1 /*
2 * Copyright (C) 2010-2020 Arm Limited or its affiliates. All rights reserved.
3 *
4 * SPDX-License-Identifier: Apache-2.0
5 *
6 * Licensed under the Apache License, Version 2.0 (the License); you may
7 * not use this file except in compliance with the License.
8 * You may obtain a copy of the License at
9 *
10 * www.apache.org/licenses/LICENSE-2.0
11 *
12 * Unless required by applicable law or agreed to in writing, software
13 * distributed under the License is distributed on an AS IS BASIS, WITHOUT
14 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15 * See the License for the specific language governing permissions and
16 * limitations under the License.
17 */
18
19 /* ----------------------------------------------------------------------
20 * Project: CMSIS NN Library
21 * Title: arm_q7_to_q15_reordered_with_offset.c
22 * Description: Converts the elements of the Q7 vector to a reordered Q15 vector with an added offset. The re-ordering
23 * is a signature of sign extension intrinsic(DSP extension).
24 *
25 * $Date: May 29, 2020
26 * $Revision: V.2.0.3
27 *
28 * Target Processor: Cortex-M cores
29 *
30 * -------------------------------------------------------------------- */
31
32 #include "arm_nnsupportfunctions.h"
33
34 /**
35 * @ingroup groupSupport
36 */
37
38 /**
39 * @addtogroup nndata_convert
40 * @{
41 */
42
43 /**
44 * @brief Converts the elements of the Q7 vector to a reordered Q15 vector with an added offset.
45 *
46 * @note Refer header file for details.
47 *
48 */
49
arm_q7_to_q15_reordered_with_offset(const q7_t * src,q15_t * dst,uint32_t block_size,q15_t offset)50 void arm_q7_to_q15_reordered_with_offset(const q7_t *src, q15_t *dst, uint32_t block_size, q15_t offset)
51 {
52
53 #if defined(ARM_MATH_DSP)
54 uint32_t block_cnt;
55 /* Run the below code for cores that support SIMD instructions */
56 q31_t in_q7x4;
57 q31_t out_q15x2_1;
58 q31_t out_q15x2_2;
59
60 /*loop unrolling */
61 block_cnt = block_size >> 2u;
62
63 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. */
64 const q31_t offset_q15x2 = (q31_t)__PKHBT(offset, offset, 16);
65 while (block_cnt > 0u)
66 {
67 /* convert from q7 to q15 and then store the results in the destination buffer */
68 in_q7x4 = arm_nn_read_q7x4_ia(&src);
69
70 /* Extract and sign extend each of the four q7 values to q15 */
71 out_q15x2_1 = __SXTAB16(offset_q15x2, __ROR((uint32_t)in_q7x4, 8));
72 out_q15x2_2 = __SXTAB16(offset_q15x2, in_q7x4);
73
74 arm_nn_write_q15x2_ia(&dst, out_q15x2_2);
75 arm_nn_write_q15x2_ia(&dst, out_q15x2_1);
76
77 block_cnt--;
78 }
79 /* Handle left over samples */
80 block_cnt = block_size % 0x4u;
81
82 while (block_cnt > 0u)
83 {
84 *dst++ = (q15_t)*src++ + offset;
85
86 /* Decrement the loop counter */
87 block_cnt--;
88 }
89 #else
90 (void)src;
91 (void)dst;
92 (void)block_size;
93 (void)offset;
94 /* Not available */
95 #endif
96 }
97
98 /**
99 * @} end of nndata_convert group
100 */
101
