diff options
Diffstat (limited to 'Drivers/CMSIS/DSP/Source/SupportFunctions/arm_q7_to_float.c')
-rw-r--r-- | Drivers/CMSIS/DSP/Source/SupportFunctions/arm_q7_to_float.c | 138 |
1 files changed, 99 insertions, 39 deletions
diff --git a/Drivers/CMSIS/DSP/Source/SupportFunctions/arm_q7_to_float.c b/Drivers/CMSIS/DSP/Source/SupportFunctions/arm_q7_to_float.c index ace437fcf..6bd86bfeb 100644 --- a/Drivers/CMSIS/DSP/Source/SupportFunctions/arm_q7_to_float.c +++ b/Drivers/CMSIS/DSP/Source/SupportFunctions/arm_q7_to_float.c @@ -3,13 +3,13 @@ * Title: arm_q7_to_float.c * Description: Converts the elements of the Q7 vector to floating-point vector * - * $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 * @@ -29,7 +29,7 @@ #include "arm_math.h" /** - * @ingroup groupSupport + @ingroup groupSupport */ /** @@ -37,83 +37,143 @@ */ /** - * @addtogroup q7_to_x - * @{ + @addtogroup q7_to_x + @{ */ /** - * @brief Converts the elements of the Q7 vector to floating-point vector. - * @param[in] *pSrc points to the Q7 input vector - * @param[out] *pDst points to the floating-point output vector - * @param[in] blockSize length of the input vector - * @return none. - * - * \par Description: - * - * The equation used for the conversion process is: - * - * <pre> - * pDst[n] = (float32_t) pSrc[n] / 128; 0 <= n < blockSize. - * </pre> - * + @brief Converts the elements of the Q7 vector to floating-point vector. + @param[in] pSrc points to the Q7 input vector + @param[out] pDst points to the floating-point output vector + @param[in] blockSize number of samples in each vector + @return none + + @par Details + The equation used for the conversion process is: + <pre> + pDst[n] = (float32_t) pSrc[n] / 128; 0 <= n < blockSize. + </pre> */ - +#if defined(ARM_MATH_NEON) void arm_q7_to_float( - q7_t * pSrc, + const q7_t * pSrc, float32_t * pDst, uint32_t blockSize) { - q7_t *pIn = pSrc; /* Src pointer */ + const q7_t *pIn = pSrc; /* Src pointer */ uint32_t blkCnt; /* loop counter */ + int8x16_t inV; + int16x8_t inVLO, inVHI; + int32x4_t inVLL, inVLH, inVHL, inVHH; + float32x4_t outV; + + blkCnt = blockSize >> 4U; + + /* Compute 16 outputs at a time. + ** a second loop below computes the remaining 1 to 15 samples. */ + while (blkCnt > 0U) + { + /* C = (float32_t) A / 128 */ + /* Convert from q7 to float and then store the results in the destination buffer */ + inV = vld1q_s8(pIn); + pIn += 16; + + inVLO = vmovl_s8(vget_low_s8(inV)); + inVHI = vmovl_s8(vget_high_s8(inV)); -#if defined (ARM_MATH_DSP) + inVLL = vmovl_s16(vget_low_s16(inVLO)); + inVLH = vmovl_s16(vget_high_s16(inVLO)); + inVHL = vmovl_s16(vget_low_s16(inVHI)); + inVHH = vmovl_s16(vget_high_s16(inVHI)); - /* Run the below code for Cortex-M4 and Cortex-M3 */ + outV = vcvtq_n_f32_s32(inVLL,7); + vst1q_f32(pDst, outV); + pDst += 4; - /*loop Unrolling */ + outV = vcvtq_n_f32_s32(inVLH,7); + vst1q_f32(pDst, outV); + pDst += 4; + + outV = vcvtq_n_f32_s32(inVHL,7); + vst1q_f32(pDst, outV); + pDst += 4; + + outV = vcvtq_n_f32_s32(inVHH,7); + vst1q_f32(pDst, outV); + pDst += 4; + + /* 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 & 0xF; + + while (blkCnt > 0U) + { + /* C = (float32_t) A / 128 */ + /* Convert from q7 to float and then store the results in the destination buffer */ + *pDst++ = ((float32_t) * pIn++ / 128.0f); + + /* Decrement the loop counter */ + blkCnt--; + } +} +#else +void arm_q7_to_float( + const q7_t * pSrc, + float32_t * pDst, + uint32_t blockSize) +{ + uint32_t blkCnt; /* Loop counter */ + const q7_t *pIn = pSrc; /* Source pointer */ + +#if defined (ARM_MATH_LOOPUNROLL) + + /* Loop unrolling: Compute 4 outputs at a time */ 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 = (float32_t) A / 128 */ - /* convert from q7 to float and then store the results in the destination buffer */ + + /* Convert from q7 to float and store result in destination buffer */ *pDst++ = ((float32_t) * pIn++ / 128.0f); *pDst++ = ((float32_t) * pIn++ / 128.0f); *pDst++ = ((float32_t) * pIn++ / 128.0f); *pDst++ = ((float32_t) * pIn++ / 128.0f); - /* Decrement the loop counter */ + /* Decrement loop counter */ blkCnt--; } - /* If the blockSize is not a multiple of 4, compute any remaining output samples here. - ** No loop unrolling is used. */ + /* Loop unrolling: Compute remaining outputs */ blkCnt = blockSize % 0x4U; #else - /* Run the below code for Cortex-M0 */ - - /* Loop over blockSize number of values */ + /* Initialize blkCnt with number of samples */ blkCnt = blockSize; -#endif /* #if defined (ARM_MATH_DSP) */ +#endif /* #if defined (ARM_MATH_LOOPUNROLL) */ while (blkCnt > 0U) { /* C = (float32_t) A / 128 */ - /* convert from q7 to float and then store the results in the destination buffer */ + + /* Convert from q7 to float and store result in destination buffer */ *pDst++ = ((float32_t) * pIn++ / 128.0f); - /* Decrement the loop counter */ + /* Decrement loop counter */ blkCnt--; } + } +#endif /* #if defined(ARM_MATH_NEON) */ /** - * @} end of q7_to_x group + @} end of q7_to_x group */ |