Lines Matching refs:q4_in
568 static WEBP_INLINE int16x8x4_t DistoTranspose4x4S16_NEON(int16x8x4_t q4_in) { in DistoTranspose4x4S16_NEON() argument
569 const int16x8x2_t q2_tmp0 = vtrnq_s16(q4_in.val[0], q4_in.val[1]); in DistoTranspose4x4S16_NEON()
570 const int16x8x2_t q2_tmp1 = vtrnq_s16(q4_in.val[2], q4_in.val[3]); in DistoTranspose4x4S16_NEON()
575 q4_in.val[0] = vreinterpretq_s16_s32(q2_tmp2.val[0]); in DistoTranspose4x4S16_NEON()
576 q4_in.val[2] = vreinterpretq_s16_s32(q2_tmp2.val[1]); in DistoTranspose4x4S16_NEON()
577 q4_in.val[1] = vreinterpretq_s16_s32(q2_tmp3.val[0]); in DistoTranspose4x4S16_NEON()
578 q4_in.val[3] = vreinterpretq_s16_s32(q2_tmp3.val[1]); in DistoTranspose4x4S16_NEON()
579 return q4_in; in DistoTranspose4x4S16_NEON()
583 const int16x8x4_t q4_in) { in DistoHorizontalPass_NEON() argument
586 const int16x8_t q_a0 = vaddq_s16(q4_in.val[0], q4_in.val[2]); in DistoHorizontalPass_NEON()
587 const int16x8_t q_a1 = vaddq_s16(q4_in.val[1], q4_in.val[3]); in DistoHorizontalPass_NEON()
588 const int16x8_t q_a3 = vsubq_s16(q4_in.val[0], q4_in.val[2]); in DistoHorizontalPass_NEON()
589 const int16x8_t q_a2 = vsubq_s16(q4_in.val[1], q4_in.val[3]); in DistoHorizontalPass_NEON()
602 static WEBP_INLINE int16x8x4_t DistoVerticalPass_NEON(const uint8x8x4_t q4_in) { in DistoVerticalPass_NEON() argument
603 const int16x8_t q_a0 = vreinterpretq_s16_u16(vaddl_u8(q4_in.val[0], in DistoVerticalPass_NEON()
604 q4_in.val[2])); in DistoVerticalPass_NEON()
605 const int16x8_t q_a1 = vreinterpretq_s16_u16(vaddl_u8(q4_in.val[1], in DistoVerticalPass_NEON()
606 q4_in.val[3])); in DistoVerticalPass_NEON()
607 const int16x8_t q_a2 = vreinterpretq_s16_u16(vsubl_u8(q4_in.val[1], in DistoVerticalPass_NEON()
608 q4_in.val[3])); in DistoVerticalPass_NEON()
609 const int16x8_t q_a3 = vreinterpretq_s16_u16(vsubl_u8(q4_in.val[0], in DistoVerticalPass_NEON()
610 q4_in.val[2])); in DistoVerticalPass_NEON()
631 static WEBP_INLINE int32x2_t DistoSum_NEON(const int16x8x4_t q4_in, in DistoSum_NEON() argument
638 int32x4_t q_sum0 = vmull_s16(d4_w.val[0], vget_low_s16(q4_in.val[0])); in DistoSum_NEON()
639 int32x4_t q_sum1 = vmull_s16(d4_w.val[1], vget_low_s16(q4_in.val[1])); in DistoSum_NEON()
640 int32x4_t q_sum2 = vmull_s16(d4_w.val[2], vget_low_s16(q4_in.val[2])); in DistoSum_NEON()
641 int32x4_t q_sum3 = vmull_s16(d4_w.val[3], vget_low_s16(q4_in.val[3])); in DistoSum_NEON()
642 q_sum0 = vmlsl_s16(q_sum0, d4_w.val[0], vget_high_s16(q4_in.val[0])); in DistoSum_NEON()
643 q_sum1 = vmlsl_s16(q_sum1, d4_w.val[1], vget_high_s16(q4_in.val[1])); in DistoSum_NEON()
644 q_sum2 = vmlsl_s16(q_sum2, d4_w.val[2], vget_high_s16(q4_in.val[2])); in DistoSum_NEON()
645 q_sum3 = vmlsl_s16(q_sum3, d4_w.val[3], vget_high_s16(q4_in.val[3])); in DistoSum_NEON()