24.04
a64_sgemv_pretransposed.hpp
Go to the documentation of this file.
1
/*
2
* Copyright (c) 2017,2020 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
#pragma once
25
26
#ifdef __aarch64__
27
28
namespace
arm_gemm
{
29
30
// Actual kernel implementations
31
void
a64_sgemv_pretransposed(
const
float
*,
int
,
const
float
*,
float
*,
float
,
int
,
int
);
32
33
// Pretransposed SGEMV strategy class.
34
class
sgemv_pretransposed {
35
public
:
36
typedef
float
operand_type;
37
typedef
float
result_type;
38
39
typedef
void (*kern_type)(
const
float
*, int,
const
float
*,
float
*, float, int, int);
40
41
/* Describes the data layout for matrix (A) input */
42
43
/* Note that often GEMV is expressed as a GEMM with M=1, i.e. A is the
44
* (row) vector and B is the matrix, but the standard GEMV arrangement
45
* is matrix A times (column) vector X. "A_transpose" is expressed in
46
* terms of this standard arrangement, so if the A matrix is in fact the
47
* B matrix from a GEMM call, the sense of the transpose needs to be
48
* reversed. */
49
static
constexpr
unsigned
int
A_interleave() {
50
return
32;
51
}
52
53
static
constexpr
unsigned
int
A_block() {
54
return
1;
55
}
56
57
static
constexpr
bool
A_transpose() {
58
return
false
;
59
}
60
61
/* Kernel blocking parameters */
62
static
constexpr
unsigned
int
out_width() {
63
return
32;
64
}
65
66
static
constexpr
unsigned
int
k_unroll() {
67
return
1;
68
}
69
70
kern_type kernel = a64_sgemv_pretransposed;
71
72
sgemv_pretransposed(
const
CPUInfo
*) { }
73
};
74
75
}
// namespace arm_gemm
76
77
#endif // __aarch64__
arm_compute::CPUInfo
Definition:
CPPTypes.h:66
arm_gemm
Definition:
barrier.hpp:30
src
core
NEON
kernels
arm_gemm
kernels
a64_sgemv_pretransposed.hpp
Generated on Mon Apr 29 2024 10:53:56 for Compute Library by
1.8.17