Lines Matching defs:a0
86 // a0 a1 a2 a3 | b0 b1 b2 b3 => a0 b0 c0 d0 | a1 b1 c1 d1
88 const int16x8x2_t tmp0 = vzipq_s16(in0, in1); // a0 c0 a1 c1 a2 c2 ...
308 const int16x8_t a0a1 = vaddq_s16(d0d1, d3d2); // d0+d3 | d1+d2 (=a0|a1)
327 const int16x8_t a0a1 = vaddq_s16(d0d1, d3d2); // d0+d3 | d1+d2 (=a0|a1)
406 "vadd.s16 d4, d0, d3 \n" // a0 = d0 + d3
411 "vadd.s16 d0, d4, d5 \n" // a0 + a1
412 "vshl.s16 d0, d0, #3 \n" // temp[0+i*4] = (a0+a1) << 3
413 "vsub.s16 d2, d4, d5 \n" // a0 - a1
414 "vshl.s16 d2, d2, #3 \n" // (temp[2+i*4] = (a0-a1) << 3
505 // a0 = in[0 * 16] + in[2 * 16]
509 const int32x4_t a0 = vaddl_s16(in.val[0], in.val[2]);
513 tmp0.val[0] = vaddq_s32(a0, a1);
516 tmp0.val[3] = vsubq_s32(a0, a1);
520 // a0 = tmp[0 + i] + tmp[ 8 + i]
524 const int32x4_t a0 = vaddq_s32(tmp1.val[0], tmp1.val[2]);
528 const int32x4_t b0 = vhaddq_s32(a0, a1); // (a0 + a1) >> 1
531 const int32x4_t b3 = vhsubq_s32(a0, a1); // (a0 - a1) >> 1
593 // {a0, a1} = {in[0] + in[2], in[1] + in[3]}
604 // tmp[0] = a0 + a1
607 // tmp[3] = a0 - a1
736 const int16x8_t a0 = vld1q_s16(out + 0);
738 const uint16x8_t a1 = vreinterpretq_u16_s16(vabsq_s16(a0));
760 const uint8x16_t a0 = vld1q_u8(a);
762 const uint8x16_t abs_diff = vabdq_u8(a0, b0);
797 const uint8x8_t a0 = vld1_u8(a + y * BPS);
799 const uint8x8_t abs_diff = vabd_u8(a0, b0);
807 const uint8x16_t a0 = Load4x4(a);
809 const uint8x16_t abs_diff = vabdq_u8(a0, b0);
