Lines Matching +full:- +full:d
1 /* ----------------------------------------------------------------------
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
12 * Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved.
14 * SPDX-License-Identifier: Apache-2.0
20 * www.apache.org/licenses/LICENSE-2.0
101 { ( 0 - 16) * (int32_t)sizeof(float16_t *) in _arm_radix4_butterfly_f16_mve()
102 , ( 4 - 16) * (int32_t)sizeof(float16_t *) in _arm_radix4_butterfly_f16_mve()
103 , ( 8 - 16) * (int32_t)sizeof(float16_t *) in _arm_radix4_butterfly_f16_mve()
104 , (12 - 16) * (int32_t)sizeof(float16_t *)}; in _arm_radix4_butterfly_f16_mve()
112 &S->rearranged_twiddle_stride1[ in _arm_radix4_butterfly_f16_mve()
113 S->rearranged_twiddle_tab_stride1_arr[stage]]; in _arm_radix4_butterfly_f16_mve()
115 &S->rearranged_twiddle_stride2[ in _arm_radix4_butterfly_f16_mve()
116 S->rearranged_twiddle_tab_stride2_arr[stage]]; in _arm_radix4_butterfly_f16_mve()
118 &S->rearranged_twiddle_stride3[ in _arm_radix4_butterfly_f16_mve()
119 S->rearranged_twiddle_tab_stride3_arr[stage]]; in _arm_radix4_butterfly_f16_mve()
144 vecDiff0 = vecA - vecC; /* vecSum0 = vsubq(vecA, vecC) */ in _arm_radix4_butterfly_f16_mve()
147 vecDiff1 = vecB - vecD; in _arm_radix4_butterfly_f16_mve()
149 * [ 1 1 1 1 ] * [ A B C D ]' .* 1 in _arm_radix4_butterfly_f16_mve()
156 * [ 1 -1 1 -1 ] * [ A B C D ]' in _arm_radix4_butterfly_f16_mve()
158 vecTmp0 = vecSum0 - vecSum1; in _arm_radix4_butterfly_f16_mve()
160 * [ 1 -1 1 -1 ] * [ A B C D ]'.* W2 in _arm_radix4_butterfly_f16_mve()
169 * [ 1 -i -1 +i ] * [ A B C D ]' in _arm_radix4_butterfly_f16_mve()
173 * [ 1 -i -1 +i ] * [ A B C D ]'.* W1 in _arm_radix4_butterfly_f16_mve()
182 * [ 1 +i -1 -i ] * [ A B C D ]' in _arm_radix4_butterfly_f16_mve()
186 * [ 1 +i -1 -i ] * [ A B C D ]'.* W3 in _arm_radix4_butterfly_f16_mve()
197 blkCnt--; in _arm_radix4_butterfly_f16_mve()
221 vecDiff0 = vecA - vecC; /* vecSum0 = vsubq(vecA, vecC) */ in _arm_radix4_butterfly_f16_mve()
227 vecDiff1 = vecB - vecD; in _arm_radix4_butterfly_f16_mve()
229 /* pre-load for next iteration */ in _arm_radix4_butterfly_f16_mve()
234 vstrwq_scatter_base_f32(vecScGathAddr, -64, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_f16_mve()
236 vecTmp0 = vecSum0 - vecSum1; in _arm_radix4_butterfly_f16_mve()
237 vstrwq_scatter_base_f32(vecScGathAddr, -64 + 4, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_f16_mve()
240 vstrwq_scatter_base_f32(vecScGathAddr, -64 + 8, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_f16_mve()
243 vstrwq_scatter_base_f32(vecScGathAddr, -64 + 12, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_f16_mve()
245 blkCnt--; in _arm_radix4_butterfly_f16_mve()
256 float16_t const *pCoef = S->pTwiddle; in arm_cfft_radix4by2_f16_mve()
287 blkCnt--; in arm_cfft_radix4by2_f16_mve()
307 ( 0 - 16) * (int32_t)sizeof(q31_t *), in _arm_radix4_butterfly_inverse_f16_mve()
308 ( 4 - 16) * (int32_t)sizeof(q31_t *), in _arm_radix4_butterfly_inverse_f16_mve()
309 ( 8 - 16) * (int32_t)sizeof(q31_t *), in _arm_radix4_butterfly_inverse_f16_mve()
310 (12 - 16) * (int32_t)sizeof(q31_t *) in _arm_radix4_butterfly_inverse_f16_mve()
319 &S->rearranged_twiddle_stride1[ in _arm_radix4_butterfly_inverse_f16_mve()
320 S->rearranged_twiddle_tab_stride1_arr[stage]]; in _arm_radix4_butterfly_inverse_f16_mve()
322 &S->rearranged_twiddle_stride2[ in _arm_radix4_butterfly_inverse_f16_mve()
323 S->rearranged_twiddle_tab_stride2_arr[stage]]; in _arm_radix4_butterfly_inverse_f16_mve()
325 &S->rearranged_twiddle_stride3[ in _arm_radix4_butterfly_inverse_f16_mve()
326 S->rearranged_twiddle_tab_stride3_arr[stage]]; in _arm_radix4_butterfly_inverse_f16_mve()
352 vecDiff0 = vecA - vecC; /* vecSum0 = vsubq(vecA, vecC) */ in _arm_radix4_butterfly_inverse_f16_mve()
355 vecDiff1 = vecB - vecD; in _arm_radix4_butterfly_inverse_f16_mve()
357 * [ 1 1 1 1 ] * [ A B C D ]' .* 1 in _arm_radix4_butterfly_inverse_f16_mve()
363 * [ 1 -1 1 -1 ] * [ A B C D ]' in _arm_radix4_butterfly_inverse_f16_mve()
365 vecTmp0 = vecSum0 - vecSum1; in _arm_radix4_butterfly_inverse_f16_mve()
367 * [ 1 -1 1 -1 ] * [ A B C D ]'.* W1 in _arm_radix4_butterfly_inverse_f16_mve()
376 * [ 1 -i -1 +i ] * [ A B C D ]' in _arm_radix4_butterfly_inverse_f16_mve()
380 * [ 1 -i -1 +i ] * [ A B C D ]'.* W2 in _arm_radix4_butterfly_inverse_f16_mve()
389 * [ 1 +i -1 -i ] * [ A B C D ]' in _arm_radix4_butterfly_inverse_f16_mve()
393 * [ 1 +i -1 -i ] * [ A B C D ]'.* W3 in _arm_radix4_butterfly_inverse_f16_mve()
404 blkCnt--; in _arm_radix4_butterfly_inverse_f16_mve()
430 vecDiff0 = vecA - vecC; /* vecSum0 = vsubq(vecA, vecC) */ in _arm_radix4_butterfly_inverse_f16_mve()
436 vecDiff1 = vecB - vecD; in _arm_radix4_butterfly_inverse_f16_mve()
443 vstrwq_scatter_base_f32(vecScGathAddr, -64, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_inverse_f16_mve()
445 vecTmp0 = vecSum0 - vecSum1; in _arm_radix4_butterfly_inverse_f16_mve()
447 vstrwq_scatter_base_f32(vecScGathAddr, -64 + 4, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_inverse_f16_mve()
451 vstrwq_scatter_base_f32(vecScGathAddr, -64 + 8, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_inverse_f16_mve()
455 vstrwq_scatter_base_f32(vecScGathAddr, -64 + 12, (f32x4_t)vecTmp0); in _arm_radix4_butterfly_inverse_f16_mve()
457 blkCnt--; in _arm_radix4_butterfly_inverse_f16_mve()
468 float16_t const *pCoef = S->pTwiddle; in arm_cfft_radix4by2_inverse_f16_mve()
500 blkCnt--; in arm_cfft_radix4by2_inverse_f16_mve()
515 @brief Processing function for the floating-point complex FFT.
516 @param[in] S points to an instance of the floating-point CFFT structure
517 … points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place
519 - value = 0: forward transform
520 - value = 1: inverse transform
522 - value = 0: disables bit reversal of output
523 - value = 1: enables bit reversal of output
533 uint32_t fftLen = S->fftLen; in arm_cfft_f16()
543 … _arm_radix4_butterfly_inverse_f16_mve(S, pSrc, fftLen, arm_inverse_fft_length_f16(S->fftLen)); in arm_cfft_f16()
576 arm_bitreversal_16_inpl_mve((uint16_t*)pSrc, S->bitRevLength, S->pBitRevTable); in arm_cfft_f16()
608 @brief Processing function for the floating-point complex FFT.
609 @param[in] S points to an instance of the floating-point CFFT structure
610 … points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place
612 - value = 0: forward transform
613 - value = 1: inverse transform
615 - value = 0: disables bit reversal of output
616 - value = 1: enables bit reversal of output
625 uint32_t L = S->fftLen, l; in arm_cfft_f16()
634 *pSrc = -(_Float16)*pSrc; in arm_cfft_f16()
647 arm_radix4_butterfly_f16 (p1, L, (float16_t*)S->pTwiddle, 1U); in arm_cfft_f16()
654 arm_cfft_radix4by2_f16 ( p1, L, (float16_t*)S->pTwiddle); in arm_cfft_f16()
660 arm_bitreversal_16((uint16_t*)p1, S->bitRevLength,(uint16_t*)S->pBitRevTable); in arm_cfft_f16()
670 *pSrc = -(_Float16)(*pSrc) * (_Float16)invL; in arm_cfft_f16()