void arm_negate_q7(
q7_t * pSrc,
q7_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
#ifndef ARM_MATH_CM0
/* Run the below code for Cortex-M4 and Cortex-M3 */
q7_t in1; /* Input value1 */
q7_t in2; /* Input value2 */
q7_t in3; /* Input value3 */
q7_t in4; /* Input value4 */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u) {
/* C = -A */
/* Read four inputs */
in1 = *pSrc++;
in2 = *pSrc++;
in3 = *pSrc++;
in4 = *pSrc++;
/* Store the Negated results in the destination buffer in a single cycle by packing the results */
*__SIMD32(pDst)++ =
__PACKq7(__SSAT(-in1, 8), __SSAT(-in2, 8), __SSAT(-in3, 8),
__SSAT(-in4, 8));
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
#else
/* Run the below code for Cortex-M0 */
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
#endif /* #ifndef ARM_MATH_CM0 */
while(blkCnt > 0u) {
/* C = -A */
/* Negate and then store the results in the destination buffer. */
*pDst++ = __SSAT(-*pSrc++, 8);
/* Decrement the loop counter */
blkCnt--;
}
}
buffer_s16_t FM::execute(
const buffer_c16_t& src,
const buffer_s16_t& dst
) {
auto z = z_;
const auto src_p = src.p;
const auto src_end = &src.p[src.count];
auto dst_p = dst.p;
while(src_p < src_end) {
const auto s0 = *__SIMD32(src_p)++;
const auto s1 = *__SIMD32(src_p)++;
const auto t0 = multiply_conjugate_s16_s32(s0, z);
const auto t1 = multiply_conjugate_s16_s32(s1, s0);
z = s1;
const int32_t theta0_int = angle_approx_0deg27(t0) * ks16;
const int32_t theta0_sat = __SSAT(theta0_int, 16);
const int32_t theta1_int = angle_approx_0deg27(t1) * ks16;
const int32_t theta1_sat = __SSAT(theta1_int, 16);
*__SIMD32(dst_p)++ = __PKHBT(
theta0_sat,
theta1_sat,
16
);
}
z_ = z;
return { dst.p, src.count, src.sampling_rate };
}
void arm_pid_init_q15(
arm_pid_instance_q15 * S,
int32_t resetStateFlag)
{
#ifndef ARM_MATH_CM0_FAMILY
/* Run the below code for Cortex-M4 and Cortex-M3 */
/* Derived coefficient A0 */
S->A0 = __QADD16(__QADD16(S->Kp, S->Ki), S->Kd);
/* Derived coefficients and pack into A1 */
#ifndef ARM_MATH_BIG_ENDIAN
S->A1 = __PKHBT(-__QADD16(__QADD16(S->Kd, S->Kd), S->Kp), S->Kd, 16);
#else
S->A1 = __PKHBT(S->Kd, -__QADD16(__QADD16(S->Kd, S->Kd), S->Kp), 16);
#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
/* Check whether state needs reset or not */
if(resetStateFlag)
{
/* Clear the state buffer. The size will be always 3 samples */
memset(S->state, 0, 3u * sizeof(q15_t));
}
#else
/* Run the below code for Cortex-M0 */
q31_t temp; /*to store the sum */
/* Derived coefficient A0 */
temp = S->Kp + S->Ki + S->Kd;
S->A0 = (q15_t) __SSAT(temp, 16);
/* Derived coefficients and pack into A1 */
temp = -(S->Kd + S->Kd + S->Kp);
S->A1 = (q15_t) __SSAT(temp, 16);
S->A2 = S->Kd;
/* Check whether state needs reset or not */
if(resetStateFlag)
{
/* Clear the state buffer. The size will be always 3 samples */
memset(S->state, 0, 3u * sizeof(q15_t));
}
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
}
void arm_abs_q7(
q7_t * pSrc,
q7_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
q7_t in1; /* Input value1 */
q7_t in2; /* Input value2 */
q7_t in3; /* Input value3 */
q7_t in4; /* Input value4 */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C = |A| */
/* Read 4 inputs */
in1 = *pSrc++;
in2 = *pSrc++;
in3 = *pSrc++;
in4 = *pSrc++;
/* Store the Absolute result in the destination buffer by packing the 4 values in single cycle */
*__SIMD32(pDst)++ =
__PACKq7(((in1 > 0) ? in1 : __SSAT(-in1, 8)),
((in2 > 0) ? in2 : __SSAT(-in2, 8)),
((in3 > 0) ? in3 : __SSAT(-in3, 8)),
((in4 > 0) ? in4 : __SSAT(-in4, 8)));
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u)
{
/* C = |A| */
/* Read the input */
in1 = *pSrc++;
/* Store the Absolute result in the destination buffer */
*pDst++ = (in1 > 0) ? in1 : __SSAT(-in1, 8);
/* Decrement the loop counter */
blkCnt--;
}
}
void arm_sub_q15(
q15_t * pSrcA,
q15_t * pSrcB,
q15_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
#ifndef ARM_MATH_CM0
/* Run the below code for Cortex-M4 and Cortex-M3 */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u) {
/* C = A - B */
/* Subtract and then store the results in the destination buffer two samples at a time. */
*__SIMD32(pDst)++ = __QSUB16(*__SIMD32(pSrcA)++, *__SIMD32(pSrcB)++);
*__SIMD32(pDst)++ = __QSUB16(*__SIMD32(pSrcA)++, *__SIMD32(pSrcB)++);
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u) {
/* C = A - B */
/* Subtract and then store the result in the destination buffer. */
*pDst++ = (q15_t) __QSUB16(*pSrcA++, *pSrcB++);
/* Decrement the loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
while(blkCnt > 0u) {
/* C = A - B */
/* Subtract and then store the result in the destination buffer. */
*pDst++ = (q15_t) __SSAT(((q31_t) * pSrcA++ - *pSrcB++), 16);
/* Decrement the loop counter */
blkCnt--;
}
#endif /* #ifndef ARM_MATH_CM0 */
}
void arm_negate_q15(
q15_t * pSrc,
q15_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
q15_t in1, in2; /* Temporary variables */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C = ~A */
/* Read two inputs */
in1 = *pSrc++;
in2 = *pSrc++;
/* Negate and then store the results in the destination buffer by packing. */
*__SIMD32(pDst)++ = __PKHBT(__SSAT(-in1, 16), __SSAT(-in2, 16), 16);
in1 = *pSrc++;
in2 = *pSrc++;
*__SIMD32(pDst)++ = __PKHBT(__SSAT(-in1, 16), __SSAT(-in2, 16), 16);
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u)
{
/* C = ~A */
/* Negate and then store the result in the destination buffer. */
*pDst++ = __SSAT(-*pSrc++, 16);
/* Decrement the loop counter */
blkCnt--;
}
}
void arm_cmplx_conj_q15(
q15_t * pSrc,
q15_t * pDst,
uint32_t numSamples)
{
uint32_t blkCnt; /* loop counter */
/*loop Unrolling */
blkCnt = numSamples >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C[0]+jC[1] = A[0]+ j (-1) A[1] */
/* Calculate Complex Conjugate and then store the results in the destination buffer. */
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
/* Decrement the loop counter */
blkCnt--;
}
/* If the numSamples is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = numSamples % 0x4u;
while(blkCnt > 0u)
{
/* C[0]+jC[1] = A[0]+ j (-1) A[1] */
/* Calculate Complex Conjugate and then store the results in the destination buffer. */
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
/* Decrement the loop counter */
blkCnt--;
}
}
void AudioOutput::fill_audio_buffer(const buffer_f32_t& audio, const bool send_to_fifo) {
std::array<int16_t, 32> audio_int;
auto audio_buffer = audio::dma::tx_empty_buffer();
for(size_t i=0; i<audio_buffer.count; i++) {
const int32_t sample_int = audio.p[i] * k;
const int32_t sample_saturated = __SSAT(sample_int, 16);
audio_buffer.p[i].left = audio_buffer.p[i].right = sample_saturated;
audio_int[i] = sample_saturated;
}
if( stream && send_to_fifo ) {
stream->write(audio_int.data(), audio_buffer.count * sizeof(audio_int[0]));
}
feed_audio_stats(audio);
}
void arm_softmax_q15(const q15_t * vec_in, const uint16_t dim_vec, q15_t * p_out)
{
q31_t sum;
int16_t i;
q31_t min, max;
max = -1 * 0x100000;
min = 0x100000;
for (i = 0; i < dim_vec; i++)
{
if (vec_in[i] > max)
{
max = vec_in[i];
}
if (vec_in[i] < min)
{
min = vec_in[i];
}
}
/* we ignore really small values
* anyway, they will be 0 after shrinking
* to q7_t
*/
if (max - min > 16)
{
min = max - 16;
}
sum = 0;
for (i = 0; i < dim_vec; i++)
{
sum += 0x1 << (vec_in[i] - min);
}
for (i = 0; i < dim_vec; i++)
{
/* we leave 7-bit dynamic range, so that 128 -> 100% confidence */
p_out[i] = (q15_t) __SSAT(((0x1 << (vec_in[i] - min + 14)) / sum), 16);
}
}
void arm_mult_q31(
q31_t * pSrcA,
q31_t * pSrcB,
q31_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counters */
#ifndef ARM_MATH_CM0_FAMILY
/* Run the below code for Cortex-M4 and Cortex-M3 */
q31_t inA1, inA2, inA3, inA4; /* temporary input variables */
q31_t inB1, inB2, inB3, inB4; /* temporary input variables */
q31_t out1, out2, out3, out4; /* temporary output variables */
/* loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C = A * B */
/* Multiply the inputs and then store the results in the destination buffer. */
inA1 = *pSrcA++;
inA2 = *pSrcA++;
inA3 = *pSrcA++;
inA4 = *pSrcA++;
inB1 = *pSrcB++;
inB2 = *pSrcB++;
inB3 = *pSrcB++;
inB4 = *pSrcB++;
out1 = ((q63_t) inA1 * inB1) >> 32;
out2 = ((q63_t) inA2 * inB2) >> 32;
out3 = ((q63_t) inA3 * inB3) >> 32;
out4 = ((q63_t) inA4 * inB4) >> 32;
out1 = __SSAT(out1, 31);
out2 = __SSAT(out2, 31);
out3 = __SSAT(out3, 31);
out4 = __SSAT(out4, 31);
*pDst++ = out1 << 1u;
*pDst++ = out2 << 1u;
*pDst++ = out3 << 1u;
*pDst++ = out4 << 1u;
/* Decrement the blockSize loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u)
{
/* C = A * B */
/* Multiply the inputs and then store the results in the destination buffer. */
inA1 = *pSrcA++;
inB1 = *pSrcB++;
out1 = ((q63_t) inA1 * inB1) >> 32;
out1 = __SSAT(out1, 31);
*pDst++ = out1 << 1u;
/* Decrement the blockSize loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
while(blkCnt > 0u)
{
/* C = A * B */
/* Multiply the inputs and then store the results in the destination buffer. */
*pDst++ =
(q31_t) clip_q63_to_q31(((q63_t) (*pSrcA++) * (*pSrcB++)) >> 31);
/* Decrement the blockSize loop counter */
blkCnt--;
}
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
}
void arm_sub_q7(
q7_t * pSrcA,
q7_t * pSrcB,
q7_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
q31_t inA1, inB1, inA2, inB2; /* temporary input variabels */
q7_t inA, inB; /* temporary variables */
q31_t out1, out2, out3, out4; /* temporary output variables */
/*loop Unrolling */
blkCnt = blockSize >> 4u;
/* First part of the processing with loop unrolling. Compute 16 outputs at a time.
** a second loop below computes the remaining 1 to 15 samples. */
while(blkCnt > 0u)
{
/* C = A - B */
/* Subtract and then store the results in the destination buffer 4 samples at a time. */
/* read 4 samples at a time from sourceA */
inA1 = _SIMD32_OFFSET(pSrcA);
/* read 4 samples at a time from sourceB */
inB1 = _SIMD32_OFFSET(pSrcB);
/* read 4 samples at a time from sourceA */
inA2 = _SIMD32_OFFSET(pSrcA + 4);
/* out = saturate(sourceA - sourceB) four samples at a time */
out1 = __QSUB8(inA1, inB1);
/* read 4 samples at a time from sourceB */
inB2 = _SIMD32_OFFSET(pSrcB + 4);
/* store result to destination four samples at a time */
_SIMD32_OFFSET(pDst) = out1;
/* out = saturate(sourceA - sourceB) four samples at a time */
out2 = __QSUB8(inA2, inB2);
/* read 4 samples at a time from sourceA */
inA1 = _SIMD32_OFFSET(pSrcA + 8);
/* read 4 samples at a time from sourceB */
inB1 = _SIMD32_OFFSET(pSrcB + 8);
/* read 4 samples at a time from sourceA */
inA2 = _SIMD32_OFFSET(pSrcA + 12);
/* out = saturate(sourceA - sourceB) four samples at a time */
out3 = __QSUB8(inA1, inB1);
/* read 4 samples at a time from sourceB */
inB2 = _SIMD32_OFFSET(pSrcB + 12);
/* increment sourceA pointer by 16 to process next samples */
pSrcA += 16u;
/* store result to destination four samples at a time */
_SIMD32_OFFSET(pDst + 4) = out2;
/* out = saturate(sourceA - sourceB) four samples at a time */
out4 = __QSUB8(inA2, inB2);
/* store result to destination four samples at a time */
_SIMD32_OFFSET(pDst + 8) = out3;
/* Update source pointer to process next sampels */
pSrcB += 16u;
/* store result to destination four samples at a time */
_SIMD32_OFFSET(pDst + 12) = out4;
/* Update destination pointer to process next sampels */
pDst += 16u;
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 16, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x10u;
while(blkCnt > 0u)
{
/* C = A - B */
/* Subtract and then store the result in the destination buffer. */
inA = *pSrcA++;
inB = *pSrcB++;
#ifdef CCS
*pDst++ = __SSATA(inA - inB, 0, 8);
#else
*pDst++ = __SSAT(inA - inB, 8);
#endif //#ifdef CCS
/* Decrement the loop counter */
blkCnt--;
}
}
void arm_sub_q15(
const q15_t * pSrcA,
const q15_t * pSrcB,
q15_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* Loop counter */
#if defined (ARM_MATH_LOOPUNROLL)
#if defined (ARM_MATH_DSP)
q31_t inA1, inA2;
q31_t inB1, inB2;
#endif
/* Loop unrolling: Compute 4 outputs at a time */
blkCnt = blockSize >> 2U;
while (blkCnt > 0U)
{
/* C = A - B */
#if defined (ARM_MATH_DSP)
/* read 2 times 2 samples at a time from sourceA */
inA1 = read_q15x2_ia ((q15_t **) &pSrcA);
inA2 = read_q15x2_ia ((q15_t **) &pSrcA);
/* read 2 times 2 samples at a time from sourceB */
inB1 = read_q15x2_ia ((q15_t **) &pSrcB);
inB2 = read_q15x2_ia ((q15_t **) &pSrcB);
/* Subtract and store 2 times 2 samples at a time */
write_q15x2_ia (&pDst, __QSUB16(inA1, inB1));
write_q15x2_ia (&pDst, __QSUB16(inA2, inB2));
#else
*pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
*pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
*pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
*pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
#endif
/* Decrement loop counter */
blkCnt--;
}
/* Loop unrolling: Compute remaining outputs */
blkCnt = blockSize % 0x4U;
#else
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
while (blkCnt > 0U)
{
/* C = A - B */
/* Subtract and store result in destination buffer. */
#if defined (ARM_MATH_DSP)
*pDst++ = (q15_t) __QSUB16(*pSrcA++, *pSrcB++);
#else
*pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
#endif
/* Decrement loop counter */
blkCnt--;
}
}
void arm_cmplx_conj_q15(
q15_t * pSrc,
q15_t * pDst,
uint32_t numSamples)
{
#ifndef ARM_MATH_CM0
/* Run the below code for Cortex-M4 and Cortex-M3 */
uint32_t blkCnt; /* loop counter */
/*loop Unrolling */
blkCnt = numSamples >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C[0]+jC[1] = A[0]+ j (-1) A[1] */
/* Calculate Complex Conjugate and then store the results in the destination buffer. */
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
/* Decrement the loop counter */
blkCnt--;
}
/* If the numSamples is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = numSamples % 0x4u;
while(blkCnt > 0u)
{
/* C[0]+jC[1] = A[0]+ j (-1) A[1] */
/* Calculate Complex Conjugate and then store the results in the destination buffer. */
*pDst++ = *pSrc++;
*pDst++ = __SSAT(-*pSrc++, 16);
/* Decrement the loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
while(numSamples > 0u)
{
/* realOut + j (imagOut) = realIn+ j (-1) imagIn */
/* Calculate Complex Conjugate and then store the results in the destination buffer. */
*pDst++ = *pSrc++;
*pDst++ = -*pSrc++;
/* Decrement the loop counter */
numSamples--;
}
#endif /* #ifndef ARM_MATH_CM0 */
}
void arm_negate_q15(
q15_t * pSrc,
q15_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
#ifndef ARM_MATH_CM0
/* Run the below code for Cortex-M4 and Cortex-M3 */
q15_t in1, in2; /* Temporary variables */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C = -A */
/* Read two inputs */
in1 = *pSrc++;
in2 = *pSrc++;
/* Negate and then store the results in the destination buffer by packing. */
#ifndef ARM_MATH_BIG_ENDIAN
*__SIMD32(pDst)++ = __PKHBT(__SSAT(-in1, 16), __SSAT(-in2, 16), 16);
#else
*__SIMD32(pDst)++ = __PKHBT(__SSAT(-in2, 16), __SSAT(-in1, 16), 16);
#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
in1 = *pSrc++;
in2 = *pSrc++;
#ifndef ARM_MATH_BIG_ENDIAN
*__SIMD32(pDst)++ = __PKHBT(__SSAT(-in1, 16), __SSAT(-in2, 16), 16);
#else
*__SIMD32(pDst)++ = __PKHBT(__SSAT(-in2, 16), __SSAT(-in1, 16), 16);
#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
#else
/* Run the below code for Cortex-M0 */
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
#endif /* #ifndef ARM_MATH_CM0 */
while(blkCnt > 0u)
{
/* C = -A */
/* Negate and then store the result in the destination buffer. */
*pDst++ = __SSAT(-*pSrc++, 16);
/* Decrement the loop counter */
blkCnt--;
}
}
extern "C" void stm32_adc12(void) {
PERF_COUNT_START
static float last_v_ab[2];
static float prev_v_ab[2];
static uint8_t last_pwm_sector = 1u;
static uint8_t prev_pwm_sector = 1u;
int16_t phase_current_lsb[3];
uint16_t phase_oc[3];
float out_v_ab[2], i_ab[2];
float temp;
hal_read_phase_shunts_(phase_current_lsb, prev_pwm_sector);
/*
Clarke transformation for balanced systems
i_alpha = i_a,
i_beta = (2 * i_b + i_a) / sqrt(3)
Multiply by 8 because the phase current readings are right-aligned.
*/
i_ab[0] = float(phase_current_lsb[0]) *
float(hal_full_scale_current_a * 8.0 / 32768.0);
temp = float(phase_current_lsb[1]) *
float(hal_full_scale_current_a * 8.0 / 32768.0);
i_ab[1] = (0.57735026919f * i_ab[0] + 1.15470053838f * temp);
out_v_ab[0] = out_v_ab[1] = 0.0f;
if (high_frequency_task_) {
high_frequency_task_(out_v_ab, prev_v_ab, i_ab, vbus_v_);
}
prev_v_ab[0] = last_v_ab[0];
prev_v_ab[1] = last_v_ab[1];
last_v_ab[0] = out_v_ab[0];
last_v_ab[1] = out_v_ab[1];
prev_pwm_sector = last_pwm_sector;
/*
Convert alpha-beta frame voltage fractions to SVM output compare values
for each phase.
*/
temp = vbus_inv_;
last_pwm_sector = svm_duty_cycle_from_v_alpha_beta(
phase_oc,
int16_t(__SSAT(int32_t(temp * out_v_ab[0]), 16u)),
int16_t(__SSAT(int32_t(temp * out_v_ab[1]), 16u)),
hal_pwm_period_ticks);
/* Update the timer */
hal_update_timer_(last_pwm_sector, phase_oc);
/*
Clear the JEOS event, and prepare for the next hardware trigger
ADC_ClearFlag(ADC1, ADC_FLAG_JEOS);
*/
putreg32(ADC_INT_JEOS, STM32_ADC1_ISR);
/*
Allow the next ADC conversion to happen based on the TIM1 CC4 event
ADC_StartInjectedConversion(ADC1);
*/
putreg32(getreg32(STM32_ADC1_CR) | ADC_CR_JADSTART, STM32_ADC1_CR);
PERF_COUNT_END
}
arm_status arm_mat_sub_q15(
const arm_matrix_instance_q15 * pSrcA,
const arm_matrix_instance_q15 * pSrcB,
arm_matrix_instance_q15 * pDst)
{
q15_t *pInA = pSrcA->pData; /* input data matrix pointer A */
q15_t *pInB = pSrcB->pData; /* input data matrix pointer B */
q15_t *pOut = pDst->pData; /* output data matrix pointer */
uint32_t numSamples; /* total number of elements in the matrix */
uint32_t blkCnt; /* loop counters */
arm_status status; /* status of matrix subtraction */
#ifdef ARM_MATH_MATRIX_CHECK
/* Check for matrix mismatch condition */
if((pSrcA->numRows != pSrcB->numRows) ||
(pSrcA->numCols != pSrcB->numCols) ||
(pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols)) {
/* Set status as ARM_MATH_SIZE_MISMATCH */
status = ARM_MATH_SIZE_MISMATCH;
} else
#endif /* #ifdef ARM_MATH_MATRIX_CHECK */
{
/* Total number of samples in the input matrix */
numSamples = (uint32_t) pSrcA->numRows * pSrcA->numCols;
#ifndef ARM_MATH_CM0
/* Run the below code for Cortex-M4 and Cortex-M3 */
/* Apply loop unrolling */
blkCnt = numSamples >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u) {
/* C(m,n) = A(m,n) - B(m,n) */
/* Subtract, Saturate and then store the results in the destination buffer. */
*__SIMD32(pOut)++ = __QSUB16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
*__SIMD32(pOut)++ = __QSUB16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = numSamples % 0x4u;
while(blkCnt > 0u) {
/* C(m,n) = A(m,n) - B(m,n) */
/* Subtract and then store the results in the destination buffer. */
*pOut++ = (q15_t) __QSUB16(*pInA++, *pInB++);
/* Decrement the loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
/* Initialize blkCnt with number of samples */
blkCnt = numSamples;
while(blkCnt > 0u) {
/* C(m,n) = A(m,n) - B(m,n) */
/* Subtract and then store the results in the destination buffer. */
*pOut++ = (q15_t) __SSAT(((q31_t) * pInA++ - *pInB++), 16);
/* Decrement the loop counter */
blkCnt--;
}
#endif /* #ifndef ARM_MATH_CM0 */
/* Set status as ARM_MATH_SUCCESS */
status = ARM_MATH_SUCCESS;
}
/* Return to application */
return (status);
}
void arm_mult_q15(
q15_t * pSrcA,
q15_t * pSrcB,
q15_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counters */
q31_t inA1, inA2, inB1, inB2; /* temporary input variables */
q15_t out1, out2, out3, out4; /* temporary output variables */
q31_t mul1, mul2, mul3, mul4; /* temporary variables */
/* loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* read two samples at a time from sourceA */
inA1 = *__SIMD32(pSrcA)++;
/* read two samples at a time from sourceB */
inB1 = *__SIMD32(pSrcB)++;
/* read two samples at a time from sourceA */
inA2 = *__SIMD32(pSrcA)++;
/* read two samples at a time from sourceB */
inB2 = *__SIMD32(pSrcB)++;
/* multiply mul = sourceA * sourceB */
mul1 = (q31_t)((q15_t)(inA1 >> 16)* (q15_t)(inB1>> 16));
mul2 = (q31_t)((q15_t)inA1 * (q15_t)inB1);
mul3 = (q31_t)((q15_t)(inA2 >> 16)* (q15_t)(inB2>> 16));
mul4 = (q31_t)((q15_t)inA2 * (q15_t)inB2);
/* shift result by 15 to get 16 bit result */
mul1 = mul1 >> 15;
mul2 = mul2 >> 15;
mul3 = mul3 >> 15;
mul4 = mul4 >> 15;
/* saturate result to 16 bit */
#ifdef CCS
out1 = (q15_t) __SSATA(mul1, 0, 16);
out2 = (q15_t) __SSATA(mul2, 0, 16);
out3 = (q15_t) __SSATA(mul3, 0, 16);
out4 = (q15_t) __SSATA(mul4, 0, 16);
#else
out1 = (q15_t) __SSAT(mul1, 16);
out2 = (q15_t) __SSAT(mul2, 16);
out3 = (q15_t) __SSAT(mul3, 16);
out4 = (q15_t) __SSAT(mul4, 16);
#endif // #ifdef CCS
/* store the result */
#ifndef ARM_MATH_BIG_ENDIAN
*__SIMD32(pDst)++ = __PKHBT(out2, out1, 16);
*__SIMD32(pDst)++ = __PKHBT(out4, out3, 16);
#else
*__SIMD32(pDst)++ = __PKHBT(out2, out1, 16);
*__SIMD32(pDst)++ = __PKHBT(out4, out3, 16);
#endif // #ifndef ARM_MATH_BIG_ENDIAN
/* Decrement the blockSize loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u)
{
/* C = A * B */
/* Multiply the inputs and store the result in the destination buffer */
#ifdef CCS
*pDst++ = (q15_t) __SSATA(((q31_t) ((*pSrcA++) * (*pSrcB++)) >> 15), 0, 16);
#else
*pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
#endif // #ifdef CCS
/* Decrement the blockSize loop counter */
blkCnt--;
}
}
void arm_float_to_q7(
float32_t * pSrc,
q7_t * pDst,
uint32_t blockSize)
{
float32_t *pIn = pSrc; /* Src pointer */
uint32_t blkCnt; /* loop counter */
#ifdef ARM_MATH_ROUNDING
float32_t in;
#endif
/*loop Unrolling */
blkCnt = blockSize >> 3u;
/* First part of the processing with loop unrolling. Compute 8 outputs at a time.
** a second loop below computes the remaining 1 to 4 samples. */
while(blkCnt > 0u)
{
#ifdef ARM_MATH_ROUNDING
/* C = A * 128 */
/* convert from float to q7 and then store the results in the destination buffer */
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
#else
/* C = A * 128 */
/* convert from float to q7 and then store the results in the destination buffer */
#ifdef CCS
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
#else
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
#endif // #ifdef CCS
#endif // #ifdef ARM_MATH_ROUNDING
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 48 compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x8u;
while(blkCnt > 0u)
{
#ifdef ARM_MATH_ROUNDING
/* C = A * 128 */
/* convert from float to q7 and then store the results in the destination buffer */
in = *pIn++;
in = (in * 128);
in += in > 0 ? 0.5 : -0.5;
#ifdef CCS
*pDst++ = (q7_t) (__SSATA((q15_t) (in), 0, 8));
#else
*pDst++ = (q7_t) (__SSAT((q15_t) (in), 8));
#endif // #ifdef CCS
#else
/* C = A * 128 */
/* convert from float to q7 and then store the results in the destination buffer */
#ifdef CCS
*pDst++ = __SSATA((q31_t) (*pIn++ * 128.0f), 0, 8);
#else
*pDst++ = __SSAT((q31_t) (*pIn++ * 128.0f), 8);
#endif // #ifdef CCS
#endif // #ifdef ARM_MATH_ROUNDING
/* Decrement the loop counter */
blkCnt--;
}
}
void arm_offset_q15(
q15_t * pSrc,
q15_t offset,
q15_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* loop counter */
#ifndef ARM_MATH_CM0_FAMILY
/* Run the below code for Cortex-M4 and Cortex-M3 */
q31_t offset_packed; /* Offset packed to 32 bit */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* Offset is packed to 32 bit in order to use SIMD32 for addition */
offset_packed = __PKHBT(offset, offset, 16);
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C = A + offset */
/* Add offset and then store the results in the destination buffer, 2 samples at a time. */
*__SIMD32(pDst)++ = __QADD16(*__SIMD32(pSrc)++, offset_packed);
*__SIMD32(pDst)++ = __QADD16(*__SIMD32(pSrc)++, offset_packed);
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u)
{
/* C = A + offset */
/* Add offset and then store the results in the destination buffer. */
*pDst++ = (q15_t) __QADD16(*pSrc++, offset);
/* Decrement the loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
while(blkCnt > 0u)
{
/* C = A + offset */
/* Add offset and then store the results in the destination buffer. */
*pDst++ = (q15_t) __SSAT(((q31_t) * pSrc++ + offset), 16);
/* Decrement the loop counter */
blkCnt--;
}
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
}
void arm_cmplx_mult_real_q31(
q31_t * pSrcCmplx,
q31_t * pSrcReal,
q31_t * pCmplxDst,
uint32_t numSamples)
{
q31_t inA1; /* Temporary variable to store input value */
#ifndef ARM_MATH_CM0_FAMILY
/* Run the below code for Cortex-M4 and Cortex-M3 */
uint32_t blkCnt; /* loop counters */
q31_t inA2, inA3, inA4; /* Temporary variables to hold input data */
q31_t inB1, inB2; /* Temporary variabels to hold input data */
q31_t out1, out2, out3, out4; /* Temporary variables to hold output data */
/* loop Unrolling */
blkCnt = numSamples >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
/* C[2 * i] = A[2 * i] * B[i]. */
/* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
/* read real input from complex input buffer */
inA1 = *pSrcCmplx++;
inA2 = *pSrcCmplx++;
/* read input from real input bufer */
inB1 = *pSrcReal++;
inB2 = *pSrcReal++;
/* read imaginary input from complex input buffer */
inA3 = *pSrcCmplx++;
inA4 = *pSrcCmplx++;
/* multiply complex input with real input */
out1 = ((q63_t) inA1 * inB1) >> 32;
out2 = ((q63_t) inA2 * inB1) >> 32;
out3 = ((q63_t) inA3 * inB2) >> 32;
out4 = ((q63_t) inA4 * inB2) >> 32;
/* sature the result */
out1 = __SSAT(out1, 31);
out2 = __SSAT(out2, 31);
out3 = __SSAT(out3, 31);
out4 = __SSAT(out4, 31);
/* get result in 1.31 format */
out1 = out1 << 1;
out2 = out2 << 1;
out3 = out3 << 1;
out4 = out4 << 1;
/* store the result to destination buffer */
*pCmplxDst++ = out1;
*pCmplxDst++ = out2;
*pCmplxDst++ = out3;
*pCmplxDst++ = out4;
/* read real input from complex input buffer */
inA1 = *pSrcCmplx++;
inA2 = *pSrcCmplx++;
/* read input from real input bufer */
inB1 = *pSrcReal++;
inB2 = *pSrcReal++;
/* read imaginary input from complex input buffer */
inA3 = *pSrcCmplx++;
inA4 = *pSrcCmplx++;
/* multiply complex input with real input */
out1 = ((q63_t) inA1 * inB1) >> 32;
out2 = ((q63_t) inA2 * inB1) >> 32;
out3 = ((q63_t) inA3 * inB2) >> 32;
out4 = ((q63_t) inA4 * inB2) >> 32;
/* sature the result */
out1 = __SSAT(out1, 31);
out2 = __SSAT(out2, 31);
out3 = __SSAT(out3, 31);
out4 = __SSAT(out4, 31);
/* get result in 1.31 format */
out1 = out1 << 1;
out2 = out2 << 1;
out3 = out3 << 1;
out4 = out4 << 1;
/* store the result to destination buffer */
*pCmplxDst++ = out1;
*pCmplxDst++ = out2;
*pCmplxDst++ = out3;
*pCmplxDst++ = out4;
/* Decrement the numSamples loop counter */
blkCnt--;
}
/* If the numSamples is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = numSamples % 0x4u;
while(blkCnt > 0u)
{
/* C[2 * i] = A[2 * i] * B[i]. */
/* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
/* read real input from complex input buffer */
inA1 = *pSrcCmplx++;
inA2 = *pSrcCmplx++;
/* read input from real input bufer */
inB1 = *pSrcReal++;
/* multiply complex input with real input */
out1 = ((q63_t) inA1 * inB1) >> 32;
out2 = ((q63_t) inA2 * inB1) >> 32;
/* sature the result */
out1 = __SSAT(out1, 31);
out2 = __SSAT(out2, 31);
/* get result in 1.31 format */
out1 = out1 << 1;
out2 = out2 << 1;
/* store the result to destination buffer */
*pCmplxDst++ = out1;
*pCmplxDst++ = out2;
/* Decrement the numSamples loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
while(numSamples > 0u)
{
/* realOut = realA * realB. */
/* imagReal = imagA * realB. */
inA1 = *pSrcReal++;
/* store the result in the destination buffer. */
*pCmplxDst++ =
(q31_t) clip_q63_to_q31(((q63_t) * pSrcCmplx++ * inA1) >> 31);
*pCmplxDst++ =
(q31_t) clip_q63_to_q31(((q63_t) * pSrcCmplx++ * inA1) >> 31);
/* Decrement the numSamples loop counter */
numSamples--;
}
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
}
void arm_float_to_q15(
float32_t * pSrc,
q15_t * pDst,
uint32_t blockSize)
{
float32_t *pIn = pSrc; /* Src pointer */
uint32_t blkCnt; /* loop counter */
#ifdef ARM_MATH_ROUNDING
float32_t in;
#endif /* #ifdef ARM_MATH_ROUNDING */
#ifndef ARM_MATH_CM0
/* Run the below code for Cortex-M4 and Cortex-M3 */
/*loop Unrolling */
blkCnt = blockSize >> 2u;
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
** a second loop below computes the remaining 1 to 3 samples. */
while(blkCnt > 0u)
{
#ifdef ARM_MATH_ROUNDING
/* C = A * 32768 */
/* convert from float to q15 and then store the results in the destination buffer */
in = *pIn++;
in = (in * 32768.0f);
in += in > 0 ? 0.5 : -0.5;
*pDst++ = (q15_t) (__SSAT((q31_t) (in), 16));
in = *pIn++;
in = (in * 32768.0f);
in += in > 0 ? 0.5 : -0.5;
*pDst++ = (q15_t) (__SSAT((q31_t) (in), 16));
in = *pIn++;
in = (in * 32768.0f);
in += in > 0 ? 0.5 : -0.5;
*pDst++ = (q15_t) (__SSAT((q31_t) (in), 16));
in = *pIn++;
in = (in * 32768.0f);
in += in > 0 ? 0.5 : -0.5;
*pDst++ = (q15_t) (__SSAT((q31_t) (in), 16));
#else
/* C = A * 32768 */
/* convert from float to q15 and then store the results in the destination buffer */
*pDst++ = (q15_t) __SSAT((q31_t) (*pIn++ * 32768.0f), 16);
*pDst++ = (q15_t) __SSAT((q31_t) (*pIn++ * 32768.0f), 16);
*pDst++ = (q15_t) __SSAT((q31_t) (*pIn++ * 32768.0f), 16);
*pDst++ = (q15_t) __SSAT((q31_t) (*pIn++ * 32768.0f), 16);
#endif /* #ifdef ARM_MATH_ROUNDING */
/* Decrement the loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
** No loop unrolling is used. */
blkCnt = blockSize % 0x4u;
while(blkCnt > 0u)
{
#ifdef ARM_MATH_ROUNDING
/* C = A * 32768 */
/* convert from float to q15 and then store the results in the destination buffer */
in = *pIn++;
in = (in * 32768.0f);
in += in > 0 ? 0.5 : -0.5;
*pDst++ = (q15_t) (__SSAT((q31_t) (in), 16));
#else
/* C = A * 32768 */
/* convert from float to q15 and then store the results in the destination buffer */
*pDst++ = (q15_t) __SSAT((q31_t) (*pIn++ * 32768.0f), 16);
#endif /* #ifdef ARM_MATH_ROUNDING */
/* Decrement the loop counter */
blkCnt--;
}
#else
/* Run the below code for Cortex-M0 */
/* Loop over blockSize number of values */
blkCnt = blockSize;
while(blkCnt > 0u)
{
#ifdef ARM_MATH_ROUNDING
/* C = A * 32768 */
/* convert from float to q15 and then store the results in the destination buffer */
in = *pIn++;
in = (in * 32768.0f);
in += in > 0 ? 0.5f : -0.5f;
*pDst++ = (q15_t) (__SSAT((q31_t) (in), 16));
#else
/* C = A * 32768 */
/* convert from float to q15 and then store the results in the destination buffer */
*pDst++ = (q15_t) __SSAT((q31_t) (*pIn++ * 32768.0f), 16);
#endif /* #ifdef ARM_MATH_ROUNDING */
/* Decrement the loop counter */
blkCnt--;
}
#endif /* #ifndef ARM_MATH_CM0 */
}
buffer_c16_t FIRAndDecimateComplex::execute(
const buffer_c16_t& src,
const buffer_c16_t& dst
) {
/* int16_t input (sample count "n" must be multiple of decimation_factor)
* -> int16_t output, decimated by decimation_factor.
* taps are normalized to 1 << 16 == 1.0.
*/
const auto output_sampling_rate = src.sampling_rate / decimation_factor_;
const size_t output_samples = src.count / decimation_factor_;
sample_t* dst_p = dst.p;
const buffer_c16_t result { dst.p, output_samples, output_sampling_rate };
const sample_t* src_p = src.p;
size_t outer_count = output_samples;
while(outer_count > 0) {
/* Put new samples into delay buffer */
auto z_new_p = &samples_[taps_count_ - decimation_factor_];
for(size_t i=0; i<decimation_factor_; i++) {
*__SIMD32(z_new_p)++ = *__SIMD32(src_p)++;
}
size_t loop_count = taps_count_ / 8;
auto t_p = &taps_reversed_[0];
auto z_p = &samples_[0];
int64_t t_real = 0;
int64_t t_imag = 0;
while(loop_count > 0) {
const auto tap0 = *__SIMD32(t_p)++;
const auto sample0 = *__SIMD32(z_p)++;
const auto tap1 = *__SIMD32(t_p)++;
const auto sample1 = *__SIMD32(z_p)++;
t_real = __SMLSLD(sample0, tap0, t_real);
t_imag = __SMLALDX(sample0, tap0, t_imag);
t_real = __SMLSLD(sample1, tap1, t_real);
t_imag = __SMLALDX(sample1, tap1, t_imag);
const auto tap2 = *__SIMD32(t_p)++;
const auto sample2 = *__SIMD32(z_p)++;
const auto tap3 = *__SIMD32(t_p)++;
const auto sample3 = *__SIMD32(z_p)++;
t_real = __SMLSLD(sample2, tap2, t_real);
t_imag = __SMLALDX(sample2, tap2, t_imag);
t_real = __SMLSLD(sample3, tap3, t_real);
t_imag = __SMLALDX(sample3, tap3, t_imag);
const auto tap4 = *__SIMD32(t_p)++;
const auto sample4 = *__SIMD32(z_p)++;
const auto tap5 = *__SIMD32(t_p)++;
const auto sample5 = *__SIMD32(z_p)++;
t_real = __SMLSLD(sample4, tap4, t_real);
t_imag = __SMLALDX(sample4, tap4, t_imag);
t_real = __SMLSLD(sample5, tap5, t_real);
t_imag = __SMLALDX(sample5, tap5, t_imag);
const auto tap6 = *__SIMD32(t_p)++;
const auto sample6 = *__SIMD32(z_p)++;
const auto tap7 = *__SIMD32(t_p)++;
const auto sample7 = *__SIMD32(z_p)++;
t_real = __SMLSLD(sample6, tap6, t_real);
t_imag = __SMLALDX(sample6, tap6, t_imag);
t_real = __SMLSLD(sample7, tap7, t_real);
t_imag = __SMLALDX(sample7, tap7, t_imag);
loop_count--;
}
/* TODO: Re-evaluate whether saturation is performed, normalization,
* all that jazz.
*/
const int32_t r = t_real >> 16;
const int32_t i = t_imag >> 16;
const int32_t r_sat = __SSAT(r, 16);
const int32_t i_sat = __SSAT(i, 16);
*__SIMD32(dst_p)++ = __PKHBT(
r_sat,
i_sat,
16
);
/* Shift sample buffer left/down by decimation factor. */
const size_t unroll_factor = 4;
size_t shift_count = (taps_count_ - decimation_factor_) / unroll_factor;
sample_t* t = &samples_[0];
const sample_t* s = &samples_[decimation_factor_];
while(shift_count > 0) {
*__SIMD32(t)++ = *__SIMD32(s)++;
*__SIMD32(t)++ = *__SIMD32(s)++;
*__SIMD32(t)++ = *__SIMD32(s)++;
*__SIMD32(t)++ = *__SIMD32(s)++;
shift_count--;
}
shift_count = (taps_count_ - decimation_factor_) % unroll_factor;
while(shift_count > 0) {
*(t++) = *(s++);
shift_count--;
}
outer_count--;
}
return result;
}