| /external/webrtc/webrtc/modules/audio_processing/aec/ |
| D | aec_rdft_neon.c | 26 const float32x4_t vec_swap_sign = vld1q_f32((float32_t*)k_swap_sign); in cft1st_128_neon()
30 float32x4_t a00v = vld1q_f32(&a[j + 0]); in cft1st_128_neon()
31 float32x4_t a04v = vld1q_f32(&a[j + 4]); in cft1st_128_neon()
32 float32x4_t a08v = vld1q_f32(&a[j + 8]); in cft1st_128_neon()
33 float32x4_t a12v = vld1q_f32(&a[j + 12]); in cft1st_128_neon()
34 float32x4_t a01v = vcombine_f32(vget_low_f32(a00v), vget_low_f32(a08v)); in cft1st_128_neon()
35 float32x4_t a23v = vcombine_f32(vget_high_f32(a00v), vget_high_f32(a08v)); in cft1st_128_neon()
36 float32x4_t a45v = vcombine_f32(vget_low_f32(a04v), vget_low_f32(a12v)); in cft1st_128_neon()
37 float32x4_t a67v = vcombine_f32(vget_high_f32(a04v), vget_high_f32(a12v)); in cft1st_128_neon()
38 const float32x4_t wk1rv = vld1q_f32(&rdft_wk1r[k2]); in cft1st_128_neon()
[all …]
|
| D | aec_core_neon.c | 55 const float32x4_t x_fft_buf_re = vld1q_f32(&x_fft_buf[0][xPos + j]); in FilterFarNEON()
56 const float32x4_t x_fft_buf_im = vld1q_f32(&x_fft_buf[1][xPos + j]); in FilterFarNEON()
57 const float32x4_t h_fft_buf_re = vld1q_f32(&h_fft_buf[0][pos + j]); in FilterFarNEON()
58 const float32x4_t h_fft_buf_im = vld1q_f32(&h_fft_buf[1][pos + j]); in FilterFarNEON()
59 const float32x4_t y_fft_re = vld1q_f32(&y_fft[0][j]); in FilterFarNEON()
60 const float32x4_t y_fft_im = vld1q_f32(&y_fft[1][j]); in FilterFarNEON()
61 const float32x4_t a = vmulq_f32(x_fft_buf_re, h_fft_buf_re); in FilterFarNEON()
62 const float32x4_t e = vmlsq_f32(a, x_fft_buf_im, h_fft_buf_im); in FilterFarNEON()
63 const float32x4_t c = vmulq_f32(x_fft_buf_re, h_fft_buf_im); in FilterFarNEON()
64 const float32x4_t f = vmlaq_f32(c, x_fft_buf_im, h_fft_buf_re); in FilterFarNEON()
[all …]
|
| /external/clang/test/CodeGen/ |
| D | aarch64-neon-fma.c | 25 float32x4_t test_vmlaq_n_f32(float32x4_t a, float32x4_t b, float32_t c) { in test_vmlaq_n_f32()
47 float32x4_t test_vmlsq_n_f32(float32x4_t a, float32x4_t b, float32_t c) { in test_vmlsq_n_f32()
85 float32x4_t test_vmlaq_lane_f32_0(float32x4_t a, float32x4_t b, float32x2_t v) { in test_vmlaq_lane_f32_0()
94 float32x2_t test_vmla_laneq_f32_0(float32x2_t a, float32x2_t b, float32x4_t v) { in test_vmla_laneq_f32_0()
103 float32x4_t test_vmlaq_laneq_f32_0(float32x4_t a, float32x4_t b, float32x4_t v) { in test_vmlaq_laneq_f32_0()
121 float32x4_t test_vmlsq_lane_f32_0(float32x4_t a, float32x4_t b, float32x2_t v) { in test_vmlsq_lane_f32_0()
130 float32x2_t test_vmls_laneq_f32_0(float32x2_t a, float32x2_t b, float32x4_t v) { in test_vmls_laneq_f32_0()
139 float32x4_t test_vmlsq_laneq_f32_0(float32x4_t a, float32x4_t b, float32x4_t v) { in test_vmlsq_laneq_f32_0()
157 float32x4_t test_vmlaq_lane_f32(float32x4_t a, float32x4_t b, float32x2_t v) { in test_vmlaq_lane_f32()
166 float32x2_t test_vmla_laneq_f32(float32x2_t a, float32x2_t b, float32x4_t v) { in test_vmla_laneq_f32()
[all …]
|
| D | arm-neon-directed-rounding.c | 23 float32x4_t test_vrndaq_f32(float32x4_t a) { in test_vrndaq_f32()
45 float32x4_t test_vrndmq_f32(float32x4_t a) { in test_vrndmq_f32()
67 float32x4_t test_vrndnq_f32(float32x4_t a) { in test_vrndnq_f32()
89 float32x4_t test_vrndpq_f32(float32x4_t a) { in test_vrndpq_f32()
111 float32x4_t test_vrndxq_f32(float32x4_t a) { in test_vrndxq_f32()
133 float32x4_t test_vrndq_f32(float32x4_t a) { in test_vrndq_f32()
|
| D | aarch64-neon-misc.c | 197 uint32x4_t test_vceqzq_f32(float32x4_t a) { in test_vceqzq_f32()
359 uint32x4_t test_vcgezq_f32(float32x4_t a) { in test_vcgezq_f32()
475 uint32x4_t test_vclezq_f32(float32x4_t a) { in test_vclezq_f32()
591 uint32x4_t test_vcgtzq_f32(float32x4_t a) { in test_vcgtzq_f32()
707 uint32x4_t test_vcltzq_f32(float32x4_t a) { in test_vcltzq_f32()
969 float32x4_t test_vrev64q_f32(float32x4_t a) { in test_vrev64q_f32()
1406 float32x4_t test_vnegq_f32(float32x4_t a) { in test_vnegq_f32()
1490 float32x4_t test_vabsq_f32(float32x4_t a) { in test_vabsq_f32()
2362 float16x4_t test_vcvt_f16_f32(float32x4_t a) { in test_vcvt_f16_f32()
2374 float16x8_t test_vcvt_high_f16_f32(float16x4_t a, float32x4_t b) { in test_vcvt_high_f16_f32()
[all …]
|
| D | arm-neon-vcvtX.c | 28 int32x4_t test_vcvtaq_s32_f32(float32x4_t a) { in test_vcvtaq_s32_f32()
37 uint32x4_t test_vcvtaq_u32_f32(float32x4_t a) { in test_vcvtaq_u32_f32()
64 int32x4_t test_vcvtnq_s32_f32(float32x4_t a) { in test_vcvtnq_s32_f32()
73 uint32x4_t test_vcvtnq_u32_f32(float32x4_t a) { in test_vcvtnq_u32_f32()
100 int32x4_t test_vcvtpq_s32_f32(float32x4_t a) { in test_vcvtpq_s32_f32()
109 uint32x4_t test_vcvtpq_u32_f32(float32x4_t a) { in test_vcvtpq_u32_f32()
136 int32x4_t test_vcvtmq_s32_f32(float32x4_t a) { in test_vcvtmq_s32_f32()
145 uint32x4_t test_vcvtmq_u32_f32(float32x4_t a) { in test_vcvtmq_u32_f32()
|
| D | arm-neon-numeric-maxmin.c | 27 float32x4_t test_vmaxnmq_f32(float32x4_t a, float32x4_t b) { in test_vmaxnmq_f32()
53 float32x4_t test_vminnmq_f32(float32x4_t a, float32x4_t b) { in test_vminnmq_f32()
|
| D | arm-arguments.c | 197 typedef __attribute__((neon_vector_type(4))) float float32x4_t; typedef
198 static __attribute__((__always_inline__, __nodebug__)) float32x4_t vaddq_f32( in vaddq_f32()
199 float32x4_t __a, float32x4_t __b) { in vaddq_f32()
202 float32x4_t f35(int i, s35_with_align s1, s35_with_align s2) { in f35()
203 float32x4_t v = vaddq_f32(*(float32x4_t *)&s1, in f35()
204 *(float32x4_t *)&s2); in f35()
|
| D | aarch64-neon-2velem.c | 303 float32x4_t test_vfmaq_lane_f32(float32x4_t a, float32x4_t b, float32x2_t v) { in test_vfmaq_lane_f32()
317 float32x2_t test_vfma_laneq_f32(float32x2_t a, float32x2_t b, float32x4_t v) { in test_vfma_laneq_f32()
331 float32x4_t test_vfmaq_laneq_f32(float32x4_t a, float32x4_t b, float32x4_t v) { in test_vfmaq_laneq_f32()
361 float32x4_t test_vfmsq_lane_f32(float32x4_t a, float32x4_t b, float32x2_t v) { in test_vfmsq_lane_f32()
376 float32x2_t test_vfms_laneq_f32(float32x2_t a, float32x2_t b, float32x4_t v) { in test_vfms_laneq_f32()
391 float32x4_t test_vfmsq_laneq_f32(float32x4_t a, float32x4_t b, float32x4_t v) { in test_vfmsq_laneq_f32()
459 float32_t test_vfmas_laneq_f32(float32_t a, float32_t b, float32x4_t v) { in test_vfmas_laneq_f32()
481 float32_t test_vfmss_laneq_f32(float32_t a, float32_t b, float32x4_t v) { in test_vfmss_laneq_f32()
1507 float32x4_t test_vmulq_lane_f32(float32x4_t a, float32x2_t v) { in test_vmulq_lane_f32()
1523 float32x2_t test_vmul_laneq_f32(float32x2_t a, float32x4_t v) { in test_vmul_laneq_f32()
[all …]
|
| D | arm64-vrnd.c | 7 int32x4_t rnd3(float32x4_t a) { return vrndq_f32(a); } in rnd3()
15 int32x4_t rnd8(float32x4_t a) { return vrndnq_f32(a); } in rnd8()
24 int32x4_t rnd12(float32x4_t a) { return vrndmq_f32(a); } in rnd12()
33 int32x4_t rnd16(float32x4_t a) { return vrndpq_f32(a); } in rnd16()
40 int32x4_t rnd20(float32x4_t a) { return vrndaq_f32(a); } in rnd20()
47 int32x4_t rnd24(float32x4_t a) { return vrndxq_f32(a); } in rnd24()
|
| D | arm-neon-fma.c | 32 float32x4_t test_fmaq_order(float32x4_t accum, float32x4_t lhs, float32x4_t rhs) { in test_fmaq_order()
|
| D | aarch64-neon-perm.c | 115 float32x4_t test_vuzp1q_f32(float32x4_t a, float32x4_t b) { in test_vuzp1q_f32()
262 float32x4_t test_vuzp2q_f32(float32x4_t a, float32x4_t b) { in test_vuzp2q_f32()
409 float32x4_t test_vzip1q_f32(float32x4_t a, float32x4_t b) { in test_vzip1q_f32()
556 float32x4_t test_vzip2q_f32(float32x4_t a, float32x4_t b) { in test_vzip2q_f32()
703 float32x4_t test_vtrn1q_f32(float32x4_t a, float32x4_t b) { in test_vtrn1q_f32()
850 float32x4_t test_vtrn2q_f32(float32x4_t a, float32x4_t b) { in test_vtrn2q_f32()
1299 float32x4x2_t test_vuzpq_f32(float32x4_t a, float32x4_t b) { in test_vuzpq_f32()
1763 float32x4x2_t test_vzipq_f32(float32x4_t a, float32x4_t b) { in test_vzipq_f32()
2227 float32x4x2_t test_vtrnq_f32(float32x4_t a, float32x4_t b) { in test_vtrnq_f32()
|
| D | arm_neon_intrinsics.c | 395 float32x4_t test_vabdq_f32(float32x4_t a, float32x4_t b) { in test_vabdq_f32()
545 float32x4_t test_vabsq_f32(float32x4_t a) { in test_vabsq_f32()
644 float32x4_t test_vaddq_f32(float32x4_t a, float32x4_t b) { in test_vaddq_f32()
1322 float32x4_t test_vbslq_f32(uint32x4_t a, float32x4_t b, float32x4_t c) { in test_vbslq_f32()
1363 uint32x4_t test_vcageq_f32(float32x4_t a, float32x4_t b) { in test_vcageq_f32()
1386 uint32x4_t test_vcagtq_f32(float32x4_t a, float32x4_t b) { in test_vcagtq_f32()
1409 uint32x4_t test_vcaleq_f32(float32x4_t a, float32x4_t b) { in test_vcaleq_f32()
1432 uint32x4_t test_vcaltq_f32(float32x4_t a, float32x4_t b) { in test_vcaltq_f32()
1529 uint32x4_t test_vceqq_f32(float32x4_t a, float32x4_t b) { in test_vceqq_f32()
1650 uint32x4_t test_vcgeq_f32(float32x4_t a, float32x4_t b) { in test_vcgeq_f32()
[all …]
|
| /external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
| D | depthwiseconv_float.h | 38 float32x4_t filter[2];
46 float32x4_t input[4];
52 float32x4_t acc[4];
70 float32x4_t input[2];
76 float32x4_t acc[2];
99 const float32x4_t filters_dup2 = vcombine_f32(filters, filters);
104 float32x4_t input[4];
110 float32x4_t acc[4];
127 float32x4_t input[2];
133 float32x4_t acc[2];
[all …]
|
| D | neon_tensor_utils.cc | 114 float32x4_t acc_32x4 = vmovq_n_f32(0.0); in NeonMatrixBatchVectorMultiplyAccumulate()
117 float32x4_t vector_f32x4 = vld1q_f32(vector_in_batch + c); in NeonMatrixBatchVectorMultiplyAccumulate()
118 float32x4_t matrix_f32x4 = vld1q_f32(matrix_row + c); in NeonMatrixBatchVectorMultiplyAccumulate()
525 float32x4_t acc_32x4 = vmovq_n_f32(0.0); in NeonSparseMatrixBatchVectorMultiplyAccumulate()
535 float32x4_t vector_f32x4 = vld1q_f32(vector_block_in_batch_ptr + in NeonSparseMatrixBatchVectorMultiplyAccumulate()
537 float32x4_t matrix_f32x4 = in NeonSparseMatrixBatchVectorMultiplyAccumulate()
640 float32x4_t v1_f32x4 = vld1q_f32(vector1 + v); in NeonVectorVectorCwiseProduct()
641 float32x4_t v2_f32x4 = vld1q_f32(vector2 + v); in NeonVectorVectorCwiseProduct()
643 float32x4_t mul_32x4 = vmulq_f32(v1_f32x4, v2_f32x4); in NeonVectorVectorCwiseProduct()
662 float32x4_t v1_f32x4 = vld1q_f32(vector1 + v); in NeonVectorVectorCwiseProductAccumulate()
[all …]
|
| /external/clang/test/SemaCXX/ |
| D | neon-vector-types.cpp | 7 typedef __attribute__((neon_vector_type(4))) float32_t float32x4_t; typedef
12 extern int bar (float32x4_t const *p);
15 float32x4_t const *vRealPtr = (VFLOAT *)&realBufPtr[0]; in foo()
24 float32x4_t const *vRealPtr = (VFLOAT *)&realBufPtr[0]; in autoCorrelation2nd_Neon()
30 typedef __attribute__((neon_vector_type(4))) float32_t float32x4_t; typedef
35 extern float32x4_t vec; in test()
37 float32x4_t __a = (vec); in test()
|
| /external/tensorflow/tensorflow/core/kernels/neon/ |
| D | depthwiseconv_float.h | 42 float32x4_t filter[2];
50 float32x4_t input[4];
56 float32x4_t acc[4];
74 float32x4_t input[2];
80 float32x4_t acc[2];
103 const float32x4_t filters_dup2 = vcombine_f32(filters, filters);
108 float32x4_t input[4];
114 float32x4_t acc[4];
131 float32x4_t input[2];
137 float32x4_t acc[2];
[all …]
|
| /external/tensorflow/tensorflow/core/kernels/ |
| D | quantized_instance_norm.cc | 52 float32x4_t xA[4] = {0.0f}; in ColMeanAndVariance()
53 float32x4_t M2A[4] = {0.0f}; in ColMeanAndVariance()
95 const float32x4_t xB = in ColMeanAndVariance()
99 const float32x4_t delta = vsubq_f32(xB, xA[i]); in ColMeanAndVariance()
104 const float32x4_t sub_sum_f32 = vcvtq_f32_u32(sub_sum[i]); in ColMeanAndVariance()
105 const float32x4_t sub_sum_sq = vmulq_f32(sub_sum_f32, sub_sum_f32); in ColMeanAndVariance()
108 const float32x4_t M2B = vsubq_f32(vcvtq_f32_u32(sub_sq_sum[i]), in ColMeanAndVariance()
110 const float32x4_t last_term = in ColMeanAndVariance()
143 const float32x4_t eps = vdupq_n_f32(variance_epsilon); in MinAndMax()
146 const float32x4_t mean[4] = {vld1q_f32(mean_ptr + col_offset), in MinAndMax()
[all …]
|
| /external/clang/test/Sema/ |
| D | arm64-neon-args.c | 7 void vcopy_reject(float32x4_t vOut0, float32x4_t vAlpha, int t) { in vcopy_reject()
12 float32x4_t test_vmlsq_lane(float32x4_t accum, float32x4_t lhs, float32x2_t rhs) { in test_vmlsq_lane()
|
| /external/tensorflow/tensorflow/examples/android/jni/object_tracking/ |
| D | utils_neon.cc | 30 inline static float GetSum(const float32x4_t& values) { in GetSum()
44 float32x4_t accum = vdupq_n_f32(0.0f); in ComputeMeanNeon()
77 const float32x4_t mean_vec = vdupq_n_f32(-mean); in ComputeStdDevNeon()
79 float32x4_t accum = vdupq_n_f32(0.0f); in ComputeStdDevNeon()
83 const float32x4_t deltas = in ComputeStdDevNeon()
118 float32x4_t accum = vdupq_n_f32(0.0f); in ComputeCrossCorrelationNeon()
|
| D | image_neon.cc | 207 float32x4_t xx = vdupq_n_f32(0.0f); in CalculateGNeon()
208 float32x4_t xy = vdupq_n_f32(0.0f); in CalculateGNeon()
209 float32x4_t yy = vdupq_n_f32(0.0f); in CalculateGNeon()
225 float32x4_t x = vld1q_f32(arm_vals_x + i); in CalculateGNeon()
231 float32x4_t y = vld1q_f32(arm_vals_y + i); in CalculateGNeon()
|
| /external/adhd/cras/src/dsp/ |
| D | crossover2.c | 35 float32x4_t x1 = {lp->x1L, hp->x1L, lp->x1R, hp->x1R}; in lr42_split()
36 float32x4_t x2 = {lp->x2L, hp->x2L, lp->x2R, hp->x2R}; in lr42_split()
37 float32x4_t y1 = {lp->y1L, hp->y1L, lp->y1R, hp->y1R}; in lr42_split()
38 float32x4_t y2 = {lp->y2L, hp->y2L, lp->y2R, hp->y2R}; in lr42_split()
39 float32x4_t z1 = {lp->z1L, hp->z1L, lp->z1R, hp->z1R}; in lr42_split()
40 float32x4_t z2 = {lp->z2L, hp->z2L, lp->z2R, hp->z2R}; in lr42_split()
41 float32x4_t b0 = {lp->b0, hp->b0, lp->b0, hp->b0}; in lr42_split()
42 float32x4_t b1 = {lp->b1, hp->b1, lp->b1, hp->b1}; in lr42_split()
43 float32x4_t b2 = {lp->b2, hp->b2, lp->b2, hp->b2}; in lr42_split()
44 float32x4_t a1 = {lp->a1, hp->a1, lp->a1, hp->a1}; in lr42_split()
[all …]
|
| /external/libopus/celt/arm/ |
| D | pitch_neon_intr.c | 199 float32x4_t xy_f32x4 = vdupq_n_f32(0); in celt_inner_prod_neon()
203 float32x4_t x_f32x4, y_f32x4; in celt_inner_prod_neon()
213 const float32x4_t x_f32x4 = vld1q_f32(&x[i]); in celt_inner_prod_neon()
214 const float32x4_t y_f32x4 = vld1q_f32(&y[i]); in celt_inner_prod_neon()
239 float32x4_t xy01_f32x4 = vdupq_n_f32(0); in dual_inner_prod_neon()
240 float32x4_t xy02_f32x4 = vdupq_n_f32(0); in dual_inner_prod_neon()
244 float32x4_t x_f32x4, y01_f32x4, y02_f32x4; in dual_inner_prod_neon()
258 const float32x4_t x_f32x4 = vld1q_f32(&x[i]); in dual_inner_prod_neon()
259 const float32x4_t y01_f32x4 = vld1q_f32(&y01[i]); in dual_inner_prod_neon()
260 const float32x4_t y02_f32x4 = vld1q_f32(&y02[i]); in dual_inner_prod_neon()
|
| /external/clang/test/CodeGenCXX/ |
| D | static-init-4.cpp | 3 typedef __attribute__((vector_size(4*4))) float float32x4_t; typedef
4 union QDSUnion { float32x4_t q; float s[4]; };
5 constexpr float32x4_t a = {1,2,3,4};
|
| /external/clang/test/SemaOpenCL/ |
| D | init.cl | 10 } float32x4_t;
13 float32x4_t lo;
14 float32x4_t hi;
|