| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vp8/common/arm/neon/ |
| D | dequant_idct_neon.c | 24 int16x8_t q1, q2, q3, q4, q5, q6; in vp8_dequant_idct_add_neon() local
55 q2 = vreinterpretq_s16_u16(vmulq_u16(vreinterpretq_u16_s16(q4), in vp8_dequant_idct_add_neon()
58 d12 = vqadd_s16(vget_low_s16(q1), vget_low_s16(q2)); in vp8_dequant_idct_add_neon()
59 d13 = vqsub_s16(vget_low_s16(q1), vget_low_s16(q2)); in vp8_dequant_idct_add_neon()
61 q2 = vcombine_s16(vget_high_s16(q1), vget_high_s16(q2)); in vp8_dequant_idct_add_neon()
63 q3 = vqdmulhq_n_s16(q2, sinpi8sqrt2); in vp8_dequant_idct_add_neon()
64 q4 = vqdmulhq_n_s16(q2, cospi8sqrt2minus1); in vp8_dequant_idct_add_neon()
69 q3 = vqaddq_s16(q3, q2); in vp8_dequant_idct_add_neon()
70 q4 = vqaddq_s16(q4, q2); in vp8_dequant_idct_add_neon()
88 q2 = vcombine_s16(d2tmp2.val[1], d2tmp3.val[1]); in vp8_dequant_idct_add_neon()
[all …]
|
| D | loopfilter_neon.asm | 34 vdup.u8 q2, r3 ; duplicate thresh
42 vld1.u8 {q9}, [r2@128] ; q2
71 vdup.u8 q2, r12 ; duplicate thresh
88 vld1.u8 {d18}, [r3@64], r1 ; q2
89 vld1.u8 {d19}, [r12@64], r1 ; q2
154 vdup.u8 q2, r3 ; duplicate thresh
244 vdup.u8 q2, r12 ; duplicate thresh
294 ; q2 thresh
301 ; q9 q2
310 vabd.u8 q3, q9, q8 ; abs(q2 - q1)
[all …]
|
| D | loopfiltersimplehorizontaledge_neon.asm | 49 vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
54 vmul.s16 q2, q2, q13 ; 3 * ( qs0 - ps0)
60 vaddw.s8 q2, q2, d8 ; vp8_filter + 3 * ( qs0 - ps0)
63 …vqmovn.s16 d8, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs…
68 vqadd.s8 q2, q14, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
70 vshr.s8 q2, q2, #3 ; Filter2 >>= 3
76 vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
|
| D | buildintrapredictorsmby_neon.asm | 60 vpaddl.u8 q2, q1
61 vpaddl.u16 q3, q2
180 vdup.u8 q2, r5
184 vst1.u8 {q2}, [r1]!
193 vdup.u8 q2, r5
197 vst1.u8 {q2}, [r1]!
207 vdup.u8 q2, r5
211 vst1.u8 {q2}, [r1]!
220 vdup.u8 q2, r5
224 vst1.u8 {q2}, [r1]!
[all …]
|
| D | idct_dequant_full_2x_neon.asm | 26 vld1.16 {q2, q3}, [r0] ; l q
44 vmul.i16 q2, q2, q0
51 ; q2: l0r0 q3: l8r8
66 vqadd.s16 q10, q2, q3 ; a1 = 0 + 8
67 vqsub.s16 q11, q2, q3 ; b1 = 0 - 8
87 vqsub.s16 q2, q6, q5
95 vqadd.s16 q5, q11, q2
96 vqsub.s16 q6, q11, q2
119 vqadd.s16 q2, q4, q6 ; a1 = 0 + 2
140 vqadd.s16 q4, q2, q9
[all …]
|
| D | vp8_subpixelvariance16x16s_neon.asm | 56 vext.8 q3, q2, q3, #1
62 vrhadd.u8 q1, q2, q3
63 vrhadd.u8 q2, q4, q5
72 vsubl.u8 q2, d6, d28
97 vpadal.s16 q8, q2
145 vld1.u8 {q2}, [r0], r1
153 vrhadd.u8 q0, q0, q2
155 vrhadd.u8 q2, q2, q4
165 vsubl.u8 q2, d12, d14
190 vpadal.s16 q8, q2
[all …]
|
| D | mbloopfilter_neon.asm | 34 vdup.u8 q2, r12 ; thresh
43 vld1.u8 {q9}, [r0@128], r1 ; q2
77 vdup.u8 q2, r12 ; thresh
93 vld1.u8 {d18}, [r0@64], r1 ; q2
94 vld1.u8 {d19}, [r12@64], r1 ; q2
135 vdup.s8 q2, r12 ; thresh
229 vdup.u8 q2, r12 ; thresh
318 ; q2 thresh
325 ; q9 q2
335 vabd.u8 q1, q9, q8 ; abs(q2 - q1)
[all …]
|
| D | loopfiltersimpleverticaledge_neon.asm | 69 vsubl.s8 q2, d8, d10 ; ( qs0 - ps0)
74 vmul.s16 q2, q2, q11 ; 3 * ( qs0 - ps0)
80 vaddw.s8 q2, q2, d28 ; vp8_filter + 3 * ( qs0 - ps0)
83 …vqmovn.s16 d28, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( q…
91 vqadd.s8 q2, q14, q11 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
93 vshr.s8 q2, q2, #3 ; Filter2 >>= 3
97 vqadd.s8 q11, q5, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
|
| /hardware/invensense/6515/libsensors_iio/software/core/mllite/ |
| D | ml_math_func.c | 41 float cgcross[4], q1[4], q2[4], qi[4]; in inv_compass_angle() local
53 inv_q_multf(q1, qi, q2); in inv_compass_angle()
56 if ((q2[2] == 0.f) && (q2[1] == 0.f)) in inv_compass_angle()
60 angW = -atan2f(q2[2], q2[1]); in inv_compass_angle()
164 void inv_q_mult(const long *q1, const long *q2, long *qProd) in inv_q_mult() argument
167 qProd[0] = inv_q30_mult(q1[0], q2[0]) - inv_q30_mult(q1[1], q2[1]) - in inv_q_mult()
168 inv_q30_mult(q1[2], q2[2]) - inv_q30_mult(q1[3], q2[3]); in inv_q_mult()
170 qProd[1] = inv_q30_mult(q1[0], q2[1]) + inv_q30_mult(q1[1], q2[0]) + in inv_q_mult()
171 inv_q30_mult(q1[2], q2[3]) - inv_q30_mult(q1[3], q2[2]); in inv_q_mult()
173 qProd[2] = inv_q30_mult(q1[0], q2[2]) - inv_q30_mult(q1[1], q2[3]) + in inv_q_mult()
[all …]
|
| D | ml_math_func.h | 71 void inv_q_mult(const long *q1, const long *q2, long *qProd);
72 void inv_q_add(long *q1, long *q2, long *qSum);
75 void inv_q_multf(const float *q1, const float *q2, float *qProd);
76 void inv_q_addf(const float *q1, const float *q2, float *qSum);
|
| /hardware/invensense/65xx/libsensors_iio/software/core/mllite/ |
| D | ml_math_func.c | 41 float cgcross[4], q1[4], q2[4], qi[4]; in inv_compass_angle() local
53 inv_q_multf(q1, qi, q2); in inv_compass_angle()
56 if ((q2[2] == 0.f) && (q2[1] == 0.f)) in inv_compass_angle()
60 angW = -atan2f(q2[2], q2[1]); in inv_compass_angle()
164 void inv_q_mult(const long *q1, const long *q2, long *qProd) in inv_q_mult() argument
167 qProd[0] = inv_q30_mult(q1[0], q2[0]) - inv_q30_mult(q1[1], q2[1]) - in inv_q_mult()
168 inv_q30_mult(q1[2], q2[2]) - inv_q30_mult(q1[3], q2[3]); in inv_q_mult()
170 qProd[1] = inv_q30_mult(q1[0], q2[1]) + inv_q30_mult(q1[1], q2[0]) + in inv_q_mult()
171 inv_q30_mult(q1[2], q2[3]) - inv_q30_mult(q1[3], q2[2]); in inv_q_mult()
173 qProd[2] = inv_q30_mult(q1[0], q2[2]) - inv_q30_mult(q1[1], q2[3]) + in inv_q_mult()
[all …]
|
| D | ml_math_func.h | 71 void inv_q_mult(const long *q1, const long *q2, long *qProd);
72 void inv_q_add(long *q1, long *q2, long *qSum);
75 void inv_q_multf(const float *q1, const float *q2, float *qProd);
76 void inv_q_addf(const float *q1, const float *q2, float *qSum);
|
| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vp9/common/ |
| D | vp9_loopfilter_filters.c | 25 uint8_t q2, uint8_t q3) { in filter_mask() argument
31 mask |= (abs(q2 - q1) > limit) * -1; in filter_mask()
32 mask |= (abs(q3 - q2) > limit) * -1; in filter_mask()
41 uint8_t q2, uint8_t q3) { in flat_mask4() argument
46 mask |= (abs(q2 - q0) > thresh) * -1; in flat_mask4()
56 uint8_t q1, uint8_t q2, in flat_mask5() argument
58 int8_t mask = ~flat_mask4(thresh, p3, p2, p1, p0, q0, q1, q2, q3); in flat_mask5()
114 const uint8_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p]; in vp9_lpf_horizontal_4_c() local
116 p3, p2, p1, p0, q0, q1, q2, q3); in vp9_lpf_horizontal_4_c()
139 const uint8_t q0 = s[0], q1 = s[1], q2 = s[2], q3 = s[3]; in vp9_lpf_vertical_4_c() local
[all …]
|
| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vp9/common/arm/neon/ |
| D | vp9_loopfilter_16_neon.asm | 63 vld1.u8 {q9}, [r2@64] ; q2
89 ; q2 thresh
96 ; q9 q2
111 vabd.u8 q3, q9, q8 ; m5 = abs(q2 - q1)
112 vabd.u8 q4, q10, q9 ; m6 = abs(q3 - q2)
126 vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
127 vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
130 vabd.u8 q2, q5, q8 ; a = abs(p1 - q1)
137 vshr.u8 q2, q2, #1 ; a = a / 2
141 vqadd.u8 q9, q9, q2 ; a = b + a
[all …]
|
| D | vp9_avg_neon.asm | 37 vld1.8 {q2-q3}, [r0], lr
43 vrhadd.u8 q2, q2, q10
46 vst1.8 {q2-q3}, [r2@128], r4
53 vld1.8 {q2-q3}, [r0], r1
61 vrhadd.u8 q2, q2, q10
65 vst1.8 {q2-q3}, [r2@128], r3
73 vld1.8 {q2}, [r6@128], r3
77 vrhadd.u8 q0, q0, q2
|
| D | vp9_reconintra_neon.asm | 315 vdup.u16 q2, r2
317 vadd.s16 q2, q2, q3
319 vqmovun.s16 d1, q2
327 vdup.u16 q2, r2
329 vadd.s16 q2, q2, q3
331 vqmovun.s16 d1, q2
431 vsubl.u8 q2, d2, d0
444 vadd.s16 q1, q0, q2
446 vadd.s16 q11, q8, q2
458 vadd.s16 q1, q0, q2
[all …]
|
| D | vp9_iht8x8_add_neon.asm | 131 vmull.s16 q2, d18, d0
139 vmlsl.s16 q2, d30, d1
147 vqrshrn.s32 d8, q2, #14 ; >> 14
155 vmull.s16 q2, d18, d1
163 vmlal.s16 q2, d30, d0
171 vqrshrn.s32 d14, q2, #14 ; >> 14
182 vmull.s16 q2, d16, d0
190 vmlal.s16 q2, d24, d0
201 vqrshrn.s32 d18, q2, #14 ; >> 14
209 vmull.s16 q2, d20, d0
[all …]
|
| D | vp9_idct16x16_add_neon.asm | 60 vld2.s16 {q1,q2}, [r0]!
88 vmull.s16 q2, d18, d0
96 vmlsl.s16 q2, d30, d1
107 vqrshrn.s32 d8, q2, #14 ; >> 14
124 vmull.s16 q2, d26, d2
132 vmlsl.s16 q2, d22, d3
140 vqrshrn.s32 d10, q2, #14 ; >> 14
151 vmull.s16 q2, d16, d30
166 vadd.s32 q3, q2, q0
170 vsub.s32 q13, q2, q0
[all …]
|
| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vp9/common/mips/dspr2/ |
| D | vp9_loopfilter_masks_dspr2.h | 31 uint32_t q2, uint32_t q3, in vp9_filter_hev_mask_dspr2() argument
95 [q2] "r" (q2), [q3] "r" (q3), [thresh] "r" (thresh) in vp9_filter_hev_mask_dspr2()
138 uint32_t q2, uint32_t q3, in vp9_filter_hev_mask_flatmask4_dspr2() argument
243 [q2] "r" (q2), [q3] "r" (q3), [thresh] "r" (thresh), in vp9_filter_hev_mask_flatmask4_dspr2()
285 uint32_t q1, uint32_t q2, in vp9_flatmask5() argument
362 [q2] "r" (q2), [q3] "r" (q3), [q4] "r" (q4), in vp9_flatmask5()
|
| D | vp9_mbloop_loopfilter_dspr2.c | 36 uint32_t p3, p2, p1, p0, q0, q1, q2, q3; in vp9_lpf_horizontal_8_dspr2() local
79 [q3] "=&r" (q3), [q2] "=&r" (q2), [q1] "=&r" (q1), [q0] "=&r" (q0) in vp9_lpf_horizontal_8_dspr2()
85 p1, p0, p3, p2, q0, q1, q2, q3, in vp9_lpf_horizontal_8_dspr2()
127 [q0] "r" (q0), [q1] "r" (q1), [q2] "r" (q2), in vp9_lpf_horizontal_8_dspr2()
232 [q0] "+r" (q0), [q1] "+r" (q1), [q2] "+r" (q2), in vp9_lpf_horizontal_8_dspr2()
334 uint32_t p3, p2, p1, p0, q3, q2, q1, q0; in vp9_lpf_vertical_8_dspr2() local
374 [q0] "=&r" (q0), [q1] "=&r" (q1), [q2] "=&r" (q2), [q3] "=&r" (q3) in vp9_lpf_vertical_8_dspr2()
449 [q3] "+r" (q3), [q2] "+r" (q2), [q1] "+r" (q1), [q0] "+r" (q0), in vp9_lpf_vertical_8_dspr2()
455 p1, p0, p3, p2, q0, q1, q2, q3, in vp9_lpf_vertical_8_dspr2()
570 [q0] "+r" (q0), [q1] "+r" (q1), [q2] "+r" (q2), in vp9_lpf_vertical_8_dspr2()
|
| D | vp9_loopfilter_filters_dspr2.h | 377 const uint32_t q0 = *oq0, q1 = *oq1, q2 = *oq2, q3 = *oq3; in vp9_mbfilter_dspr2() local
436 [p2] "r" (p2), [q2] "r" (q2), [p3] "r" (p3), [q3] "r" (q3), in vp9_mbfilter_dspr2()
451 uint32_t q2, uint32_t q3, in vp9_mbfilter1_dspr2() argument
514 [p2] "r" (p2), [q2] "r" (q2), [p3] "r" (p3), [q3] "r" (q3), in vp9_mbfilter1_dspr2()
536 const uint32_t q0 = *oq0, q1 = *oq1, q2 = *oq2, q3 = *oq3; in vp9_wide_mbfilter_dspr2() local
565 [q0] "r" (q0), [q1] "r" (q1), [q2] "r" (q2), [q3] "r" (q3), in vp9_wide_mbfilter_dspr2()
651 [q2] "r" (q2), [q1] "r" (q1), in vp9_wide_mbfilter_dspr2()
744 [q3] "r" (q3), [q2] "r" (q2), [q1] "r" (q1), [q0] "r" (q0), in vp9_wide_mbfilter_dspr2()
|
| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vp8/encoder/arm/neon/ |
| D | shortfdct_neon.asm | 55 vshl.s16 q2, q2, #3 ; (a1, b1) << 3
120 vld1.16 {q2}, [r0@128], r2
124 ; transpose q0=ip[0], q1=ip[1], q2=ip[2], q3=ip[3]
125 vtrn.32 q0, q2 ; [A0|B0]
128 vtrn.16 q2, q3 ; [A3|B3]
131 vadd.s16 q12, q1, q2 ; b1 = ip[1] + ip[2]
132 vsub.s16 q13, q1, q2 ; c1 = ip[1] - ip[2]
141 vsub.s16 q2, q11, q12 ; [A2 | B2] = a1 - b1
165 ; transpose q0=ip[0], q1=ip[4], q2=ip[8], q3=ip[12]
166 vtrn.32 q0, q2 ; q0=[A0 | B0]
[all …]
|
| D | fastquantizeb_neon.asm | 41 vshr.s16 q2, q0, #15 ; sz
63 veor.s16 q4, q2 ; y^sz
79 vsub.s16 q4, q2 ; x1=(y^sz)-sz = (y^sz)-(-1) (2's complement)
98 vmul.s16 q2, q6, q4 ; x * Dequant
115 vst1.s16 {q2, q3}, [r9] ; store dqcoeff = x * Dequant
124 vtst.16 q2, q10, q8 ; now find eob
129 vand q10, q6, q2 ; get all valid numbers from scan array
174 vshr.s16 q2, q0, #15 ; sz
205 veor.s16 q12, q2 ; y^sz
208 vsub.s16 q12, q2 ; x1=(y^sz)-sz = (y^sz)-(-1) (2's complement)
|
| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vpx_scale/arm/neon/ |
| D | vp8_vpxyv12_extendframeborders_neon.asm | 52 vmov q3, q2
63 vst1.8 {q2, q3}, [r6], lr
89 vld1.8 {q2, q3}, [r1]!
103 vst1.8 {q2, q3}, [r5]!
169 vst1.8 {q2}, [r5], lr
200 vld1.8 {q2, q3}, [r1]!
210 vst1.8 {q2, q3}, [r5]!
241 vld1.8 {q2}, [r2]!
249 vst1.8 {q2}, [r6], lr
251 vst1.8 {q2}, [r6], lr
[all …]
|
| /hardware/intel/common/omx-components/videocodec/libvpx_internal/libvpx/vp9/common/x86/ |
| D | vp9_loopfilter_intrin_sse2.c | 388 __m128i p4, p3, p2, p1, p0, q0, q1, q2, q3, q4; in mb_lpf_horizontal_edge_w_sse2_16() local
399 q2 = _mm_loadu_si128((__m128i *)(s + 2 * p)); in mb_lpf_horizontal_edge_w_sse2_16()
410 _mm_store_si128((__m128i *)&aq[2 * 16], q2); in mb_lpf_horizontal_edge_w_sse2_16()
444 work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(q2, q1), in mb_lpf_horizontal_edge_w_sse2_16()
445 _mm_subs_epu8(q1, q2)), in mb_lpf_horizontal_edge_w_sse2_16()
446 _mm_or_si128(_mm_subs_epu8(q3, q2), in mb_lpf_horizontal_edge_w_sse2_16()
447 _mm_subs_epu8(q2, q3))); in mb_lpf_horizontal_edge_w_sse2_16()
514 _mm_or_si128(_mm_subs_epu8(q2, q0), in mb_lpf_horizontal_edge_w_sse2_16()
515 _mm_subs_epu8(q0, q2))); in mb_lpf_horizontal_edge_w_sse2_16()
596 q2 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(&aq[2 * 16] + off)), in mb_lpf_horizontal_edge_w_sse2_16()
[all …]
|