xref: /aosp_15_r20/external/ComputeLibrary/src/cpu/kernels/CpuGemmInterleave4x4Kernel.h (revision c217d954acce2dbc11938adb493fc0abd69584f3)
1 /*
2  * Copyright (c) 2016-2022 Arm Limited.
3  *
4  * SPDX-License-Identifier: MIT
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to
8  * deal in the Software without restriction, including without limitation the
9  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10  * sell copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in all
14  * copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
24 #ifndef ARM_COMPUTE_CPU_GEMM_INTERLEAVE4x4_KERNEL_H
25 #define ARM_COMPUTE_CPU_GEMM_INTERLEAVE4x4_KERNEL_H
26 
27 #include "src/core/common/Macros.h"
28 #include "src/cpu/ICpuKernel.h"
29 
30 namespace arm_compute
31 {
32 namespace cpu
33 {
34 namespace kernels
35 {
36 /** Kernel to interleave the elements of a matrix
37  *
38  * This function puts the values in a 4x4 block of Matrix A on the same row (Interleaved values)
39  *
40  * @f[
41  * \left( \begin{array}{cccc}
42  * a00 & a01 & a02 & a03 \\
43  * a10 & a11 & a12 & a13 \\
44  * a20 & a21 & a22 & a23 \\
45  * a30 & a31 & a32 & a33 \\
46  * \end{array} \right)
47  * \rightarrow
48  * \left( \begin{array}{ccccccccccccccccc}
49  * a00 & a10 & a20 & a30 & a01 & a11 & a21 & a31 & a02 & a12 & a22 & a32 & a03 & a13 & a23 & a33 \\
50  * \end{array} \right)
51  * @f]
52  *
53  * After this operation, the dst matrix will have the following shape: [ height * 4, ceil(width / 4.0f) ]
54  */
55 class CpuGemmInterleave4x4Kernel : public ICpuKernel<CpuGemmInterleave4x4Kernel>
56 {
57 public:
58     CpuGemmInterleave4x4Kernel() = default;
59     /** Initialise the kernel's src and dst.
60      *
61      * @param[in]  src Input tensor info. Data types supported: All
62      * @param[out] dst Output tensor info which stores the interleaved matrix. Data type supported: same as @p src.
63      */
64     void configure(const ITensorInfo *src, ITensorInfo *dst);
65     /** Static function to check if given info will lead to a valid configuration of @ref CpuGemmInterleave4x4Kernel
66      *
67      * Similar to @ref CpuGemmInterleave4x4Kernel::configure()
68      *
69      * @return a status
70      */
71     static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
72 
73     // Inherited methods overridden:
74     void run_op(ITensorPack &tensors, const Window &window, const ThreadInfo &info) override;
75     const char *name() const override;
76 };
77 } // namespace kernels
78 } // namespace cpu
79 } // namespace arm_compute
80 #endif /* ARM_COMPUTE_CPU_GEMM_INTERLEAVE4x4_KERNEL_H */
81