编译器的主要目标是降低编译成本,并使调试产生预期的结果。并非所有优化都直接由标志控制,有时我们需要显式声明标志以产生优化。默认情况下,不优化。要使用抑制的优化,我们将使用编译指示。
未优化程序的示例:让我们考虑一个示例,计算最大10000000的素数。
下面是没有优化的代码:
// C++ program to calculate the Prime
// Numbers upto 10000000 using Sieve
// of Eratosthenes with NO optimization
#include
#include
#include
#define N 10000005
using namespace std;
// Boolean array for Prime Number
vector prime(N, true);
// Sieve implemented to find Prime
// Number
void sieveOfEratosthenes()
{
for (int i = 2; i <= sqrt(N); ++i) {
if (prime[i]) {
for (int j = i * i; j <= N; j += i) {
prime[j] = false;
}
}
}
}
// Driver Code
int main()
{
// Intialise clock to calculate
// time required to execute without
// optimization
clock_t start, end;
// Start clock
start = clock();
// Function call to find Prime Numbers
sieveOfEratosthenes();
// End clock
end = clock();
// Calculate the time difference
double time_taken
= double(end - start)
/ double(CLOCKS_PER_SEC);
// Print the Calculated execution time
cout << "Execution time: " << time_taken
<< " secs";
return 0;
}
输出:
Execution time: 0.592183 secs
以下是优化:
- O1:在O1上优化编译需要更多的时间和内存来分解较大的功能。编译器试图减少代码和执行时间。在O1上,几乎没有任何优化可以产生出色的结果,但是O1是尝试进行更好的优化的挫折。
以下是使用O1优化的先前程序的实现:
// C++ program to calculate the Prime // Numbers upto 10000000 using Sieve // of Eratosthenes with O1 optimization // To see the working of controlled // optimization "O1" #pragma GCC optimize("O1") #include
#include #include #define N 10000005 using namespace std; // Boolean array for Prime Number vector prime(N, true); // Sieve implemented to find Prime // Number void sieveOfEratosthenes() { for (int i = 2; i <= sqrt(N); ++i) { if (prime[i]) { for (int j = i * i; j <= N; j += i) { prime[j] = false; } } } } // Driver Code int main() { // Intialise clock to calculate // time required to execute without // optimization clock_t start, end; // Start clock start = clock(); // Function call to find Prime Numbers sieveOfEratosthenes(); // End clock end = clock(); // Calculate the time difference double time_taken = double(end - start) / double(CLOCKS_PER_SEC); // Print the Calculated execution time cout << "Execution time: " << time_taken << " secs."; return 0; } 输出:Execution time: 0.384945 secs.
- O2:在O2优化优化汇编到更大的程度。与O1相比,此选项增加了编译时间和所生成代码的性能。 O2打开由O1指定的所有优化标志。
以下是使用O2优化的先前程序的实现:
// C++ program to calculate the Prime // Numbers upto 10000000 using Sieve // of Eratosthenes with O2 optimization // To see the working of controlled // optimization "O2" #pragma GCC optimize("O2") #include
#include #include #define N 10000005 using namespace std; // Boolean array for Prime Number vector prime(N, true); // Sieve implemented to find Prime // Number void sieveOfEratosthenes() { for (int i = 2; i <= sqrt(N); ++i) { if (prime[i]) { for (int j = i * i; j <= N; j += i) { prime[j] = false; } } } } // Driver Code int main() { // Intialise clock to calculate // time required to execute without // optimization clock_t start, end; // Start clock start = clock(); // Function call to find Prime Numbers sieveOfEratosthenes(); // End clock end = clock(); // Calculate the time difference double time_taken = double(end - start) / double(CLOCKS_PER_SEC); // Print the Calculated execution time cout << "Execution time: " << time_taken << " secs."; return 0; } 输出:Execution time: 0.288337 secs.
- O3:在所有水平的臭氧和其他标志的列表中指定的O2优化也将启用。 O3中包含的标志很少是floop-interchange -floop-unroll-jam和-fpeel-loops。
以下是使用O3优化的先前程序的实现:
// C++ program to calculate the Prime // Numbers upto 10000000 using Sieve // of Eratosthenes with O3 optimization // To see the working of controlled // optimization "O3" #pragma GCC optimize("O3") #include
#include #include #define N 10000005 using namespace std; // Boolean array for Prime Number vector prime(N, true); // Sieve implemented to find Prime // Number void sieveOfEratosthenes() { for (int i = 2; i <= sqrt(N); ++i) { if (prime[i]) { for (int j = i * i; j <= N; j += i) { prime[j] = false; } } } } // Driver Code int main() { // Intialise clock to calculate // time required to execute without // optimization clock_t start, end; // Start clock start = clock(); // Function call to find Prime Numbers sieveOfEratosthenes(); // End clock end = clock(); // Calculate the time difference double time_taken = double(end - start) / double(CLOCKS_PER_SEC); // Print the Calculated execution time cout << "Execution time: " << time_taken << " secs."; return 0; } 输出:Execution time: 0.580154 secs.
- Os:针对大小进行了优化。 Os启用所有O2优化,但增加代码大小的优化除外。它还启用-finline-functions功能,使编译器调整代码大小而不是执行速度,并执行旨在减小代码大小的进一步优化。
以下是使用OS优化的先前程序的实现:
// C++ program to calculate the Prime // Numbers upto 10000000 using Sieve // of Eratosthenes with Os optimization // To see the working of controlled // optimization "Os" #pragma GCC optimize("Os") #include
#include #include #define N 10000005 using namespace std; // Boolean array for Prime Number vector prime(N, true); // Sieve implemented to find Prime // Number void sieveOfEratosthenes() { for (int i = 2; i <= sqrt(N); ++i) { if (prime[i]) { for (int j = i * i; j <= N; j += i) { prime[j] = false; } } } } // Driver Code int main() { // Intialise clock to calculate // time required to execute without // optimization clock_t start, end; // Start clock start = clock(); // Function call to find Prime Numbers sieveOfEratosthenes(); // End clock end = clock(); // Calculate the time difference double time_taken = double(end - start) / double(CLOCKS_PER_SEC); // Print the Calculated execution time cout << "Execution time: " << time_taken << " secs."; return 0; } 输出:Execution time: 0.317845 secs.
- Ofast: Ofast启用所有O3优化。它还具有产生超级优化结果的已启用标志的数量。 Ofast结合了以上每个O级别产生的优化。这种优化通常是许多有竞争力的程序员所偏爱的,因此值得推荐。如果声明了多个优化,则最后一个声明的优化将被启用。
以下是使用Ofast优化的先前程序的实现:
// C++ program to calculate the Prime // Numbers upto 10000000 using Sieve // of Eratosthenes with Ofast optimization // To see the working of controlled // optimization "Ofast" #pragma GCC optimize("Ofast") #include
#include #include #define N 10000005 using namespace std; // Boolean array for Prime Number vector prime(N, true); // Sieve implemented to find Prime // Number void sieveOfEratosthenes() { for (int i = 2; i <= sqrt(N); ++i) { if (prime[i]) { for (int j = i * i; j <= N; j += i) { prime[j] = false; } } } } // Driver Code int main() { // Intialise clock to calculate // time required to execute without // optimization clock_t start, end; // Start clock start = clock(); // Function call to find Prime Numbers sieveOfEratosthenes(); // End clock end = clock(); // Calculate the time difference double time_taken = double(end - start) / double(CLOCKS_PER_SEC); // Print the Calculated execution time cout << "Execution time: " << time_taken << " secs."; return 0; } 输出:Execution time: 0.303287 secs.
为了进一步实现体系结构级别的优化,我们可以将目标与编译指示一起使用。这些优化可以产生令人惊讶的结果。但是,建议将target与以上指定的任何优化配合使用。
以下是先前使用Target实现的程序的实现:
// C++ program to calculate the Prime
// Numbers upto 10000000 using Sieve
// of Eratosthenes with Ofast optimization along with target optimizations
// To see the working of controlled
// optimization "Ofast"
#pragma GCC optimize("Ofast")
#pragma GCC target("avx,avx2,fma")
#include
#include
#include
#define N 10000005
using namespace std;
// Boolean array for Prime Number
vector prime(N, true);
// Sieve implemented to find Prime
// Number
void sieveOfEratosthenes()
{
for (int i = 2; i <= sqrt(N); ++i) {
if (prime[i]) {
for (int j = i * i; j <= N; j += i) {
prime[j] = false;
}
}
}
}
// Driver Code
int main()
{
// Intialise clock to calculate
// time required to execute without
// optimization
clock_t start, end;
// Start clock
start = clock();
// Function call to find Prime Numbers
sieveOfEratosthenes();
// End clock
end = clock();
// Calculate the time difference
double time_taken
= double(end - start)
/ double(CLOCKS_PER_SEC);
// Print the Calculated execution time
cout << "Execution time: " << time_taken
<< " secs.";
return 0;
}
输出:
Execution time: 0.292147 secs.
如果您希望与行业专家一起参加现场课程,请参阅《 Geeks现场课程》和《 Geeks现场课程美国》。