ncnn/src/layer/vulkan/shader/convolutiondepthwise_group_pack8.comp

// 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

layout (constant_id = 0) const int kernel_w = 1;
layout (constant_id = 1) const int kernel_h = 1;
layout (constant_id = 2) const int dilation_w = 1;
layout (constant_id = 3) const int dilation_h = 1;
layout (constant_id = 4) const int stride_w = 1;
layout (constant_id = 5) const int stride_h = 1;
layout (constant_id = 6) const int bias_term = 0;
layout (constant_id = 7) const int group = 1;
layout (constant_id = 8) const int activation_type = 0;
layout (constant_id = 9) const float activation_param_0 = 0;
layout (constant_id = 10) const float activation_param_1 = 0;

#define shape_constant_id_offset 11
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;

layout (constant_id = shape_constant_id_offset + 5) const int outdims = 0;
layout (constant_id = shape_constant_id_offset + 6) const int outw = 0;
layout (constant_id = shape_constant_id_offset + 7) const int outh = 0;
layout (constant_id = shape_constant_id_offset + 8) const int outc = 0;
layout (constant_id = shape_constant_id_offset + 9) const int outcstep = 0;

layout (local_size_x_id = 233) in;
layout (local_size_y_id = 234) in;
layout (local_size_z_id = 235) in;

#if NCNN_image_shader
layout (binding = 0) uniform unfp sampler3D bottom_blob;
layout (binding = 1, imfmtc4) writeonly uniform unfp image3D top_blob;
layout (binding = 2) uniform unfp sampler3D weight_blob;
layout (binding = 3) uniform unfp sampler1D bias_blob;
#else // NCNN_image_shader
layout (binding = 0) readonly buffer bottom_blob { sfpvec8 bottom_blob_data[]; };
layout (binding = 1) writeonly buffer top_blob { sfpvec8 top_blob_data[]; };
layout (binding = 2) readonly buffer weight_blob { sfpvec8 weight_data[]; };
layout (binding = 3) readonly buffer bias_blob { sfpvec8 bias_data[]; };
#endif // NCNN_image_shader

layout (push_constant) uniform parameter
{
    int dims;
    int w;
    int h;
    int c;
    int cstep;

    int outdims;
    int outw;
    int outh;
    int outc;
    int outcstep;
} p;

void main()
{
    int gx = int(gl_GlobalInvocationID.x);
    int gy = int(gl_GlobalInvocationID.y);
    int gz = int(gl_GlobalInvocationID.z);

    if (gx >= psc(outw) || gy >= psc(outh) || gz >= psc(outc))
        return;

    afpvec8 sum;

    if (bias_term == 1)
    {
#if NCNN_image_shader
        sum = image1d_ld8(bias_blob, gz);
#else
        sum = buffer_ld8(bias_data, gz);
#endif
    }
    else
    {
        sum = afpvec8(afpvec4(0.f), afpvec4(0.f));
    }

    // group convolution
    const int channels_g = psc(c) / group;
    const int num_output_g = psc(outc) / group;

    // group id
    const int gg = gz / num_output_g;

#if NCNN_image_shader
    int sz = gg * channels_g;

    for (int z = 0; z < channels_g; z++)
    {
        int sy = gy * stride_h;
        int wx = 0;

        for (int y = 0; y < kernel_h; y++)
        {
            int sx = gx * stride_w;

            for (int x = 0; x < kernel_w; x++)
            {
                afpvec8 v = image3d_ld8(bottom_blob, ivec3(sx, sy, sz));

                afpvec8 k0 = image3d_ld8(weight_blob, ivec3(wx + 0, z, gz));
                afpvec8 k1 = image3d_ld8(weight_blob, ivec3(wx + 1, z, gz));
                afpvec8 k2 = image3d_ld8(weight_blob, ivec3(wx + 2, z, gz));
                afpvec8 k3 = image3d_ld8(weight_blob, ivec3(wx + 3, z, gz));
                afpvec8 k4 = image3d_ld8(weight_blob, ivec3(wx + 4, z, gz));
                afpvec8 k5 = image3d_ld8(weight_blob, ivec3(wx + 5, z, gz));
                afpvec8 k6 = image3d_ld8(weight_blob, ivec3(wx + 6, z, gz));
                afpvec8 k7 = image3d_ld8(weight_blob, ivec3(wx + 7, z, gz));

                // sum += v * k
                sum[0].r += dot(v[0], k0[0]) + dot(v[1], k0[1]);
                sum[0].g += dot(v[0], k1[0]) + dot(v[1], k1[1]);
                sum[0].b += dot(v[0], k2[0]) + dot(v[1], k2[1]);
                sum[0].a += dot(v[0], k3[0]) + dot(v[1], k3[1]);
                sum[1].r += dot(v[0], k4[0]) + dot(v[1], k4[1]);
                sum[1].g += dot(v[0], k5[0]) + dot(v[1], k5[1]);
                sum[1].b += dot(v[0], k6[0]) + dot(v[1], k6[1]);
                sum[1].a += dot(v[0], k7[0]) + dot(v[1], k7[1]);

                sx += dilation_w;
                wx += 8;
            }

            sy += dilation_h;
        }

        sz += 1;
    }
#else // NCNN_image_shader
    int w_offset = gz * channels_g * kernel_w * kernel_h;
    int v_offset_0 = gg * channels_g * psc(cstep);

    for (int z = 0; z < channels_g; z++)
    {
        int v_offset = v_offset_0 + gy * stride_h * psc(w) + gx * stride_w;

        for (int y = 0; y < kernel_h; y++)
        {
            for (int x = 0; x < kernel_w; x++)
            {
                afpvec8 v = buffer_ld8(bottom_blob_data, v_offset + x * dilation_w);

                afpvec8 k0 = buffer_ld8(weight_data, (w_offset + x) * 8 + 0);
                afpvec8 k1 = buffer_ld8(weight_data, (w_offset + x) * 8 + 1);
                afpvec8 k2 = buffer_ld8(weight_data, (w_offset + x) * 8 + 2);
                afpvec8 k3 = buffer_ld8(weight_data, (w_offset + x) * 8 + 3);
                afpvec8 k4 = buffer_ld8(weight_data, (w_offset + x) * 8 + 4);
                afpvec8 k5 = buffer_ld8(weight_data, (w_offset + x) * 8 + 5);
                afpvec8 k6 = buffer_ld8(weight_data, (w_offset + x) * 8 + 6);
                afpvec8 k7 = buffer_ld8(weight_data, (w_offset + x) * 8 + 7);

                // sum += v * k
                sum[0].r += dot(v[0], k0[0]) + dot(v[1], k0[1]);
                sum[0].g += dot(v[0], k1[0]) + dot(v[1], k1[1]);
                sum[0].b += dot(v[0], k2[0]) + dot(v[1], k2[1]);
                sum[0].a += dot(v[0], k3[0]) + dot(v[1], k3[1]);
                sum[1].r += dot(v[0], k4[0]) + dot(v[1], k4[1]);
                sum[1].g += dot(v[0], k5[0]) + dot(v[1], k5[1]);
                sum[1].b += dot(v[0], k6[0]) + dot(v[1], k6[1]);
                sum[1].a += dot(v[0], k7[0]) + dot(v[1], k7[1]);
            }

            v_offset += dilation_h * psc(w);
            w_offset += kernel_w;
        }

        v_offset_0 += psc(cstep);
    }
#endif // NCNN_image_shader

    if (activation_type == 1)
    {
        sum[0] = max(sum[0], afp(0.f));
        sum[1] = max(sum[1], afp(0.f));
    }
    if (activation_type == 2)
    {
        const afp slope = afp(activation_param_0);
        sum[0] = mix(sum[0], sum[0] * afp(slope), lessThan(sum[0], afpvec4(0.f)));
        sum[1] = mix(sum[1], sum[1] * afp(slope), lessThan(sum[1], afpvec4(0.f)));
    }
    if (activation_type == 3)
    {
        const afp const_min = afp(activation_param_0);
        const afp const_max = afp(activation_param_1);
        sum[0] = clamp(sum[0], const_min, const_max);
        sum[1] = clamp(sum[1], const_min, const_max);
    }
    if (activation_type == 4)
    {
        sum[0] = afp(1.f) / (afp(1.f) + exp(-sum[0]));
        sum[1] = afp(1.f) / (afp(1.f) + exp(-sum[1]));
    }

#if NCNN_image_shader
    image3d_st8(top_blob, ivec3(gx, gy, gz), sum);
#else
    const int gi = gz * psc(outcstep) + gy * psc(outw) + gx;

    buffer_st8(top_blob_data, gi, sum);
#endif
}