File size: 2,843 Bytes
be903e2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 | // Tencent is pleased to support the open source community by making ncnn available.
//
// Copyright (C) 2020 THL A29 Limited, a Tencent company. All rights reserved.
//
// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
// in compliance with the License. You may obtain a copy of the License at
//
// https://opensource.org/licenses/BSD-3-Clause
//
// Unless required by applicable law or agreed to in writing, software distributed
// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
// CONDITIONS OF ANY KIND, either express or implied. See the License for the
// specific language governing permissions and limitations under the License.
#version 450
#if NCNN_fp16_storage
#extension GL_EXT_shader_16bit_storage: require
struct sfpvec8 { f16vec4 abcd; f16vec4 efgh; };
#endif
#if NCNN_fp16_arithmetic
#extension GL_EXT_shader_explicit_arithmetic_types_float16: require
#endif
#define shape_constant_id_offset 0
layout (constant_id = shape_constant_id_offset + 0) const int dims = 0;
layout (constant_id = shape_constant_id_offset + 1) const int w = 0;
layout (constant_id = shape_constant_id_offset + 2) const int h = 0;
layout (constant_id = shape_constant_id_offset + 3) const int c = 0;
layout (constant_id = shape_constant_id_offset + 4) const int cstep = 0;
#if NCNN_image_shader
layout (binding = 0) uniform unfp sampler3D bottom_blob_3d;
layout (binding = 1, imfmtc4) writeonly uniform unfp image3D top_blob_3d;
layout (binding = 2) uniform unfp sampler3D a;
layout (binding = 3) uniform unfp sampler3D b;
#else
layout (binding = 0) buffer bottom_top_blob { sfpvec8 bottom_top_blob_data[]; };
layout (binding = 1) readonly buffer a { sfpvec8 a_data[]; };
layout (binding = 2) readonly buffer b { sfpvec8 b_data[]; };
#endif
layout (push_constant) uniform parameter
{
int dims;
int w;
int h;
int c;
int cstep;
} p;
void main()
{
int gx = int(gl_GlobalInvocationID.x);
int gy = int(gl_GlobalInvocationID.y);
int gz = int(gl_GlobalInvocationID.z);
if (gx >= psc(w) || gy >= psc(h) || gz >= psc(c))
return;
#if NCNN_image_shader
afpvec8 v = image3d_ld8(bottom_blob_3d, ivec3(gx, gy, gz));
#else
const int gi = gz * psc(cstep) + gy * psc(w) + gx;
afpvec8 v = buffer_ld8(bottom_top_blob_data, gi);
#endif
const ivec3 gxyz = ivec3(gx, gy, gz);
const int abi = gxyz[psc(dims) - 1];
#if NCNN_image_shader
afpvec8 b = image3d_ld8(b, ivec3(abi, 0, 0));
afpvec8 a = image3d_ld8(a, ivec3(abi, 0, 0));
#else
afpvec8 b = buffer_ld8(b_data, abi);
afpvec8 a = buffer_ld8(a_data, abi);
#endif
v[0] = b[0] * v[0] + a[0];
v[1] = b[1] * v[1] + a[1];
#if NCNN_image_shader
image3d_st8(top_blob_3d, ivec3(gx, gy, gz), v);
#else
buffer_st8(bottom_top_blob_data, gi, v);
#endif
}
|