1 /* ----------------------------------------------------------------------
2 * Copyright (C) 2010-2013 ARM Limited. All rights reserved.
4 * $Date: 17. January 2013
7 * Project: CMSIS DSP Library
8 * Title: arm_mult_q15.c
10 * Description: Q15 vector multiplication.
12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
14 * Redistribution and use in source and binary forms, with or without
15 * modification, are permitted provided that the following conditions
17 * - Redistributions of source code must retain the above copyright
18 * notice, this list of conditions and the following disclaimer.
19 * - Redistributions in binary form must reproduce the above copyright
20 * notice, this list of conditions and the following disclaimer in
21 * the documentation and/or other materials provided with the
23 * - Neither the name of ARM LIMITED nor the names of its contributors
24 * may be used to endorse or promote products derived from this
25 * software without specific prior written permission.
27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
30 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
31 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
32 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
33 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
34 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
35 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
37 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
38 * POSSIBILITY OF SUCH DAMAGE.
39 * -------------------------------------------------------------------- */
48 * @addtogroup BasicMult
54 * @brief Q15 vector multiplication
55 * @param[in] *pSrcA points to the first input vector
56 * @param[in] *pSrcB points to the second input vector
57 * @param[out] *pDst points to the output vector
58 * @param[in] blockSize number of samples in each vector
61 * <b>Scaling and Overflow Behavior:</b>
63 * The function uses saturating arithmetic.
64 * Results outside of the allowable Q15 range [0x8000 0x7FFF] will be saturated.
73 uint32_t blkCnt; /* loop counters */
75 #ifndef ARM_MATH_CM0_FAMILY
77 /* Run the below code for Cortex-M4 and Cortex-M3 */
78 q31_t inA1, inA2, inB1, inB2; /* temporary input variables */
79 q15_t out1, out2, out3, out4; /* temporary output variables */
80 q31_t mul1, mul2, mul3, mul4; /* temporary variables */
83 blkCnt = blockSize >> 2u;
85 /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
86 ** a second loop below computes the remaining 1 to 3 samples. */
89 /* read two samples at a time from sourceA */
90 inA1 = *__SIMD32(pSrcA)++;
91 /* read two samples at a time from sourceB */
92 inB1 = *__SIMD32(pSrcB)++;
93 /* read two samples at a time from sourceA */
94 inA2 = *__SIMD32(pSrcA)++;
95 /* read two samples at a time from sourceB */
96 inB2 = *__SIMD32(pSrcB)++;
98 /* multiply mul = sourceA * sourceB */
99 mul1 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
100 mul2 = (q31_t) ((q15_t) inA1 * (q15_t) inB1);
101 mul3 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB2 >> 16));
102 mul4 = (q31_t) ((q15_t) inA2 * (q15_t) inB2);
104 /* saturate result to 16 bit */
105 out1 = (q15_t) __SSAT(mul1 >> 15, 16);
106 out2 = (q15_t) __SSAT(mul2 >> 15, 16);
107 out3 = (q15_t) __SSAT(mul3 >> 15, 16);
108 out4 = (q15_t) __SSAT(mul4 >> 15, 16);
110 /* store the result */
111 #ifndef ARM_MATH_BIG_ENDIAN
113 *__SIMD32(pDst)++ = __PKHBT(out2, out1, 16);
114 *__SIMD32(pDst)++ = __PKHBT(out4, out3, 16);
118 *__SIMD32(pDst)++ = __PKHBT(out2, out1, 16);
119 *__SIMD32(pDst)++ = __PKHBT(out4, out3, 16);
121 #endif // #ifndef ARM_MATH_BIG_ENDIAN
123 /* Decrement the blockSize loop counter */
127 /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
128 ** No loop unrolling is used. */
129 blkCnt = blockSize % 0x4u;
133 /* Run the below code for Cortex-M0 */
135 /* Initialize blkCnt with number of samples */
138 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
144 /* Multiply the inputs and store the result in the destination buffer */
145 *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
147 /* Decrement the blockSize loop counter */
153 * @} end of BasicMult group