diff mbox series

[FFmpeg-devel,2/2] Add mutithread function for dnn_backend_native_layer_conv2d.c

Message ID 20200903155724.167477-2-xujunzz@sjtu.edu.cn
State New
Headers show
Series [FFmpeg-devel,1/2] dnn_backend_native.c: parse options in native backend | expand

Checks

Context Check Description
andriy/default pending
andriy/make_warn warning New warnings during build
andriy/make success Make finished
andriy/make_fate fail Make fate failed

Commit Message

Xu Jun Sept. 3, 2020, 3:57 p.m. UTC 
From: Xu Jun <xujunzz@sjtu.edu.cn>

Use pthread to multithread dnn_execute_layer_conv2d.
Can be tested with command "./ffmpeg_g -i input.png -vf \
format=yuvj420p,dnn_processing=dnn_backend=native:model= \
espcn.model:input=x:output=y:options=conv2d_threads=23 \
 -y sr_native.jpg -benchmark"

before patch: utime=11.238s stime=0.005s rtime=11.248s
after patch:  utime=20.817s stime=0.047s rtime=1.051s
on my 3900X 12c24t @4.2GHz

About the increase of utime, it's because that CPU HyperThreading
technology makes logical cores twice of physical cores while cpu's
counting performance improves less than double. And utime sums
all cpu's logical cores' runtime. As a result, using threads num
near cpu's logical core's number will double utime, while reduce
rtime less than half for HyperThreading CPUs.

Signed-off-by: Xu Jun <xujunzz@sjtu.edu.cn>
---
 .../dnn/dnn_backend_native_layer_conv2d.c     | 92 ++++++++++++++++---
 1 file changed, 81 insertions(+), 11 deletions(-)

Comments

Michael Niedermayer Sept. 3, 2020, 8:15 p.m. UTC | #1 
On Thu, Sep 03, 2020 at 11:57:24PM +0800, xujunzz@sjtu.edu.cn wrote:
> From: Xu Jun <xujunzz@sjtu.edu.cn>
> 
> Use pthread to multithread dnn_execute_layer_conv2d.
> Can be tested with command "./ffmpeg_g -i input.png -vf \
> format=yuvj420p,dnn_processing=dnn_backend=native:model= \
> espcn.model:input=x:output=y:options=conv2d_threads=23 \
>  -y sr_native.jpg -benchmark"
> 
> before patch: utime=11.238s stime=0.005s rtime=11.248s
> after patch:  utime=20.817s stime=0.047s rtime=1.051s
> on my 3900X 12c24t @4.2GHz
> 
> About the increase of utime, it's because that CPU HyperThreading
> technology makes logical cores twice of physical cores while cpu's
> counting performance improves less than double. And utime sums
> all cpu's logical cores' runtime. As a result, using threads num
> near cpu's logical core's number will double utime, while reduce
> rtime less than half for HyperThreading CPUs.
> 
> Signed-off-by: Xu Jun <xujunzz@sjtu.edu.cn>
> ---
>  .../dnn/dnn_backend_native_layer_conv2d.c     | 92 ++++++++++++++++---
>  1 file changed, 81 insertions(+), 11 deletions(-)

[...]

> +typedef struct thread_param{
> +    thread_common_param *thread_common_param;
> +    int thread_index

semicolon missing


[...]
> +    //join threads, res gets function return
> +    for (int i = 0; i < thread_num; i++){
> +        pthread_join(thread_id[i], &res);

this should be under something like HAVE_PTHREAD_CANCEL

thx

[...]
diff mbox series

Patch

diff --git a/libavfilter/dnn/dnn_backend_native_layer_conv2d.c b/libavfilter/dnn/dnn_backend_native_layer_conv2d.c
index d079795bf8..8da99540ed 100644
--- a/libavfilter/dnn/dnn_backend_native_layer_conv2d.c
+++ b/libavfilter/dnn/dnn_backend_native_layer_conv2d.c
@@ -19,10 +19,27 @@ 
  */
 
 #include "libavutil/avassert.h"
+#include "libavutil/thread.h"
+#include "libavutil/cpu.h"
 #include "dnn_backend_native_layer_conv2d.h"
 
 #define CLAMP_TO_EDGE(x, w) ((x) < 0 ? 0 : ((x) >= (w) ? (w - 1) : (x)))
 
+//struct to pass parameters
+typedef struct thread_common_param{
+    DnnOperand *operands;
+    const int32_t *input_operand_indexes;
+    int32_t output_operand_index;
+    const void *parameters;
+    NativeContext *ctx;
+    int thread_num;
+} thread_common_param;
+
+typedef struct thread_param{
+    thread_common_param *thread_common_param;
+    int thread_index
+} thread_param;
+
 int dnn_load_layer_conv2d(Layer *layer, AVIOContext *model_file_context, int file_size, int operands_num)
 {
     ConvolutionalParams *conv_params;
@@ -88,17 +105,20 @@  int dnn_load_layer_conv2d(Layer *layer, AVIOContext *model_file_context, int fil
     return dnn_size;
 }
 
-int dnn_execute_layer_conv2d(DnnOperand *operands, const int32_t *input_operand_indexes,
-                             int32_t output_operand_index, const void *parameters, NativeContext *ctx)
+static void * dnn_execute_layer_conv2d_thread(void *threadarg)
 {
+    //pass parameters
+    thread_param *thread_param = (struct thread_param *)threadarg;
+    thread_common_param *thread_common_param = thread_param->thread_common_param;
+    DnnOperand *operands = thread_common_param->operands;
     float *output;
-    int32_t input_operand_index = input_operand_indexes[0];
+    int32_t input_operand_index = thread_common_param->input_operand_indexes[0];
     int number = operands[input_operand_index].dims[0];
     int height = operands[input_operand_index].dims[1];
     int width = operands[input_operand_index].dims[2];
     int channel = operands[input_operand_index].dims[3];
     const float *input = operands[input_operand_index].data;
-    const ConvolutionalParams *conv_params = (const ConvolutionalParams *)parameters;
+    const ConvolutionalParams *conv_params = (const ConvolutionalParams *)(thread_common_param->parameters);
 
     int radius = conv_params->kernel_size >> 1;
     int src_linesize = width * conv_params->input_num;
@@ -106,7 +126,11 @@  int dnn_execute_layer_conv2d(DnnOperand *operands, const int32_t *input_operand_
     int filter_size = conv_params->kernel_size * filter_linesize;
     int pad_size = (conv_params->padding_method == VALID) ? (conv_params->kernel_size - 1) / 2 * conv_params->dilation : 0;
 
-    DnnOperand *output_operand = &operands[output_operand_index];
+    int thread_stride = (height - pad_size * 2) / thread_common_param->thread_num;
+    int thread_start = thread_stride * thread_param->thread_index + pad_size;
+    int thread_end = (thread_param->thread_index == thread_common_param->thread_num - 1) ? (height - pad_size) : (thread_start + thread_stride);
+
+    DnnOperand *output_operand = &operands[thread_common_param->output_operand_index];
     output_operand->dims[0] = number;
     output_operand->dims[1] = height - pad_size * 2;
     output_operand->dims[2] = width - pad_size * 2;
@@ -114,19 +138,21 @@  int dnn_execute_layer_conv2d(DnnOperand *operands, const int32_t *input_operand_
     output_operand->data_type = operands[input_operand_index].data_type;
     output_operand->length = calculate_operand_data_length(output_operand);
     if (output_operand->length <= 0) {
-        av_log(ctx, AV_LOG_ERROR, "The output data length overflow\n");
-        return DNN_ERROR;
+        av_log(thread_common_param->ctx, AV_LOG_ERROR, "The output data length overflow\n");
+        return (void *)DNN_ERROR;
     }
     output_operand->data = av_realloc(output_operand->data, output_operand->length);
     if (!output_operand->data) {
-        av_log(ctx, AV_LOG_ERROR, "Failed to reallocate memory for output\n");
-        return DNN_ERROR;
+        av_log(thread_common_param->ctx, AV_LOG_ERROR, "Failed to reallocate memory for output\n");
+        return (void *)DNN_ERROR;
     }
+
     output = output_operand->data;
+    output += (conv_params->output_num) * (width - 2 * pad_size) * (thread_start - pad_size);
 
     av_assert0(channel == conv_params->input_num);
 
-    for (int y = pad_size; y < height - pad_size; ++y) {
+    for (int y = thread_start; y < thread_end; ++y) {
         for (int x = pad_size; x < width - pad_size; ++x) {
             for (int n_filter = 0; n_filter < conv_params->output_num; ++n_filter) {
                 if (conv_params->has_bias)
@@ -174,5 +200,49 @@  int dnn_execute_layer_conv2d(DnnOperand *operands, const int32_t *input_operand_
             output += conv_params->output_num;
         }
     }
-    return 0;
+    return (void *)0;
+}
+
+
+int dnn_execute_layer_conv2d(DnnOperand *operands, const int32_t *input_operand_indexes,
+                             int32_t output_operand_index, const void *parameters, NativeContext *ctx)
+{
+    int thread_num = (ctx->options.conv2d_threads <= 0 || ctx->options.conv2d_threads > av_cpu_count())
+        ? (av_cpu_count() + 1) : (ctx->options.conv2d_threads);
+    pthread_t *thread_id = av_malloc(thread_num * sizeof(pthread_t));
+    thread_param **thread_param = av_malloc(thread_num * sizeof(*thread_param));
+    void *res;
+    int error_flag = 0;
+
+    //struct used to pass parameters
+    thread_common_param thread_common_param;
+    thread_common_param.operands = operands;
+    thread_common_param.input_operand_indexes = input_operand_indexes;
+    thread_common_param.output_operand_index = output_operand_index;
+    thread_common_param.parameters = parameters;
+    thread_common_param.ctx = ctx;
+    thread_common_param.thread_num = thread_num;
+
+    //create threads
+    for (int i = 0; i < thread_num; i++){
+        thread_param[i] = av_malloc(sizeof(thread_param));
+        thread_param[i]->thread_common_param = &thread_common_param;
+        thread_param[i]->thread_index = i;
+        pthread_create(&thread_id[i], NULL, dnn_execute_layer_conv2d_thread, (void *)thread_param[i]);
+    }
+
+    //join threads, res gets function return
+    for (int i = 0; i < thread_num; i++){
+        pthread_join(thread_id[i], &res);
+        if ((int)res != 0)
+            error_flag = (int)res;
+    }
+
+    //release memory
+    av_free(thread_id);
+    for (int i = 0; i < thread_num; i++){
+        av_free(thread_param[i]);
+    }
+    av_free(thread_param);
+    return error_flag;
 }