diff options
| author | Gian Marco <gianmarco.iodice@arm.com> | 2018-01-29 12:15:32 +0000 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:45:42 +0000 |
| commit | 76faef88284e6fd51f53b23063374d3d3a884e4f (patch) | |
| tree | d29dfaa733edc7928125ade473b278e1e4923b6e /src/core/CL/cl_kernels/fixed_point.h | |
| parent | b28f329d0cdc2e5f2f72fd3ecc1913c6b913aee8 (diff) | |
| download | ComputeLibrary-76faef88284e6fd51f53b23063374d3d3a884e4f.tar.gz | |
COMPMID-855 - Optimizing im2col on OpenCL (DCHW)
Introduced optimizations for 1x1, 3x3, 5x5 and 11x11
Change-Id: Ibb7f7a9fbec01a7684746ed8513634078126e452
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/118107
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Diffstat (limited to 'src/core/CL/cl_kernels/fixed_point.h')
| -rw-r--r-- | src/core/CL/cl_kernels/fixed_point.h | 18 |
1 files changed, 9 insertions, 9 deletions
diff --git a/src/core/CL/cl_kernels/fixed_point.h b/src/core/CL/cl_kernels/fixed_point.h index d55346b532..46fa645c2b 100644 --- a/src/core/CL/cl_kernels/fixed_point.h +++ b/src/core/CL/cl_kernels/fixed_point.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017 ARM Limited. + * Copyright (c) 2017-2018 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -476,19 +476,19 @@ TANHQ_IMPL(qs16, qs16x8, 8) #define floatx16 float16 #define float16_TYPE float16 -#define CONVERTQ_DOWN_IMPL(in_type, out_type) \ - inline out_type convert_##out_type##_##in_type(in_type a, int fixed_point_position) \ - { \ - return CONVERT(a * (1 << fixed_point_position) + select((in_type)-0.5, (in_type)0.5, isgreater(a, (in_type)0)), out_type); \ +#define CONVERTQ_DOWN_IMPL(in_type, out_type) \ + inline out_type convert_##out_type##_##in_type(in_type a, int fixed_point_position) \ + { \ + return CONVERT(a * (1 << fixed_point_position) + select((in_type)-0.5f, (in_type)0.5f, isgreater(a, (in_type)0)), out_type); \ } CONVERTQ_DOWN_IMPL(float16, qs8x16) CONVERTQ_DOWN_IMPL(float16, qs16x16) -#define CONVERTQ_DOWN_SAT_IMPL(in_type, out_type) \ - inline out_type convert_##out_type##_##in_type##_sat(in_type a, int fixed_point_position) \ - { \ - return CONVERT_SAT(a * (1 << fixed_point_position) + select((in_type)-0.5, (in_type)0.5, isgreater(a, (in_type)0)), out_type); \ +#define CONVERTQ_DOWN_SAT_IMPL(in_type, out_type) \ + inline out_type convert_##out_type##_##in_type##_sat(in_type a, int fixed_point_position) \ + { \ + return CONVERT_SAT(a * (1 << fixed_point_position) + select((in_type)-0.5f, (in_type)0.5f, isgreater(a, (in_type)0)), out_type); \ } CONVERTQ_DOWN_SAT_IMPL(float16, qs8x16) |