std::list里面的push_back和resize效率问题

2024年8月6日：

向list里面循环添加数据，两种写法

for(;;){myList.resize(myList.size()+1);MyObject &obj= *myList.rbegin();obj.a= itervalue;
}

第二种写法是push_back:

for(;;){MyObject obj；obj.a= itervalue;myList.push_back(obj);
}

用第一种用法，是避免了push_back的拷贝构造函数，用的缺省构造函数，对于COPY构造函数效率毕竟低的场景，也许有效。但是大部分时间应该没有啥用。

效率上，找了一个评测网站：

Quick C++ Benchmarks

代码：

#include <vector>
#include <list>constexpr size_t loopCount = 1000;struct BigObject{double LongDouble[10000];
};static void VectorPushBack(benchmark::State& state) {for (auto _ : state) {std::vector<BigObject> vec;for (int i = 0; i < loopCount; i++) {BigObject bo;vec.push_back(bo);}benchmark::DoNotOptimize(vec);}
}
BENCHMARK(VectorPushBack);static void VectorEmplaceBack(benchmark::State& state) {for (auto _ : state) {std::vector<BigObject> vec;for (int i = 0; i < loopCount; i++) {vec.emplace_back();}benchmark::DoNotOptimize(vec);}
}
BENCHMARK(VectorEmplaceBack);static void ListPushBack(benchmark::State& state) {for (auto _ : state) {std::list<BigObject> list;for (int i = 0; i < loopCount; i++) {BigObject bo;list.push_back(bo);}benchmark::DoNotOptimize(list);}
}
BENCHMARK(ListPushBack);static void ListEmplaceBack(benchmark::State& state) {for (auto _ : state) {std::list<BigObject> list;for (int i = 0; i < loopCount; i++) {list.emplace_back();}benchmark::DoNotOptimize(list);}
}
BENCHMARK(ListEmplaceBack);
static void ListResize(benchmark::State& state) {for (auto _ : state) {std::list<BigObject> list;for (int i = 0; i < loopCount; i++) {BigObject bo;//list.push_back(bo);list.resize(list.size()+1);BigObject &obj = *list.rbegin();obj=bo;}benchmark::DoNotOptimize(list);}
}
BENCHMARK(ListResize);

结果：push_back效率最高，高的原因不详

参考：

https://www.reddit.com/r/cpp_questions/comments/15uekey/push_back_big_object_into_stdvector_by_creating/?rdt=60610