Anthony Barbier | 6ff3b19 | 2017-09-04 18:44:23 +0100 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (c) 2017 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_CLGEMMTRANSPOSE1XWKERNEL_H__ |
| 25 | #define __ARM_COMPUTE_CLGEMMTRANSPOSE1XWKERNEL_H__ |
| 26 | |
| 27 | #include "arm_compute/core/CL/ICLSimple2DKernel.h" |
| 28 | |
| 29 | namespace arm_compute |
| 30 | { |
| 31 | class ICLTensor; |
| 32 | |
Gian Marco Iodice | 9f89bae | 2017-06-22 12:09:49 +0100 | [diff] [blame] | 33 | /** OpenCL kernel which transposes the elements of a matrix in chunks of 1xW, where W is equal to (16 / element size of the tensor) |
Anthony Barbier | 6ff3b19 | 2017-09-04 18:44:23 +0100 | [diff] [blame] | 34 | * |
| 35 | * Following an example of how the transposition1xW works when the input data type is F32 |
| 36 | * |
| 37 | * @f[ |
| 38 | * \left( \begin{array}{cccc} |
| 39 | * a00 & a01 & a02 & a03 \\ |
| 40 | * a10 & a11 & a12 & a13 \\ |
| 41 | * a20 & a21 & a22 & a23 \\ |
| 42 | * a30 & a31 & a32 & a33 \\ |
| 43 | * \end{array} \right) |
| 44 | * \rightarrow |
| 45 | * \left( \begin{array}{ccccccccccccccccc} |
| 46 | * a00 & a01 & a02 & a03 & a10 & a11 & a12 & a13 & a20 & a21 & a22 & a23 & a30 & a31 & a32 & a33 \\ |
| 47 | * \end{array} \right) |
| 48 | * @f] |
| 49 | * |
| 50 | * Following an example of how the transposition1xW works when the input data type is F16 |
| 51 | * |
| 52 | * @f[ |
| 53 | * \left( \begin{array}{cccccccc} |
| 54 | * a00 & a01 & a02 & a03 & a04 & a05 & a06 & a7 \\ |
| 55 | * a10 & a11 & a12 & a13 & a14 & a15 & a16 & 17 \\ |
| 56 | * a20 & a21 & a22 & a23 & a24 & a25 & a26 & 27 \\ |
| 57 | * a30 & a31 & a32 & a33 & a34 & a35 & a36 & 37 \\ |
| 58 | * \end{array} \right) |
| 59 | * \rightarrow |
| 60 | * \left( \begin{array}{cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc} |
| 61 | * a00 & a01 & a02 & a03 & a04 & a05 & a06 & a07 & a10 & a11 & a12 & a13 & a14 & a15 & a16 & a17 & a20 & a21 & a22 & a23 & a24 & a25 & a26 & a27 & a30 & a31 & a32 & a33 & a34 & a35 & a36 & a37\\ |
| 62 | * \end{array} \right) |
| 63 | * @f] |
| 64 | * |
Gian Marco Iodice | 9f89bae | 2017-06-22 12:09:49 +0100 | [diff] [blame] | 65 | * @note The output matrix will have the following shape: [ height * W, ceil(width / W) ], where W = (16 / element size of the tensor) |
Anthony Barbier | 6ff3b19 | 2017-09-04 18:44:23 +0100 | [diff] [blame] | 66 | * |
| 67 | */ |
| 68 | class CLGEMMTranspose1xWKernel : public ICLSimple2DKernel |
| 69 | { |
| 70 | public: |
| 71 | /** Initialise the kernel's input and output. |
| 72 | * |
Gian Marco Iodice | 8a38369 | 2017-07-03 17:41:47 +0100 | [diff] [blame^] | 73 | * @param[in] input Input tensor. Data types supported: U8/S8/QS8/U16/S16/QS16/F16/U32/S32/F32 |
Anthony Barbier | 6ff3b19 | 2017-09-04 18:44:23 +0100 | [diff] [blame] | 74 | * @param[out] output Output tensor. Data type supported: same as @p input |
| 75 | */ |
| 76 | void configure(const ICLTensor *input, ICLTensor *output); |
| 77 | |
| 78 | // Inherited methods overridden: |
| 79 | void run(const Window &window, cl::CommandQueue &queue) override; |
| 80 | }; |
| 81 | } |
| 82 | #endif /* __ARM_COMPUTE_CLGEMMTRANSPOSE1XWKERNEL_H__ */ |