[feat] add pipeline.trim function.

ChunelFeng · Sep 1, 2024 · 3a6d690 · 3a6d690
1 parent 3351726
commit 3a6d690
Show file tree

Hide file tree

Showing 12 changed files with 160 additions and 31 deletions.
diff --git a/README.md b/README.md
@@ -325,6 +325,7 @@ int main() {
 [2024.07.27 - v2.6.1 - Chunel]
 * 提供`pipeline`的静态执行的方式，提供微任务机制
 * 优化`event`(事件)机制，异步事件可以等待结束
+* 提供`pipeline`剪裁功能，用于删除`element`之间重复的依赖
 
 </details>
 

diff --git a/src/GraphCtrl/GraphElement/GElement.cpp b/src/GraphCtrl/GraphElement/GElement.cpp
@@ -566,4 +566,18 @@ CBool GElement::isDefaultBinding() const {
     return CGRAPH_DEFAULT_BINDING_INDEX == binding_index_;
 }
 
+
+CBool GElement::removeDepend(GElementPtr element) {
+    CGRAPH_ASSERT_NOT_NULL_THROW_ERROR(element)
+    CGRAPH_ASSERT_INIT_THROW_ERROR(false)
+    if (!dependence_.hasValue(element)) {
+        return false;
+    }
+
+    dependence_.remove(element);
+    element->run_before_.remove(this);
+    left_depend_.store(dependence_.size(), std::memory_order_release);
+    return true;
+}
+
 CGRAPH_NAMESPACE_END
diff --git a/src/GraphCtrl/GraphElement/GElement.h b/src/GraphCtrl/GraphElement/GElement.h
@@ -419,6 +419,13 @@ class GElement : public GElementObject,
      */
     CBool isDefaultBinding() const;
 
+    /**
+     * 删除一个依赖的节点信息
+     * @param element
+     * @return
+     */
+    CBool removeDepend(GElement* element);
+
 private:
     /** 状态相关信息 */
     CBool done_ { false };                                                    // 判定被执行结束
@@ -444,7 +451,7 @@ class GElement : public GElementObject,
     CBool is_prepared_ { false };                                             // 判断是否已经执行过 prepareRun() 方法
 
     /** 图相关信息 */
-    std::atomic<CSize> left_depend_ { 0 };                                 // 当 left_depend_ 值为0的时候，即可以执行该element信息
+    std::atomic<CSize> left_depend_ { 0 };                                    // 当 left_depend_ 值为0的时候，即可以执行该element信息
     USmallVector<GElement *> run_before_;                                     // 被依赖的节点（后继）
     USmallVector<GElement *> dependence_;                                     // 依赖的节点信息（前驱）
     GElement* belong_ { nullptr };                                            // 从属的element 信息，如为nullptr，则表示从属于 pipeline
@@ -477,6 +484,7 @@ class GElement : public GElementObject,
     friend class GAspectObject;
     friend class GOptimizer;
     friend class GMaxParaOptimizer;
+    friend class GTrimOptimizer;
     friend class GSeparateOptimizer;
     friend class GElementRepository;
     friend class GPerf;

diff --git a/src/GraphCtrl/GraphElement/GElementManager.cpp b/src/GraphCtrl/GraphElement/GElementManager.cpp
@@ -176,6 +176,11 @@ CBool GElementManager::checkSerializable() {
 }
 
 
+CSize GElementManager::trim() {
+    return GTrimOptimizer::trim(manager_elements_);
+}
+
+
 CStatus GElementManager::process(const GSortedGElementPtrSet& elements) {
     CGRAPH_FUNCTION_BEGIN
     CGRAPH_ASSERT_NOT_NULL(engine_)

diff --git a/src/GraphCtrl/GraphElement/GElementManager.h b/src/GraphCtrl/GraphElement/GElementManager.h
@@ -85,6 +85,12 @@ class GElementManager : public GElementObject,
      */
     CBool checkSerializable();
 
+    /**
+     * 剪裁多余的连边信息
+     * @return
+     */
+    CSize trim();
+
     /**
      * 加入数据，并且执行
      * @param elements

diff --git a/src/GraphCtrl/GraphElement/_GOptimizer/GMaxParaOptimizer.h b/src/GraphCtrl/GraphElement/_GOptimizer/GMaxParaOptimizer.h
@@ -41,39 +41,12 @@ class GMaxParaOptimizer : public GOptimizer {
          * 2. 根据路径，生成一张全连通图，然后将图取反，得到对应的补图(reGraph)
          * 3. 计算补图的最大团中元素个数(maxCliqueSize)，即为当前dag的最大并行度
          */
-        const CSize size = elements.size();
-        const auto& paths = GOptimizer::collectPaths(elements);    // 根据传入的elements 的关系，分析出所有完整路径信息
-
-        std::vector<std::vector<int>> reGraph(size, std::vector<int>(size, 1));
-        buildReverseGraph(elements, paths, reGraph);    // 根据路径信息，求出全连接图的补图
-
-        CSize maxCliqueSize = calcMaxCliqueSize(reGraph);    // 计算最大团信息
+        const auto& paths = collectPaths(elements);
+        const auto& reGraph = buildGraph(elements, paths, 0, 0, 1);
+        CSize maxCliqueSize = calcMaxCliqueSize(reGraph);
         return maxCliqueSize;
     }
 
-    /**
-     * 基于原有dag，计算出所有链路的全连接的补图，记作graph
-     * @param elements
-     * @param paths
-     * @param graph
-     * @return
-     */
-    static CVoid buildReverseGraph(const GSortedGElementPtrSet& elements,
-                                   const std::vector<std::vector<GElementPtr>>& paths,
-                                   std::vector<std::vector<int>>& graph) {
-        for (auto& path : paths) {
-            for (int i = 0; i < path.size() - 1; i++) {
-                // 这里的 find是一定能找到的。因为path的数据，是从elements中记录的
-                int height = (int)std::distance(elements.begin(), elements.find(path[i]));
-                for (int j = i + 1; j < path.size(); j++) {
-                    int column = (int)std::distance(elements.begin(), elements.find(path[j]));
-                    graph[height][column] = 0;
-                    graph[column][height] = 0;    // 因为需要记录的是补图，所以将默认值设置为1，符合条件的设置为0
-                }
-            }
-        }
-    }
-
     /**
      * 计算当前最大团的大小
      * @param graph

diff --git a/src/GraphCtrl/GraphElement/_GOptimizer/GOptimizer.h b/src/GraphCtrl/GraphElement/_GOptimizer/GOptimizer.h
@@ -9,6 +9,8 @@
 #ifndef CGRAPH_GOPTIMIZER_H
 #define CGRAPH_GOPTIMIZER_H
 
+#include <vector>
+
 #include "../GElementObject.h"
 
 CGRAPH_NAMESPACE_BEGIN
@@ -54,6 +56,34 @@ class GOptimizer : public GElementObject {
         return paths;
     }
 
+    /**
+     * 构造对应的二维矩阵图
+     * @param elements
+     * @param paths
+     * @param father
+     * @param son
+     * @param unlink
+     * @return
+     */
+    static std::vector<std::vector<int>> buildGraph(const GSortedGElementPtrSet& elements,
+                            const std::vector<std::vector<GElementPtr>>& paths,
+                            int father, int son, int unlink) {
+        const CSize size = elements.size();
+        std::vector<std::vector<int>> graph(size, std::vector<int>(size, unlink));
+        for (auto& path : paths) {
+            for (int i = 0; i < path.size() - 1; i++) {
+                // 这里的 find是一定能找到的。因为path的数据，是从elements中记录的
+                int height = (int)std::distance(elements.begin(), elements.find(path[i]));
+                for (int j = i + 1; j < path.size(); j++) {
+                    int column = (int)std::distance(elements.begin(), elements.find(path[j]));
+                    graph[height][column] = father;
+                    graph[column][height] = son;
+                }
+            }
+        }
+        return graph;
+    }
+
     friend class GPerf;
 };
 

diff --git a/src/GraphCtrl/GraphElement/_GOptimizer/GOptimizerInclude.h b/src/GraphCtrl/GraphElement/_GOptimizer/GOptimizerInclude.h
@@ -12,5 +12,6 @@
 #include "GOptimizer.h"
 #include "GMaxParaOptimizer.h"
 #include "GSeparateOptimizer.h"
+#include "GTrimOptimizer.h"
 
 #endif //CGRAPH_GOPTIMIZERINCLUDE_H
diff --git a/src/GraphCtrl/GraphElement/_GOptimizer/GTrimOptimizer.h b/src/GraphCtrl/GraphElement/_GOptimizer/GTrimOptimizer.h
@@ -0,0 +1,59 @@
+/***************************
+@Author: Chunel
+@Contact: [email protected]
+@File: GTrimOptimizer.h
+@Time: 2024/9/1 13:59
+@Desc: 
+***************************/
+
+#ifndef CGRAPH_GTRIMOPTIMIZER_H
+#define CGRAPH_GTRIMOPTIMIZER_H
+
+#include <algorithm>
+#include <iterator>
+
+#include "GOptimizer.h"
+
+CGRAPH_NAMESPACE_BEGIN
+
+class GTrimOptimizer : public GOptimizer {
+    /**
+     * 针对图结构，做剪裁
+     * @param elements
+     * @return
+     */
+    static CSize trim(GSortedGElementPtrSet& elements) {
+        CSize trimNum = 0;
+        const auto& paths = collectPaths(elements);
+        auto graph = buildGraph(elements, paths, 1, 0, 0);
+
+        for (auto* cur : elements) {
+            int idx = (int)std::distance(elements.begin(), elements.find(cur));
+            GElementPtrArr candidates;
+            for (int i = 0; i < (int)cur->dependence_.size(); i++) {
+                int x = (int)std::distance(elements.begin(), elements.find(cur->dependence_[i]));
+                for (int j = i; j < (int)cur->dependence_.size(); j++) {
+                    int y = (int)std::distance(elements.begin(), elements.find(cur->dependence_[j]));
+                    if (1 == graph[x][y]) {
+                        graph[x][idx] = 0;
+                        candidates.push_back(cur->dependence_[i]);
+                    }
+                }
+            }
+
+            for (auto* candidate : candidates) {
+                if (cur->removeDepend(candidate)) {
+                    trimNum++;
+                }
+            }
+        }
+
+        return trimNum;
+    }
+
+    friend class GElementManager;
+};
+
+CGRAPH_NAMESPACE_END
+
+#endif //CGRAPH_GTRIMOPTIMIZER_H
diff --git a/src/GraphCtrl/GraphPipeline/GPipeline.cpp b/src/GraphCtrl/GraphPipeline/GPipeline.cpp
@@ -262,11 +262,19 @@ GPipelinePtr GPipeline::setAutoCheck(CBool enable) {
 
 
 CSize GPipeline::getMaxPara() {
+    CGRAPH_ASSERT_INIT_THROW_ERROR(false)
     CGRAPH_ASSERT_NOT_NULL_THROW_ERROR(element_manager_)
     return element_manager_->calcMaxParaSize();
 }
 
 
+CSize GPipeline::trim() {
+    CGRAPH_ASSERT_INIT_THROW_ERROR(false)
+    CGRAPH_ASSERT_NOT_NULL_THROW_ERROR(element_manager_)
+    return element_manager_->trim();
+}
+
+
 CStatus GPipeline::makeSerial() {
     CGRAPH_FUNCTION_BEGIN
     CGRAPH_ASSERT_INIT(false)

diff --git a/src/GraphCtrl/GraphPipeline/GPipeline.h b/src/GraphCtrl/GraphPipeline/GPipeline.h
@@ -354,6 +354,12 @@ class GPipeline : public GPipelineObject,
      */
     CSize getMaxPara();
 
+    /**
+     * 针对图结构，做多余边剪裁
+     * @return 返回值表示裁剪了多少条信息
+     */
+    CSize trim();
+
     /**
      * 将符合串行执行条件的pipeline，设定为串行执行的模式。可以大幅度提升运行性能。
      * @return

diff --git a/src/UtilsCtrl/Container/USmallVector.h b/src/UtilsCtrl/Container/USmallVector.h
@@ -68,6 +68,22 @@ class USmallVector : public UtilsObject {
         memset(data_, 0, sizeof(T) * capacity_);
     }
 
+    CBool remove(const T& val) {
+        CBool result = false;
+        for (CSize i = 0; i < cur_index_; i++) {
+            if (data_[i] == val) {
+                for (CSize j = i; j < cur_index_ - 1; j++) {
+                    data_[j] = data_[j + 1];
+                }
+                cur_index_--;
+                result = true;
+                break;
+            }
+        }
+        return result;
+    }
+
+
     /**
      * 判断内部是否包含该值
      * @param val
@@ -122,6 +138,8 @@ class USmallVector : public UtilsObject {
     UIter end() const { return UIter(data_ + cur_index_); }
     T front() const { return data_[0]; }
     T back() const { return data_[cur_index_ - 1]; }
+    T& operator[](CSize index) { return data_[index]; }
+    const T& operator[](CSize index) const { return data_[index]; }
 
 private:
     T* data_ { nullptr };          // 存放具体数据