/external/libvpx/libvpx/vp9/common/arm/neon/ |
D | vp9_iht4x4_add_neon.c | 28 int16x8_t *q9s16) { in TRANSPOSE4X4() argument 34 d1x2s16 = vtrn_s16(vget_low_s16(*q9s16), vget_high_s16(*q9s16)); in TRANSPOSE4X4() 41 *q9s16 = vreinterpretq_s16_s32(q0x2s32.val[1]); in TRANSPOSE4X4() 72 int16x8_t *q9s16) { in IDCT4x4_1D() argument 80 d18s16 = vget_low_s16(*q9s16); in IDCT4x4_1D() 81 d19s16 = vget_high_s16(*q9s16); in IDCT4x4_1D() 101 *q9s16 = vsubq_s16(q13s16, q14s16); in IDCT4x4_1D() 102 *q9s16 = vcombine_s16(vget_high_s16(*q9s16), in IDCT4x4_1D() 103 vget_low_s16(*q9s16)); // vswp in IDCT4x4_1D() 113 int16x8_t *q9s16) { in IADST4x4_1D() argument [all …]
|
D | vp9_iht8x8_add_neon.c | 36 int16x8_t *q9s16, in TRANSPOSE8X8() argument 50 d18s16 = vget_low_s16(*q9s16); in TRANSPOSE8X8() 51 d19s16 = vget_high_s16(*q9s16); in TRANSPOSE8X8() 66 *q9s16 = vcombine_s16(d18s16, d26s16); // vswp d19, d26 in TRANSPOSE8X8() 76 q1x2s32 = vtrnq_s32(vreinterpretq_s32_s16(*q9s16), in TRANSPOSE8X8() 93 *q9s16 = q0x2s16.val[1]; in TRANSPOSE8X8() 105 int16x8_t *q9s16, in IDCT8x8_1D() argument 127 d18s16 = vget_low_s16(*q9s16); in IDCT8x8_1D() 128 d19s16 = vget_high_s16(*q9s16); in IDCT8x8_1D() 195 *q9s16 = vcombine_s16(d18s16, d19s16); in IDCT8x8_1D() [all …]
|
/external/libvpx/libvpx/vpx_dsp/arm/ |
D | idct8x8_add_neon.c | 18 int16x8_t *q9s16, in TRANSPOSE8X8() argument 32 d18s16 = vget_low_s16(*q9s16); in TRANSPOSE8X8() 33 d19s16 = vget_high_s16(*q9s16); in TRANSPOSE8X8() 48 *q9s16 = vcombine_s16(d18s16, d26s16); // vswp d19, d26 in TRANSPOSE8X8() 58 q1x2s32 = vtrnq_s32(vreinterpretq_s32_s16(*q9s16), in TRANSPOSE8X8() 75 *q9s16 = q0x2s16.val[1]; in TRANSPOSE8X8() 87 int16x8_t *q9s16, in IDCT8x8_1D() argument 109 d18s16 = vget_low_s16(*q9s16); in IDCT8x8_1D() 110 d19s16 = vget_high_s16(*q9s16); in IDCT8x8_1D() 177 *q9s16 = vcombine_s16(d18s16, d19s16); in IDCT8x8_1D() [all …]
|
D | idct4x4_add_neon.c | 22 int16x8_t q8s16, q9s16, q13s16, q14s16; in vpx_idct4x4_16_add_neon() local 34 q9s16 = vld1q_s16(input + 8); in vpx_idct4x4_16_add_neon() 38 d18s16 = vget_low_s16(q9s16); in vpx_idct4x4_16_add_neon() 39 d19s16 = vget_high_s16(q9s16); in vpx_idct4x4_16_add_neon() 44 q9s16 = vcombine_s16(d1x2s16.val[0], d1x2s16.val[1]); in vpx_idct4x4_16_add_neon() 50 vreinterpretq_s32_s16(q9s16)); in vpx_idct4x4_16_add_neon() 79 q9s16 = vsubq_s16(q13s16, q14s16); in vpx_idct4x4_16_add_neon() 83 d18s16 = vget_high_s16(q9s16); // vswp d18 d19 in vpx_idct4x4_16_add_neon() 84 d19s16 = vget_low_s16(q9s16); in vpx_idct4x4_16_add_neon() 89 q9s16 = vcombine_s16(d1x2s16.val[0], d1x2s16.val[1]); in vpx_idct4x4_16_add_neon() [all …]
|
D | idct32x32_add_neon.c | 30 q6s16, q7s16, q8s16, q9s16); 38 int16x8_t q9s16) { in __STORE_COMBINE_CENTER_RESULTS() argument 50 q9s16 = vrshrq_n_s16(q9s16, 6); in __STORE_COMBINE_CENTER_RESULTS() 57 q9s16 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q9s16), in __STORE_COMBINE_CENTER_RESULTS() 64 d11s16 = vreinterpret_s16_u8(vqmovun_s16(q9s16)); in __STORE_COMBINE_CENTER_RESULTS() 176 int16x8_t q8s16, q9s16, q10s16, q11s16, q12s16, q13s16, q14s16, q15s16; in idct32_transpose_pair() local 184 q9s16 = vld1q_s16(in); in idct32_transpose_pair() 200 d18s16 = vget_low_s16(q9s16); in idct32_transpose_pair() 201 d19s16 = vget_high_s16(q9s16); in idct32_transpose_pair() 216 q9s16 = vcombine_s16(d18s16, d26s16); // vswp d19, d26 in idct32_transpose_pair() [all …]
|
D | idct16x16_add_neon.c | 18 int16x8_t *q9s16, in TRANSPOSE8X8() argument 32 d18s16 = vget_low_s16(*q9s16); in TRANSPOSE8X8() 33 d19s16 = vget_high_s16(*q9s16); in TRANSPOSE8X8() 48 *q9s16 = vcombine_s16(d18s16, d26s16); // vswp d19, d26 in TRANSPOSE8X8() 58 q1x2s32 = vtrnq_s32(vreinterpretq_s32_s16(*q9s16), in TRANSPOSE8X8() 75 *q9s16 = q0x2s16.val[1]; in TRANSPOSE8X8() 96 int16x8_t q8s16, q9s16, q10s16, q11s16, q12s16, q13s16, q14s16, q15s16; in vpx_idct16x16_256_add_neon_pass1() local 105 q9s16 = q0x2s16.val[0]; in vpx_idct16x16_256_add_neon_pass1() 125 TRANSPOSE8X8(&q8s16, &q9s16, &q10s16, &q11s16, in vpx_idct16x16_256_add_neon_pass1() 130 d18s16 = vget_low_s16(q9s16); in vpx_idct16x16_256_add_neon_pass1() [all …]
|
/external/libvpx/libvpx/vp8/common/arm/neon/ |
D | sixtappredict_neon.c | 39 int16x8_t q8s16, q9s16, q10s16, q11s16, q12s16; in vp8_sixtap_predict4x4_neon() local 229 q9s16 = vreinterpretq_s16_u16(q9u16); in vp8_sixtap_predict4x4_neon() 231 q7s16 = vqaddq_s16(q7s16, q9s16); in vp8_sixtap_predict4x4_neon() 334 q9s16 = vreinterpretq_s16_u16(q9u16); in vp8_sixtap_predict4x4_neon() 338 q7s16 = vqaddq_s16(q7s16, q9s16); in vp8_sixtap_predict4x4_neon() 417 int16x8_t q8s16, q9s16, q10s16, q11s16, q12s16; in vp8_sixtap_predict8x4_neon() local 492 q9s16 = vreinterpretq_s16_u16(q9u16); in vp8_sixtap_predict8x4_neon() 497 q9s16 = vqaddq_s16(q9s16, q5s16); in vp8_sixtap_predict8x4_neon() 502 d8u8 = vqrshrun_n_s16(q9s16, 7); in vp8_sixtap_predict8x4_neon() 604 q9s16 = vreinterpretq_s16_u16(q9u16); in vp8_sixtap_predict8x4_neon() [all …]
|