xmm6[2],xmm8[2],xmm6[3],xmm8[3],xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
-; SSE2-NEXT: punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm4[4],xmm6[5],xmm4[5],xmm6[6],xmm4[6],xmm6[7],xmm4[7]
-; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm6[0,1,2,1]
-; SSE2-NEXT: pshufhw {{.*#+}} xmm6 = xmm4[0,1,2,3,4,5,7,7]
-; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535,0,65535,65535,65535,0,65535]
-; SSE2-NEXT: movdqa %xmm4, %xmm7
-; SSE2-NEXT: pandn %xmm6, %xmm7
-; SSE2-NEXT: movdqa %xmm2, %xmm6
-; SSE2-NEXT: punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm8[8],xmm6[9],xmm8[9],xmm6[10],xmm8[10],xmm6[11],xmm8[11],xmm6[12],xmm8[12],xmm6[13],xmm8[13],xmm6[14],xmm8[14],xmm6[15],xmm8[15]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm6 = xmm6[3,1,2,3,4,5,6,7]
-; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[0,3,2,3]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm6 = xmm6[1,0,2,3,4,5,6,7]
-; SSE2-NEXT: movdqa %xmm2, %xmm5
-; SSE2-NEXT: punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm8[0],xmm5[1],xmm8[1],xmm5[2],xmm8[2],xmm5[3],xmm8[3],xmm5[4],xmm8[4],xmm5[5],xmm8[5],xmm5[6],xmm8[6],xmm5[7],xmm8[7]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm5 = xmm5[3,1,2,3,4,5,6,7]
-; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[0,3,2,3]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm5 = xmm5[1,0,2,3,4,5,6,7]
-; SSE2-NEXT: punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm6[0]
-; SSE2-NEXT: pand %xmm4, %xmm5
-; SSE2-NEXT: por %xmm7, %xmm5
-; SSE2-NEXT: packuswb %xmm0, %xmm5
-; SSE2-NEXT: movdqa {{.*#+}} xmm9 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT: pand %xmm9, %xmm2
-; SSE2-NEXT: movdqa {{.*#+}} xmm7 = [0,65535,65535,65535,0,65535,65535,65535]
-; SSE2-NEXT: movdqa %xmm7, %xmm6
-; SSE2-NEXT: pandn %xmm2, %xmm6
-; SSE2-NEXT: pand %xmm9, %xmm0
-; SSE2-NEXT: pand %xmm7, %xmm0
-; SSE2-NEXT: por %xmm6, %xmm0
-; SSE2-NEXT: packuswb %xmm0, %xmm0
-; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
-; SSE2-NEXT: movdqa %xmm1, %xmm2
-; SSE2-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm8[8],xmm2[9],xmm8[9],xmm2[10],xmm8[10],xmm2[11],xmm8[11],xmm2[12],xmm8[12],xmm2[13],xmm8[13],xmm2[14],xmm8[14],xmm2[15],xmm8[15]
-; SSE2-NEXT: movdqa %xmm1, %xmm5
-; SSE2-NEXT: punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm8[0],xmm5[1],xmm8[1],xmm5[2],xmm8[2],xmm5[3],xmm8[3],xmm5[4],xmm8[4],xmm5[5],xmm8[5],xmm5[6],xmm8[6],xmm5[7],xmm8[7]
-; SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm2[4],xmm5[5],xmm2[5],xmm5[6],xmm2[6],xmm5[7],xmm2[7]
-; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm5[0,1,2,1]
-; SSE2-NEXT: pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,7,7]
-; SSE2-NEXT: movdqa %xmm3, %xmm5
-; SSE2-NEXT: punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm8[8],xmm5[9],xmm8[9],xmm5[10],xmm8[10],xmm5[11],xmm8[11],xmm5[12],xmm8[12],xmm5[13],xmm8[13],xmm5[14],xmm8[14],xmm5[15],xmm8[15]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm5 = xmm5[3,1,2,3,4,5,6,7]
-; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[0,3,2,3]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm5 = xmm5[1,0,2,3,4,5,6,7]
-; SSE2-NEXT: movdqa %xmm3, %xmm6
-; SSE2-NEXT: punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm8[0],xmm6[1],xmm8[1],xmm6[2],xmm8[2],xmm6[3],xmm8[3],xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm6 = xmm6[3,1,2,3,4,5,6,7]
-; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[0,3,2,3]
-; SSE2-NEXT: pshuflw {{.*#+}} xmm6 = xmm6[1,0,2,3,4,5,6,7]
-; SSE2-NEXT: punpcklqdq {{.*#+}} xmm6 = xmm6[0],xmm5[0]
-; SSE2-NEXT: pand %xmm4, %xmm6
-; SSE2-NEXT: pandn %xmm2, %xmm4
-; SSE2-NEXT: por %xmm6, %xmm4
-; SSE2-NEXT: packuswb %xmm0, %xmm4
-; SSE2-NEXT: pand %xmm9, %xmm3
-; SSE2-NEXT: pand %xmm9, %xmm1
-; SSE2-NEXT: pand %xmm7, %xmm1
-; SSE2-NEXT: pandn %xmm3, %xmm7
-; SSE2-NEXT: por %xmm7, %xmm1
-; SSE2-NEXT: packuswb %xmm0, %xmm1
-; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]