/third_party/cmsis/CMSIS/DSP/Include/ |
D | arm_helium_utils.h | 112 float16x8_t vecIn) in __mve_cmplx_sum_intra_vec_f16() argument 117 vecTmp = (float16x8_t) vrev64q_s32((int32x4_t) vecIn); in __mve_cmplx_sum_intra_vec_f16() 125 vecTmp = vaddq_f16(vecTmp, vecIn); in __mve_cmplx_sum_intra_vec_f16() 153 float16x8_t vecIn, in mve_cmplx_sum_intra_vec_f16() argument 156 float16x8_t vecOut = __mve_cmplx_sum_intra_vec_f16(vecIn); in mve_cmplx_sum_intra_vec_f16() 202 uint32x4_t vecIn = vldrwq_u32((uint32_t const *)pDataSrc); in arm_mat_trans_32bit_2x2_mve() local 203 vstrwq_scatter_shifted_offset_u32(pDataDest, vecOffs, vecIn); in arm_mat_trans_32bit_2x2_mve() 244 uint32x4x4_t vecIn; in arm_mat_trans_32bit_4x4_mve() local 246 vecIn = vld4q((uint32_t const *) pDataSrc); in arm_mat_trans_32bit_4x4_mve() 247 vstrwq(pDataDest, vecIn.val[0]); in arm_mat_trans_32bit_4x4_mve() [all …]
|
D | arm_vec_math_f16.h | 158 __STATIC_INLINE float16x8_t vlogq_f16(float16x8_t vecIn) in vlogq_f16() argument 168 VMANT_EXP_F16(vecIn); in vlogq_f16() 214 vecAcc0 = vdupq_m(vecAcc0, -F16INFINITY, vcmpeqq(vecIn, 0.0f)); in vlogq_f16() 255 __STATIC_INLINE f16x8_t vrecip_f16(f16x8_t vecIn) in vrecip_f16() argument 260 vecSx = vabsq(vecIn); in vrecip_f16() 262 v.f = vecIn; in vrecip_f16() 277 v.f = vdupq_m(v.f, F16INFINITY, vcmpeqq(vecIn, 0.0f)); in vrecip_f16() 281 v.f = vnegq_m(v.f, v.f, vcmpltq(vecIn, 0.0f)); in vrecip_f16()
|
D | arm_vec_math.h | 160 __STATIC_INLINE f32x4_t vlogq_f32(f32x4_t vecIn) in vlogq_f32() argument 170 vecTmpFlt1 = vmant_exp_f32(vecIn, &vecExpUnBiased); in vlogq_f32() 216 vecAcc0 = vdupq_m(vecAcc0, -INFINITY, vcmpeqq(vecIn, 0.0f)); in vlogq_f32() 248 __STATIC_INLINE f32x4_t vrecip_f32(f32x4_t vecIn) in vrecip_f32() argument 253 vecSx = vabsq(vecIn); in vrecip_f32() 255 v.f = vecIn; in vrecip_f32() 270 v.f = vdupq_m(v.f, INFINITY, vcmpeqq(vecIn, 0.0f)); in vrecip_f32() 274 v.f = vnegq_m(v.f, v.f, vcmpltq(vecIn, 0.0f)); in vrecip_f32()
|
/third_party/cmsis/CMSIS/DSP/Source/SVMFunctions/ |
D | arm_svm_linear_predict_f16.c | 80 f16x8_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_linear_predict_f16() local 111 vecIn = vld1q(pInVec); in arm_svm_linear_predict_f16() 115 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_linear_predict_f16() 118 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_linear_predict_f16() 121 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_linear_predict_f16() 124 acc3 = vfmaq(acc3, vecIn, vecA); in arm_svm_linear_predict_f16() 137 vecIn = vldrhq_z_f16(pInVec, p0); in arm_svm_linear_predict_f16() 139 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_linear_predict_f16() 141 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_linear_predict_f16() 143 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_linear_predict_f16() [all …]
|
D | arm_svm_sigmoid_predict_f16.c | 83 f16x8_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_sigmoid_predict_f16() local 114 vecIn = vld1q(pInVec); in arm_svm_sigmoid_predict_f16() 118 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_sigmoid_predict_f16() 121 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_sigmoid_predict_f16() 124 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_sigmoid_predict_f16() 127 acc3 = vfmaq(acc3, vecIn, vecA); in arm_svm_sigmoid_predict_f16() 140 vecIn = vldrhq_z_f16(pInVec, p0); in arm_svm_sigmoid_predict_f16() 142 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_sigmoid_predict_f16() 144 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_sigmoid_predict_f16() 146 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_sigmoid_predict_f16() [all …]
|
D | arm_svm_polynomial_predict_f16.c | 83 f16x8_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_polynomial_predict_f16() local 114 vecIn = vld1q(pInVec); in arm_svm_polynomial_predict_f16() 118 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_polynomial_predict_f16() 121 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_polynomial_predict_f16() 124 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_polynomial_predict_f16() 127 acc3 = vfmaq(acc3, vecIn, vecA); in arm_svm_polynomial_predict_f16() 140 vecIn = vldrhq_z_f16(pInVec, p0); in arm_svm_polynomial_predict_f16() 142 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_polynomial_predict_f16() 144 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_polynomial_predict_f16() 146 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_polynomial_predict_f16() [all …]
|
D | arm_svm_rbf_predict_f16.c | 83 f16x8_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_rbf_predict_f16() local 115 vecIn = vld1q(pInVec); in arm_svm_rbf_predict_f16() 119 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() 123 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() 127 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() 131 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() 146 vecIn = vldrhq_z_f16(pInVec, p0); in arm_svm_rbf_predict_f16() 148 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() 151 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() 154 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f16() [all …]
|
D | arm_svm_linear_predict_f32.c | 77 f32x4_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_linear_predict_f32() local 108 vecIn = vld1q(pInVec); in arm_svm_linear_predict_f32() 112 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_linear_predict_f32() 115 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_linear_predict_f32() 118 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_linear_predict_f32() 121 acc3 = vfmaq(acc3, vecIn, vecA); in arm_svm_linear_predict_f32() 134 vecIn = vldrwq_z_f32(pInVec, p0); in arm_svm_linear_predict_f32() 136 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_linear_predict_f32() 138 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_linear_predict_f32() 140 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_linear_predict_f32() [all …]
|
D | arm_svm_polynomial_predict_f32.c | 83 f32x4_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_polynomial_predict_f32() local 114 vecIn = vld1q(pInVec); in arm_svm_polynomial_predict_f32() 118 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_polynomial_predict_f32() 121 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_polynomial_predict_f32() 124 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_polynomial_predict_f32() 127 acc3 = vfmaq(acc3, vecIn, vecA); in arm_svm_polynomial_predict_f32() 140 vecIn = vldrwq_z_f32(pInVec, p0); in arm_svm_polynomial_predict_f32() 142 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_polynomial_predict_f32() 144 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_polynomial_predict_f32() 146 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_polynomial_predict_f32() [all …]
|
D | arm_svm_sigmoid_predict_f32.c | 80 f32x4_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_sigmoid_predict_f32() local 111 vecIn = vld1q(pInVec); in arm_svm_sigmoid_predict_f32() 115 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_sigmoid_predict_f32() 118 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_sigmoid_predict_f32() 121 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_sigmoid_predict_f32() 124 acc3 = vfmaq(acc3, vecIn, vecA); in arm_svm_sigmoid_predict_f32() 137 vecIn = vldrwq_z_f32(pInVec, p0); in arm_svm_sigmoid_predict_f32() 139 acc0 = vfmaq(acc0, vecIn, vecA); in arm_svm_sigmoid_predict_f32() 141 acc1 = vfmaq(acc1, vecIn, vecA); in arm_svm_sigmoid_predict_f32() 143 acc2 = vfmaq(acc2, vecIn, vecA); in arm_svm_sigmoid_predict_f32() [all …]
|
D | arm_svm_rbf_predict_f32.c | 80 f32x4_t vecIn, acc0, acc1, acc2, acc3; in arm_svm_rbf_predict_f32() local 112 vecIn = vld1q(pInVec); in arm_svm_rbf_predict_f32() 116 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() 120 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() 124 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() 128 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() 143 vecIn = vldrwq_z_f32(pInVec, p0); in arm_svm_rbf_predict_f32() 145 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() 148 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() 151 vecDif = vsubq(vecIn, vecA); in arm_svm_rbf_predict_f32() [all …]
|
/third_party/cmsis/CMSIS/DSP/Source/MatrixFunctions/ |
D | arm_mat_vec_mult_f16.c | 79 f16x8_t vecIn, acc0, acc1, acc2, acc3; in arm_mat_vec_mult_f16() local 111 vecIn = vld1q(pInVec); in arm_mat_vec_mult_f16() 115 acc0 = vfmaq(acc0, vecIn, vecA); in arm_mat_vec_mult_f16() 118 acc1 = vfmaq(acc1, vecIn, vecA); in arm_mat_vec_mult_f16() 121 acc2 = vfmaq(acc2, vecIn, vecA); in arm_mat_vec_mult_f16() 124 acc3 = vfmaq(acc3, vecIn, vecA); in arm_mat_vec_mult_f16() 138 vecIn = vldrhq_z_f16(pInVec, p0); in arm_mat_vec_mult_f16() 140 acc0 = vfmaq(acc0, vecIn, vecA); in arm_mat_vec_mult_f16() 142 acc1 = vfmaq(acc1, vecIn, vecA); in arm_mat_vec_mult_f16() 144 acc2 = vfmaq(acc2, vecIn, vecA); in arm_mat_vec_mult_f16() [all …]
|
D | arm_mat_vec_mult_f32.c | 83 f32x4_t vecIn, acc0, acc1, acc2, acc3; in arm_mat_vec_mult_f32() local 115 vecIn = vld1q(pInVec); in arm_mat_vec_mult_f32() 119 acc0 = vfmaq(acc0, vecIn, vecA); in arm_mat_vec_mult_f32() 122 acc1 = vfmaq(acc1, vecIn, vecA); in arm_mat_vec_mult_f32() 125 acc2 = vfmaq(acc2, vecIn, vecA); in arm_mat_vec_mult_f32() 128 acc3 = vfmaq(acc3, vecIn, vecA); in arm_mat_vec_mult_f32() 142 vecIn = vldrwq_z_f32(pInVec, p0); in arm_mat_vec_mult_f32() 144 acc0 = vfmaq(acc0, vecIn, vecA); in arm_mat_vec_mult_f32() 146 acc1 = vfmaq(acc1, vecIn, vecA); in arm_mat_vec_mult_f32() 148 acc2 = vfmaq(acc2, vecIn, vecA); in arm_mat_vec_mult_f32() [all …]
|
D | arm_mat_vec_mult_q31.c | 75 q31x4_t vecMatA0, vecMatA1, vecMatA2, vecIn; in arm_mat_vec_mult_q31() local 104 vecIn = vld1q(pVec); in arm_mat_vec_mult_q31() 107 acc0 = vmlaldavaq(acc0, vecIn, vecMatA0); in arm_mat_vec_mult_q31() 108 acc1 = vmlaldavaq(acc1, vecIn, vecMatA1); in arm_mat_vec_mult_q31() 109 acc2 = vmlaldavaq(acc2, vecIn, vecMatA2); in arm_mat_vec_mult_q31() 125 vecIn = vldrwq_z_s32(pVec, p0); in arm_mat_vec_mult_q31() 127 acc0 = vmlaldavaq(acc0, vecIn, vecMatA0); in arm_mat_vec_mult_q31() 128 acc1 = vmlaldavaq(acc1, vecIn, vecMatA1); in arm_mat_vec_mult_q31() 129 acc2 = vmlaldavaq(acc2, vecIn, vecMatA2); in arm_mat_vec_mult_q31() 151 q31x4_t vecMatA0, vecMatA1, vecIn; in arm_mat_vec_mult_q31() local [all …]
|
D | arm_mat_vec_mult_q15.c | 78 q15x8_t vecMatA0, vecMatA1, vecMatA2, vecIn; in arm_mat_vec_mult_q15() local 107 vecIn = vld1q(pVec); in arm_mat_vec_mult_q15() 110 acc0 = vmlaldavaq(acc0, vecIn, vecMatA0); in arm_mat_vec_mult_q15() 111 acc1 = vmlaldavaq(acc1, vecIn, vecMatA1); in arm_mat_vec_mult_q15() 112 acc2 = vmlaldavaq(acc2, vecIn, vecMatA2); in arm_mat_vec_mult_q15() 128 vecIn = vldrhq_z_s16(pVec, p0); in arm_mat_vec_mult_q15() 130 acc0 = vmlaldavaq(acc0, vecIn, vecMatA0); in arm_mat_vec_mult_q15() 131 acc1 = vmlaldavaq(acc1, vecIn, vecMatA1); in arm_mat_vec_mult_q15() 132 acc2 = vmlaldavaq(acc2, vecIn, vecMatA2); in arm_mat_vec_mult_q15() 154 q15x8_t vecMatA0, vecMatA1, vecIn; in arm_mat_vec_mult_q15() local [all …]
|
D | arm_mat_vec_mult_q7.c | 78 q7x16_t vecMatA0, vecMatA1, vecMatA2, vecMatA3, vecIn; in arm_mat_vec_mult_q7() local 112 vecIn = vld1q(pVec); in arm_mat_vec_mult_q7() 115 acc0 = vmladavaq(acc0, vecIn, vecMatA0); in arm_mat_vec_mult_q7() 116 acc1 = vmladavaq(acc1, vecIn, vecMatA1); in arm_mat_vec_mult_q7() 117 acc2 = vmladavaq(acc2, vecIn, vecMatA2); in arm_mat_vec_mult_q7() 118 acc3 = vmladavaq(acc3, vecIn, vecMatA3); in arm_mat_vec_mult_q7() 135 vecIn = vldrbq_z_s8(pVec, p0); in arm_mat_vec_mult_q7() 137 acc0 = vmladavaq(acc0, vecIn, vecMatA0); in arm_mat_vec_mult_q7() 138 acc1 = vmladavaq(acc1, vecIn, vecMatA1); in arm_mat_vec_mult_q7() 139 acc2 = vmladavaq(acc2, vecIn, vecMatA2); in arm_mat_vec_mult_q7() [all …]
|
D | arm_mat_trans_q7.c | 57 uint16x8_t vecIn; in arm_mat_trans_q7() local 84 vecIn = vldrbq_gather_offset_u16(pDataC, vecOffs); in arm_mat_trans_q7() 85 vstrbq_u16(pDataDestR, vecIn); in arm_mat_trans_q7() 102 vecIn = vldrbq_gather_offset_u16(pDataC, vecOffs); in arm_mat_trans_q7() 103 vstrbq_p_u16(pDataDestR, vecIn, p0); in arm_mat_trans_q7()
|
D | arm_mat_scale_f16.c | 75 f16x8_t vecIn, vecOut, vecScale; in arm_mat_scale_f16() local 92 vecIn = vld1q(pInVec); in arm_mat_scale_f16() 95 vecOut = vmulq_f16(vecIn, vecScale); in arm_mat_scale_f16() 111 vecIn = vld1q(pInVec); in arm_mat_scale_f16() 112 vecOut = vecIn * scale; in arm_mat_scale_f16()
|
D | arm_mat_scale_f32.c | 88 f32x4_t vecIn, vecOut; in arm_mat_scale_f32() local 103 vecIn = vld1q(pInVec); in arm_mat_scale_f32() 106 vecOut = vecIn * scale; in arm_mat_scale_f32() 122 vecIn = vld1q(pInVec); in arm_mat_scale_f32() 123 vecOut = vecIn * scale; in arm_mat_scale_f32()
|
D | arm_mat_scale_q31.c | 65 q31x4_t vecIn, vecOut; in arm_mat_scale_q31() local 96 vecIn = vld1q(pInVec); in arm_mat_scale_q31() 99 vecOut = vmulhq(vecIn, vdupq_n_s32(scaleFract)); in arm_mat_scale_q31() 117 vecIn = vld1q(pInVec); in arm_mat_scale_q31() 119 vecOut = vmulhq(vecIn, vdupq_n_s32(scaleFract)); in arm_mat_scale_q31()
|
D | arm_mat_scale_q15.c | 66 q15x8_t vecIn, vecOut; in arm_mat_scale_q15() local 97 vecIn = vld1q(pInVec); pInVec += 8; in arm_mat_scale_q15() 100 vecOut = vmulhq(vecIn, vdupq_n_s16(scaleFract)); in arm_mat_scale_q15() 119 vecIn = vld1q(pInVec); pInVec += 8; in arm_mat_scale_q15() 120 vecOut = vmulhq(vecIn, vdupq_n_s16(scaleFract)); in arm_mat_scale_q15()
|
/third_party/cmsis/CMSIS/DSP/PrivateInclude/ |
D | arm_vec_fft.h | 252 uint64x2_t vecIn; in arm_bitreversal_32_outpl_mve() local 254 vecIn = vldrdq_gather_offset_u64(pSrc, (int64x2_t) bitRevOffs0); in arm_bitreversal_32_outpl_mve() 256 vst1q(pDst32, (uint32x4_t) vecIn); in arm_bitreversal_32_outpl_mve() 260 vecIn = vldrdq_gather_offset_u64(pSrc, (int64x2_t) bitRevOffs1); in arm_bitreversal_32_outpl_mve() 262 vst1q(pDst32, (uint32x4_t) vecIn); in arm_bitreversal_32_outpl_mve() 298 uint32x4_t vecIn; in arm_bitreversal_16_outpl_mve() local 300 vecIn = vldrwq_gather_offset_s32(pSrc, bitRevOffs0); in arm_bitreversal_16_outpl_mve() 302 vst1q(pDst16, (uint16x8_t) vecIn); in arm_bitreversal_16_outpl_mve() 306 vecIn = vldrwq_gather_offset_s32(pSrc, bitRevOffs1); in arm_bitreversal_16_outpl_mve() 308 vst1q(pDst16, (uint16x8_t) vecIn); in arm_bitreversal_16_outpl_mve()
|
/third_party/cmsis/CMSIS/DSP/Source/StatisticsFunctions/ |
D | arm_entropy_f16.c | 81 f16x8_t vecIn = vld1q(pSrcA); in arm_entropy_f16() local 83 vSum = vaddq_f16(vSum, vmulq(vecIn, vlogq_f16(vecIn))); in arm_entropy_f16()
|
D | arm_entropy_f32.c | 68 f32x4_t vecIn = vld1q(pSrcA); in arm_entropy_f32() local 70 vSum = vaddq_f32(vSum, vmulq(vecIn, vlogq_f32(vecIn))); in arm_entropy_f32()
|
/third_party/cmsis/CMSIS/DSP/Source/FilteringFunctions/ |
D | arm_biquad_cascade_df1_32x64_q31.c | 299 q31x4_t vecCoef, vecIn; in arm_biquad_cas_df1_32x64_q31() local 355 vecIn = vld1q(pIn1); in arm_biquad_cas_df1_32x64_q31() 357 Yn1 = vmlaldavaq(Yn1, vecIn, vecCoef); in arm_biquad_cas_df1_32x64_q31() 378 vecIn = vld1q(pIn1); in arm_biquad_cas_df1_32x64_q31() 410 Yn1 = vmlaldavaq(Yn1, vecIn, vecCoef); in arm_biquad_cas_df1_32x64_q31() 423 Xn2 = vecIn[2]; in arm_biquad_cas_df1_32x64_q31() 424 Xn1 = vecIn[3]; in arm_biquad_cas_df1_32x64_q31()
|