mirror of
https://github.com/opencv/opencv.git
synced 2026-07-29 15:23:05 +04:00
WuZhiwen
3d44e9ad92
* dnn/Vulkan: fix GPU hang for heavy convolution tasks Intel i915 driver will declare GPU hang if the compute shader takes too long to complete. See https://bugs.freedesktop.org/show_bug.cgi?id=108947 for details. The idea in this commit is to divide heavy task into several light ones and run compute shader multiple times to make each run take short time enough. TODO: Add more efficient compute shader Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: add a more efficient conv shader
77 lines
2.2 KiB
Plaintext
77 lines
2.2 KiB
Plaintext
#version 450
|
|
#define LOCAL_SZ_X 256
|
|
layout(binding = 0) readonly buffer Input0{
|
|
float image_data[];
|
|
};
|
|
layout(binding = 1) readonly buffer Input1 {
|
|
float bias_data[];
|
|
};
|
|
layout(binding = 2) readonly buffer Input3{
|
|
float weight_data[];
|
|
};
|
|
layout(binding = 3) writeonly buffer Output{
|
|
float convolved_image_data[];
|
|
};
|
|
|
|
layout(push_constant) uniform pushBlock {
|
|
int in_h;
|
|
int in_w;
|
|
int out_h;
|
|
int out_w;
|
|
int stride_h;
|
|
int stride_w;
|
|
int pad_h;
|
|
int pad_w;
|
|
int filter_h;
|
|
int filter_w;
|
|
int dilation_h;
|
|
int dilation_w;
|
|
int channels;
|
|
int batch;
|
|
int has_bias;
|
|
int M;
|
|
int K;
|
|
int N;
|
|
int basic_shader_batch_idx;
|
|
int basic_shader_partition_idx;
|
|
int basic_shader_partition_size;
|
|
} p;
|
|
|
|
layout(local_size_x = LOCAL_SZ_X, local_size_y = 1, local_size_z = 1) in;
|
|
void main()
|
|
{
|
|
int gx = int(gl_GlobalInvocationID.x);
|
|
int gy = int(gl_GlobalInvocationID.y) + p.basic_shader_partition_idx * p.basic_shader_partition_size;
|
|
int gz = p.basic_shader_batch_idx;
|
|
if(gx < p.M && gy < p.N)
|
|
{
|
|
float sum = 0.0f;
|
|
int output_y = gx / p.out_w;
|
|
int output_x = gx % p.out_w;
|
|
int org_y = output_y * p.stride_h - p.pad_h;
|
|
int org_x = output_x * p.stride_w - p.pad_w;
|
|
int weight_off = gy * p.K;
|
|
int input_off = gz * p.in_h * p.in_w * p.channels + (org_y * p.in_w + org_x);
|
|
for(int c = 0; c < p.channels; c++)
|
|
{
|
|
for(int y = 0; y < p.filter_h; y++)
|
|
{
|
|
for(int x = 0; x < p.filter_w; x++)
|
|
{
|
|
if((org_y + y * p.dilation_h >= 0) && (org_y + y * p.dilation_h < p.in_h) && (org_x + x * p.dilation_w >= 0) && (org_x + x * p.dilation_w < p.in_w))
|
|
{
|
|
sum += image_data[input_off + x * p.dilation_w] * weight_data[weight_off + x];
|
|
}
|
|
}
|
|
input_off += p.in_w * p.dilation_h;
|
|
weight_off += p.filter_w;
|
|
}
|
|
input_off += p.in_h * p.in_w - p.in_w * p.filter_h * p.dilation_h;
|
|
}
|
|
int offset = gz * p.M * p.N + gx + gy * p.M;
|
|
if (p.has_bias == 1)
|
|
sum += bias_data[gy];
|
|
convolved_image_data[offset] = sum;
|
|
}
|
|
}
|