Welcome to mirror list, hosted at ThFree Co, Russian Federation.

github.com/Flipper-Zero/STM32CubeWB.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
path: root/Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c
diff options
context:
space:
mode:
Diffstat (limited to 'Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c')
-rw-r--r--Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c459
1 files changed, 223 insertions, 236 deletions
diff --git a/Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c b/Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c
index da76ddcc7..28ad1e184 100644
--- a/Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c
+++ b/Drivers/CMSIS/DSP/Source/TransformFunctions/arm_cfft_q15.c
@@ -3,13 +3,13 @@
* Title: arm_cfft_q15.c
* Description: Combined Radix Decimation in Q15 Frequency CFFT processing function
*
- * $Date: 27. January 2017
- * $Revision: V.1.5.1
+ * $Date: 18. March 2019
+ * $Revision: V1.6.0
*
* Target Processor: Cortex-M cores
* -------------------------------------------------------------------- */
/*
- * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
*
* SPDX-License-Identifier: Apache-2.0
*
@@ -29,317 +29,304 @@
#include "arm_math.h"
extern void arm_radix4_butterfly_q15(
- q15_t * pSrc,
- uint32_t fftLen,
- q15_t * pCoef,
- uint32_t twidCoefModifier);
+ q15_t * pSrc,
+ uint32_t fftLen,
+ const q15_t * pCoef,
+ uint32_t twidCoefModifier);
extern void arm_radix4_butterfly_inverse_q15(
- q15_t * pSrc,
- uint32_t fftLen,
- q15_t * pCoef,
- uint32_t twidCoefModifier);
+ q15_t * pSrc,
+ uint32_t fftLen,
+ const q15_t * pCoef,
+ uint32_t twidCoefModifier);
extern void arm_bitreversal_16(
- uint16_t * pSrc,
- const uint16_t bitRevLen,
- const uint16_t * pBitRevTable);
+ uint16_t * pSrc,
+ const uint16_t bitRevLen,
+ const uint16_t * pBitRevTable);
void arm_cfft_radix4by2_q15(
- q15_t * pSrc,
- uint32_t fftLen,
- const q15_t * pCoef);
+ q15_t * pSrc,
+ uint32_t fftLen,
+ const q15_t * pCoef);
void arm_cfft_radix4by2_inverse_q15(
- q15_t * pSrc,
- uint32_t fftLen,
- const q15_t * pCoef);
+ q15_t * pSrc,
+ uint32_t fftLen,
+ const q15_t * pCoef);
/**
-* @ingroup groupTransforms
-*/
+ @ingroup groupTransforms
+ */
/**
-* @addtogroup ComplexFFT
-* @{
-*/
+ @addtogroup ComplexFFT
+ @{
+ */
/**
-* @details
-* @brief Processing function for the Q15 complex FFT.
-* @param[in] *S points to an instance of the Q15 CFFT structure.
-* @param[in, out] *p1 points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place.
-* @param[in] ifftFlag flag that selects forward (ifftFlag=0) or inverse (ifftFlag=1) transform.
-* @param[in] bitReverseFlag flag that enables (bitReverseFlag=1) or disables (bitReverseFlag=0) bit reversal of output.
-* @return none.
-*/
+ @brief Processing function for Q15 complex FFT.
+ @param[in] S points to an instance of Q15 CFFT structure
+ @param[in,out] p1 points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place
+ @param[in] ifftFlag flag that selects transform direction
+ - value = 0: forward transform
+ - value = 1: inverse transform
+ @param[in] bitReverseFlag flag that enables / disables bit reversal of output
+ - value = 0: disables bit reversal of output
+ - value = 1: enables bit reversal of output
+ @return none
+ */
void arm_cfft_q15(
- const arm_cfft_instance_q15 * S,
- q15_t * p1,
- uint8_t ifftFlag,
- uint8_t bitReverseFlag)
+ const arm_cfft_instance_q15 * S,
+ q15_t * p1,
+ uint8_t ifftFlag,
+ uint8_t bitReverseFlag)
{
- uint32_t L = S->fftLen;
-
- if (ifftFlag == 1U)
- {
- switch (L)
- {
- case 16:
- case 64:
- case 256:
- case 1024:
- case 4096:
- arm_radix4_butterfly_inverse_q15 ( p1, L, (q15_t*)S->pTwiddle, 1 );
- break;
-
- case 32:
- case 128:
- case 512:
- case 2048:
- arm_cfft_radix4by2_inverse_q15 ( p1, L, S->pTwiddle );
- break;
- }
- }
- else
- {
- switch (L)
- {
- case 16:
- case 64:
- case 256:
- case 1024:
- case 4096:
- arm_radix4_butterfly_q15 ( p1, L, (q15_t*)S->pTwiddle, 1 );
- break;
-
- case 32:
- case 128:
- case 512:
- case 2048:
- arm_cfft_radix4by2_q15 ( p1, L, S->pTwiddle );
- break;
- }
- }
-
- if ( bitReverseFlag )
- arm_bitreversal_16((uint16_t*)p1,S->bitRevLength,S->pBitRevTable);
+ uint32_t L = S->fftLen;
+
+ if (ifftFlag == 1U)
+ {
+ switch (L)
+ {
+ case 16:
+ case 64:
+ case 256:
+ case 1024:
+ case 4096:
+ arm_radix4_butterfly_inverse_q15 ( p1, L, (q15_t*)S->pTwiddle, 1 );
+ break;
+
+ case 32:
+ case 128:
+ case 512:
+ case 2048:
+ arm_cfft_radix4by2_inverse_q15 ( p1, L, S->pTwiddle );
+ break;
+ }
+ }
+ else
+ {
+ switch (L)
+ {
+ case 16:
+ case 64:
+ case 256:
+ case 1024:
+ case 4096:
+ arm_radix4_butterfly_q15 ( p1, L, (q15_t*)S->pTwiddle, 1 );
+ break;
+
+ case 32:
+ case 128:
+ case 512:
+ case 2048:
+ arm_cfft_radix4by2_q15 ( p1, L, S->pTwiddle );
+ break;
+ }
+ }
+
+ if ( bitReverseFlag )
+ arm_bitreversal_16 ((uint16_t*) p1, S->bitRevLength, S->pBitRevTable);
}
/**
-* @} end of ComplexFFT group
-*/
+ @} end of ComplexFFT group
+ */
void arm_cfft_radix4by2_q15(
- q15_t * pSrc,
- uint32_t fftLen,
- const q15_t * pCoef)
+ q15_t * pSrc,
+ uint32_t fftLen,
+ const q15_t * pCoef)
{
- uint32_t i;
- uint32_t n2;
- q15_t p0, p1, p2, p3;
+ uint32_t i;
+ uint32_t n2;
+ q15_t p0, p1, p2, p3;
#if defined (ARM_MATH_DSP)
- q31_t T, S, R;
- q31_t coeff, out1, out2;
- const q15_t *pC = pCoef;
- q15_t *pSi = pSrc;
- q15_t *pSl = pSrc + fftLen;
+ q31_t T, S, R;
+ q31_t coeff, out1, out2;
+ const q15_t *pC = pCoef;
+ q15_t *pSi = pSrc;
+ q15_t *pSl = pSrc + fftLen;
#else
- uint32_t ia, l;
- q15_t xt, yt, cosVal, sinVal;
+ uint32_t l;
+ q15_t xt, yt, cosVal, sinVal;
#endif
- n2 = fftLen >> 1;
+ n2 = fftLen >> 1U;
#if defined (ARM_MATH_DSP)
- for (i = n2; i > 0; i--)
- {
- coeff = _SIMD32_OFFSET(pC);
- pC += 2;
-
- T = _SIMD32_OFFSET(pSi);
- T = __SHADD16(T, 0); // this is just a SIMD arithmetic shift right by 1
+ for (i = n2; i > 0; i--)
+ {
+ coeff = read_q15x2_ia ((q15_t **) &pC);
- S = _SIMD32_OFFSET(pSl);
- S = __SHADD16(S, 0); // this is just a SIMD arithmetic shift right by 1
+ T = read_q15x2 (pSi);
+ T = __SHADD16(T, 0); /* this is just a SIMD arithmetic shift right by 1 */
- R = __QSUB16(T, S);
+ S = read_q15x2 (pSl);
+ S = __SHADD16(S, 0); /* this is just a SIMD arithmetic shift right by 1 */
- _SIMD32_OFFSET(pSi) = __SHADD16(T, S);
- pSi += 2;
+ R = __QSUB16(T, S);
- #ifndef ARM_MATH_BIG_ENDIAN
+ write_q15x2_ia (&pSi, __SHADD16(T, S));
- out1 = __SMUAD(coeff, R) >> 16;
- out2 = __SMUSDX(coeff, R);
-
- #else
+#ifndef ARM_MATH_BIG_ENDIAN
+ out1 = __SMUAD(coeff, R) >> 16U;
+ out2 = __SMUSDX(coeff, R);
+#else
+ out1 = __SMUSDX(R, coeff) >> 16U;
+ out2 = __SMUAD(coeff, R);
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
- out1 = __SMUSDX(R, coeff) >> 16U;
- out2 = __SMUAD(coeff, R);
+ write_q15x2_ia (&pSl, (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF));
+ }
- #endif // #ifndef ARM_MATH_BIG_ENDIAN
+#else /* #if defined (ARM_MATH_DSP) */
- _SIMD32_OFFSET(pSl) =
- (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF);
- pSl += 2;
- }
+ for (i = 0; i < n2; i++)
+ {
+ cosVal = pCoef[2 * i];
+ sinVal = pCoef[2 * i + 1];
-#else // #if defined (ARM_MATH_DSP)
+ l = i + n2;
- ia = 0;
- for (i = 0; i < n2; i++)
- {
- cosVal = pCoef[ia * 2];
- sinVal = pCoef[(ia * 2) + 1];
- ia++;
+ xt = (pSrc[2 * i] >> 1U) - (pSrc[2 * l] >> 1U);
+ pSrc[2 * i] = ((pSrc[2 * i] >> 1U) + (pSrc[2 * l] >> 1U)) >> 1U;
- l = i + n2;
+ yt = (pSrc[2 * i + 1] >> 1U) - (pSrc[2 * l + 1] >> 1U);
+ pSrc[2 * i + 1] = ((pSrc[2 * l + 1] >> 1U) + (pSrc[2 * i + 1] >> 1U)) >> 1U;
- xt = (pSrc[2 * i] >> 1U) - (pSrc[2 * l] >> 1U);
- pSrc[2 * i] = ((pSrc[2 * i] >> 1U) + (pSrc[2 * l] >> 1U)) >> 1U;
+ pSrc[2 * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16U)) +
+ ((int16_t) (((q31_t) yt * sinVal) >> 16U)) );
- yt = (pSrc[2 * i + 1] >> 1U) - (pSrc[2 * l + 1] >> 1U);
- pSrc[2 * i + 1] =
- ((pSrc[2 * l + 1] >> 1U) + (pSrc[2 * i + 1] >> 1U)) >> 1U;
+ pSrc[2 * l + 1] = (((int16_t) (((q31_t) yt * cosVal) >> 16U)) -
+ ((int16_t) (((q31_t) xt * sinVal) >> 16U)) );
+ }
- pSrc[2U * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16)) +
- ((int16_t) (((q31_t) yt * sinVal) >> 16)));
+#endif /* #if defined (ARM_MATH_DSP) */
- pSrc[2U * l + 1U] = (((int16_t) (((q31_t) yt * cosVal) >> 16)) -
- ((int16_t) (((q31_t) xt * sinVal) >> 16)));
- }
+ /* first col */
+ arm_radix4_butterfly_q15( pSrc, n2, (q15_t*)pCoef, 2U);
-#endif // #if defined (ARM_MATH_DSP)
+ /* second col */
+ arm_radix4_butterfly_q15( pSrc + fftLen, n2, (q15_t*)pCoef, 2U);
- // first col
- arm_radix4_butterfly_q15( pSrc, n2, (q15_t*)pCoef, 2U);
- // second col
- arm_radix4_butterfly_q15( pSrc + fftLen, n2, (q15_t*)pCoef, 2U);
+ n2 = fftLen >> 1U;
+ for (i = 0; i < n2; i++)
+ {
+ p0 = pSrc[4 * i + 0];
+ p1 = pSrc[4 * i + 1];
+ p2 = pSrc[4 * i + 2];
+ p3 = pSrc[4 * i + 3];
- for (i = 0; i < fftLen >> 1; i++)
- {
- p0 = pSrc[4*i+0];
- p1 = pSrc[4*i+1];
- p2 = pSrc[4*i+2];
- p3 = pSrc[4*i+3];
+ p0 <<= 1U;
+ p1 <<= 1U;
+ p2 <<= 1U;
+ p3 <<= 1U;
- p0 <<= 1;
- p1 <<= 1;
- p2 <<= 1;
- p3 <<= 1;
+ pSrc[4 * i + 0] = p0;
+ pSrc[4 * i + 1] = p1;
+ pSrc[4 * i + 2] = p2;
+ pSrc[4 * i + 3] = p3;
+ }
- pSrc[4*i+0] = p0;
- pSrc[4*i+1] = p1;
- pSrc[4*i+2] = p2;
- pSrc[4*i+3] = p3;
- }
}
void arm_cfft_radix4by2_inverse_q15(
- q15_t * pSrc,
- uint32_t fftLen,
- const q15_t * pCoef)
+ q15_t * pSrc,
+ uint32_t fftLen,
+ const q15_t * pCoef)
{
- uint32_t i;
- uint32_t n2;
- q15_t p0, p1, p2, p3;
+ uint32_t i;
+ uint32_t n2;
+ q15_t p0, p1, p2, p3;
#if defined (ARM_MATH_DSP)
- q31_t T, S, R;
- q31_t coeff, out1, out2;
- const q15_t *pC = pCoef;
- q15_t *pSi = pSrc;
- q15_t *pSl = pSrc + fftLen;
+ q31_t T, S, R;
+ q31_t coeff, out1, out2;
+ const q15_t *pC = pCoef;
+ q15_t *pSi = pSrc;
+ q15_t *pSl = pSrc + fftLen;
#else
- uint32_t ia, l;
- q15_t xt, yt, cosVal, sinVal;
+ uint32_t l;
+ q15_t xt, yt, cosVal, sinVal;
#endif
- n2 = fftLen >> 1;
+ n2 = fftLen >> 1U;
#if defined (ARM_MATH_DSP)
- for (i = n2; i > 0; i--)
- {
- coeff = _SIMD32_OFFSET(pC);
- pC += 2;
-
- T = _SIMD32_OFFSET(pSi);
- T = __SHADD16(T, 0); // this is just a SIMD arithmetic shift right by 1
+ for (i = n2; i > 0; i--)
+ {
+ coeff = read_q15x2_ia ((q15_t **) &pC);
- S = _SIMD32_OFFSET(pSl);
- S = __SHADD16(S, 0); // this is just a SIMD arithmetic shift right by 1
+ T = read_q15x2 (pSi);
+ T = __SHADD16(T, 0); /* this is just a SIMD arithmetic shift right by 1 */
- R = __QSUB16(T, S);
+ S = read_q15x2 (pSl);
+ S = __SHADD16(S, 0); /* this is just a SIMD arithmetic shift right by 1 */
- _SIMD32_OFFSET(pSi) = __SHADD16(T, S);
- pSi += 2;
+ R = __QSUB16(T, S);
- #ifndef ARM_MATH_BIG_ENDIAN
+ write_q15x2_ia (&pSi, __SHADD16(T, S));
- out1 = __SMUSD(coeff, R) >> 16;
- out2 = __SMUADX(coeff, R);
- #else
+#ifndef ARM_MATH_BIG_ENDIAN
+ out1 = __SMUSD(coeff, R) >> 16U;
+ out2 = __SMUADX(coeff, R);
+#else
+ out1 = __SMUADX(R, coeff) >> 16U;
+ out2 = __SMUSD(__QSUB(0, coeff), R);
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
- out1 = __SMUADX(R, coeff) >> 16U;
- out2 = __SMUSD(__QSUB(0, coeff), R);
+ write_q15x2_ia (&pSl, (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF));
+ }
- #endif // #ifndef ARM_MATH_BIG_ENDIAN
+#else /* #if defined (ARM_MATH_DSP) */
- _SIMD32_OFFSET(pSl) =
- (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF);
- pSl += 2;
- }
+ for (i = 0; i < n2; i++)
+ {
+ cosVal = pCoef[2 * i];
+ sinVal = pCoef[2 * i + 1];
-#else // #if defined (ARM_MATH_DSP)
+ l = i + n2;
- ia = 0;
- for (i = 0; i < n2; i++)
- {
- cosVal = pCoef[ia * 2];
- sinVal = pCoef[(ia * 2) + 1];
- ia++;
+ xt = (pSrc[2 * i] >> 1U) - (pSrc[2 * l] >> 1U);
+ pSrc[2 * i] = ((pSrc[2 * i] >> 1U) + (pSrc[2 * l] >> 1U)) >> 1U;
- l = i + n2;
- xt = (pSrc[2 * i] >> 1U) - (pSrc[2 * l] >> 1U);
- pSrc[2 * i] = ((pSrc[2 * i] >> 1U) + (pSrc[2 * l] >> 1U)) >> 1U;
+ yt = (pSrc[2 * i + 1] >> 1U) - (pSrc[2 * l + 1] >> 1U);
+ pSrc[2 * i + 1] = ((pSrc[2 * l + 1] >> 1U) + (pSrc[2 * i + 1] >> 1U)) >> 1U;
- yt = (pSrc[2 * i + 1] >> 1U) - (pSrc[2 * l + 1] >> 1U);
- pSrc[2 * i + 1] =
- ((pSrc[2 * l + 1] >> 1U) + (pSrc[2 * i + 1] >> 1U)) >> 1U;
+ pSrc[2 * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16U)) -
+ ((int16_t) (((q31_t) yt * sinVal) >> 16U)) );
- pSrc[2U * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16)) -
- ((int16_t) (((q31_t) yt * sinVal) >> 16)));
+ pSrc[2 * l + 1] = (((int16_t) (((q31_t) yt * cosVal) >> 16U)) +
+ ((int16_t) (((q31_t) xt * sinVal) >> 16U)) );
+ }
- pSrc[2U * l + 1U] = (((int16_t) (((q31_t) yt * cosVal) >> 16)) +
- ((int16_t) (((q31_t) xt * sinVal) >> 16)));
- }
+#endif /* #if defined (ARM_MATH_DSP) */
-#endif // #if defined (ARM_MATH_DSP)
+ /* first col */
+ arm_radix4_butterfly_inverse_q15( pSrc, n2, (q15_t*)pCoef, 2U);
- // first col
- arm_radix4_butterfly_inverse_q15( pSrc, n2, (q15_t*)pCoef, 2U);
- // second col
- arm_radix4_butterfly_inverse_q15( pSrc + fftLen, n2, (q15_t*)pCoef, 2U);
+ /* second col */
+ arm_radix4_butterfly_inverse_q15( pSrc + fftLen, n2, (q15_t*)pCoef, 2U);
- for (i = 0; i < fftLen >> 1; i++)
- {
- p0 = pSrc[4*i+0];
- p1 = pSrc[4*i+1];
- p2 = pSrc[4*i+2];
- p3 = pSrc[4*i+3];
+ n2 = fftLen >> 1U;
+ for (i = 0; i < n2; i++)
+ {
+ p0 = pSrc[4 * i + 0];
+ p1 = pSrc[4 * i + 1];
+ p2 = pSrc[4 * i + 2];
+ p3 = pSrc[4 * i + 3];
- p0 <<= 1;
- p1 <<= 1;
- p2 <<= 1;
- p3 <<= 1;
+ p0 <<= 1U;
+ p1 <<= 1U;
+ p2 <<= 1U;
+ p3 <<= 1U;
- pSrc[4*i+0] = p0;
- pSrc[4*i+1] = p1;
- pSrc[4*i+2] = p2;
- pSrc[4*i+3] = p3;
- }
+ pSrc[4 * i + 0] = p0;
+ pSrc[4 * i + 1] = p1;
+ pSrc[4 * i + 2] = p2;
+ pSrc[4 * i + 3] = p3;
+ }
}
-
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-29 10:15:36 +0300