1 /* ----------------------------------------------------------------------
2 * Copyright (C) 2010 ARM Limited. All rights reserved.
7 * Project: CMSIS DSP Library
8 * Title: arm_dot_prod_q7.c
10 * Description: Q7 dot product.
12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
14 * Version 1.0.10 2011/7/15
15 * Big Endian support added and Merged M0 and M3/M4 Source code.
17 * Version 1.0.3 2010/11/29
18 * Re-organized the CMSIS folders and updated documentation.
20 * Version 1.0.2 2010/11/11
21 * Documentation updated.
23 * Version 1.0.1 2010/10/05
24 * Production release and review comments incorporated.
26 * Version 1.0.0 2010/09/20
27 * Production release and review comments incorporated.
29 * Version 0.0.7 2010/06/10
30 * Misra-C changes done
31 * -------------------------------------------------------------------- */
40 * @addtogroup dot_prod
45 * @brief Dot product of Q7 vectors.
46 * @param[in] *pSrcA points to the first input vector
47 * @param[in] *pSrcB points to the second input vector
48 * @param[in] blockSize number of samples in each vector
49 * @param[out] *result output result returned here
52 * <b>Scaling and Overflow Behavior:</b>
54 * The intermediate multiplications are in 1.7 x 1.7 = 2.14 format and these
55 * results are added to an accumulator in 18.14 format.
56 * Nonsaturating additions are used and there is no danger of wrap around as long as
57 * the vectors are less than 2^18 elements long.
58 * The return result is in 18.14 format.
67 uint32_t blkCnt; /* loop counter */
69 q31_t sum = 0; /* Temporary variables to store output */
73 /* Run the below code for Cortex-M4 and Cortex-M3 */
75 q31_t input1, input2; /* Temporary variables to store input */
76 q15_t in1, in2; /* Temporary variables to store input */
81 blkCnt = blockSize >> 2u;
83 /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
84 ** a second loop below computes the remaining 1 to 3 samples. */
87 /* Reading two inputs of SrcA buffer and packing */
88 in1 = (q15_t) * pSrcA++;
89 in2 = (q15_t) * pSrcA++;
90 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
92 /* Reading two inputs of SrcB buffer and packing */
93 in1 = (q15_t) * pSrcB++;
94 in2 = (q15_t) * pSrcB++;
95 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
97 /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
98 /* Perform Dot product of 2 packed inputs using SMLALD and store the result in a temporary variable. */
99 sum = __SMLAD(input1, input2, sum);
101 /* Reading two inputs of SrcA buffer and packing */
102 in1 = (q15_t) * pSrcA++;
103 in2 = (q15_t) * pSrcA++;
104 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
106 /* Reading two inputs of SrcB buffer and packing */
107 in1 = (q15_t) * pSrcB++;
108 in2 = (q15_t) * pSrcB++;
109 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
111 /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
112 /* Perform Dot product of 2 packed inputs using SMLALD and store the result in a temporary variable. */
113 sum = __SMLAD(input1, input2, sum);
117 /* Decrement the loop counter */
121 /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
122 ** No loop unrolling is used. */
123 blkCnt = blockSize % 0x4u;
127 /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
128 /* Dot product and then store the results in a temporary buffer. */
129 sum = __SMLAD(*pSrcA++, *pSrcB++, sum);
131 /* Decrement the loop counter */
137 /* Run the below code for Cortex-M0 */
141 /* Initialize blkCnt with number of samples */
146 /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
147 /* Dot product and then store the results in a temporary buffer. */
148 sum += (q31_t) ((q15_t) * pSrcA++ * *pSrcB++);
150 /* Decrement the loop counter */
154 #endif /* #ifndef ARM_MATH_CM0 */
157 /* Store the result in the destination buffer in 18.14 format */
162 * @} end of dot_prod group